A Study on Identifying the Spatial Characteristic Factors of Traditional Streets Based on Visitor Perception: Yuanjia Village, Shaanxi Province

Abstract

The street spaces in tourist-oriented traditional villages served both the daily lives of villagers and the leisure activities of tourists. However, due to insufficient understanding of the spatial characteristics and under-exploration of spatial genes, these spaces often suffered from homogenization during tourism development. Thus, identifying the characteristics and connotations of such streets, understanding the relationship between tourists’ perceptions and built environment elements, and developing optimization strategies for these rural street spaces were urgent issues. Many studies have evaluated street space characteristics from tourists’ behavior, but few have focused on rural areas. Especially, research combining new technologies like artificial intelligence to study the psychological perceptions of tourists is still in its infancy. This study used a typical traditional village as a case study and applied the YOLOv5 deep-learning model to build a perception evaluation system based on three dimensions: tourists’ aggregation degree, stay time, and facial expressions. The study conducted a multivariate regression analysis on 21 factors across 4 aspects: street scale morphology, environmental facilities, ground-floor interface, and street business types. Results indicated that the functional business type of the scene had the greatest impact on tourists’ perception of the street space environment, followed by ground-floor features and environmental facilities. The regression coefficient for business in situ values and spatial perception was 0.47, highlighting it as a key factor influencing characteristic perception. Landscape water systems, flat ground-floor façades, and business diversity also positively affected tourists’ perception. This study utilized advanced techniques like the YOLOv5 model, known for its speed and accuracy, to scientifically analyze tourists’ behavior and perceptions, serving as feedback and evaluation for the village’s built environment. Empirical analysis of Yuanjia Village validated the effectiveness of the multidimensional approach and spatial gene theory. Ultimately, this method identified 12 street characteristic factors significantly affecting tourists’ perceptions. The uniqueness of this study lies in its comprehensive approach, combining empirical research, spatial gene theory, and advanced object detection technology, providing new insights for village spatial planning and construction.

1. Introduction

Traditional villages contain rich cultural heritage resources, which are a critical driving force for sustainable rural development. Rural tourism brings new visitors to villages. It introduces capital and enhanced aesthetic experiences. This makes rural tourism a key engine for China’s rural revitalization and urban–rural integration [1]. With the rapid development of the economy and the tourism industry, more and more traditional villages are undergoing large-scale tourist development. These villages fully showcase their unique natural environments and cultural economic value. The combination of traditional villages and tourism has proven to be an effective development model worldwide [2,3,4].

In recent years, several government documents, such as the “Rural Revitalization Strategic Plan (2018–2022)”, “Guidelines for Promoting Sustainable Development of Rural Tourism”, and the “14th Five-Year Plan for Tourism Development”, have been issued, setting forth requirements for the planning and construction of rural tourism and rural revitalization. These documents emphasize the importance of “inheriting and enhancing excellent traditional rural culture to boost rural tourism development”, “highlighting the cultural characteristics of rural tourism”, and “preserving the unique features of rural landscapes”. The aim is to protect the cultural heritage left by previous generations and retain the unique regional environments, cultural characteristics, and architectural styles of rural areas. The street space in traditional villages is not only a place for villagers to live and work but also an important component of tourist leisure and sightseeing. However, there are some issues in the construction of street and alley spaces in traditional villages, such as over-commercialization, homogenization, and arbitrary demolition or renovation [5,6]. These problems not only affect the visitor experience but also potentially lead to the loss of unique village spatial characteristics. From a planning and design perspective, a significant reason is the lack of scientific understanding, systematic extraction, and effective application of village characteristics.

To address these issues, the academician Duan [7] proposed incorporating spatial gene theory into planning and design methodologies for villages and towns, suggesting a scientific understanding and systematic refinement of village features [8,9]. Building on this, Wang Kai proposed a “Six-Step Method” for identifying and inheriting spatial genes in characteristic villages and towns. The identification and refinement of characteristic scenes and feature factors are key steps in planning and designing these unique spaces [10]. Some specific methods have been proposed for identifying characteristic factors. For example, Liu Jia and Xiao Yun extracted village characteristic scenes using on-site photos, historical records, aerial views, video footage, and textual descriptions, and then used element analysis to determine the spatial elements that constituted the village street scene [11]. Yang Liu used declarative and semantic memory from village residents to identify characteristic scenes and employed a combination of “image + semantics” to jointly refine the characteristic factors of village street spaces [12]. However, the perception of village characteristics is closely tied to human emotional perception. The common method for identifying characteristic scenes involves interviews and questionnaires [13,14], which are time-consuming and often fail to provide a clear visual representation of tourists’ “regional cultural identity” and “sense of place” in village street spaces [15,16]. Thus, there is an urgent need for a convenient, fast, and comprehensive method to identify tourists’ perceptions of street space characteristics, which will help in comprehensively extracting the spatial feature factors that impact tourists’ perceptions.

This study aims to explore methods for identifying and optimizing the spatial characteristics of traditional tourism villages to enhance visitor experiences and preserve the unique spatial features of these villages [8]. Firstly, the research examines variations in visitors’ perceptions of village street space characteristics under different psychological and behavioral states, focusing on differences in stay duration, crowd density, and emotional perception. Secondly, it analyzes the correlation between the architectural space elements of streets and visitor perceptions, identifying and extracting key spatial feature factors that influence visitor perceptions. Furthermore, this study proposes a convenient and comprehensive method to identify visitors’ perceptions of street space characteristics, addressing the limitations of traditional interview and survey methods to improve research efficiency and intuitiveness. Finally, the research supplements and refines the multidimensional framework for identifying human characteristics in street spaces and provides a systematic method for identifying characteristic factors in the study of spatial genes in Chinese characteristic villages. By achieving these objectives, this study provides a scientific basis for the protection and development of traditional villages, promoting the sustainable development of rural tourism.

2. Literature Review

2.1. Indicators of Spatial Environmental Elements of Streets

At present, the study of streets in general can be traced back to Kevin Lynch’s classic work “The Image of the City”, which explored the important theory of urban space [17], and Yoshinobu Ashihara’s “Street Aesthetics”, which puts forward that “the street is a part of life”. In “Street Aesthetics”, Ashihara suggests that “the street is a part of life” as an understanding of street space [18], and Jane Jacobs discusses how to improve street safety and create a good street environment through the interactions of residents [19]. Characteristics of street space design play an important role in the perception of pedestrians’ walking activities, and they affect the specific feelings of pedestrians about their surroundings when they walk through the environment [20,21,22,23]. Gehl, Whyte, and Montgomery, among others, have conducted generalized research on the spatial characteristics of streets from different perspectives. They have explored the design approach of streets and investigated how to use streets to provide better urban quality and vitality [24,25,26].

Subsequently, academics have conducted diverse studies on the spatial elements of streets that affect walking. Harvey analyzed the spatial variables related to street morphology [23]. Chen Yong, Xu Leiqing, Park, Nagata, Agnieszka, and other scholars have explored the elements of the spatial environment affecting pedestrians in terms of the street’s underlying architectural interface characteristics, building façades, greenery, street trees, infrastructure, and spatial scale [27,28,29,30,31,32]. Some scholars have calculated spatial measurements such as street aspect ratio, greening ornamental index, and sky visual factor [33,34,35]. Zhang Zhang and Long Ying established micro-built-environment indicators for historical streets. These indicators covered street and building scale, street transition space, characteristics of the street floor, merchandise furnishings, etc. They analyzed these indicators using mathematical modeling combined with the walking behavior of tourists, building upon previous research [36]. Shuangjin Li and Shuang Ma investigated the relationship between street spatial quality and built environment attributes. They utilized a large dataset of streetscape images to identify the influencing elements of street location, morphology, and functional data [37]. Furthermore, Xiang Gao and Zao Li used GPS motion tracking to explore the correlation between elemental indicators of the spatial environment of traditional villages and tourists’ behavioral preferences [2]. Zhuang Weimin, Liu, and Nobuyoshi et al. explored the correlation between streets’ built environment and tourist vitality status with the help of street-view image data and deep-learning technology [38,39]. In summary, the abovementioned domestic and foreign scholars, exploring a large number of street spatial elemental indicators affecting pedestrians, have provided a sufficient basis for the study of the associations between people and the spatial environment.

2.2. Measures of Pedestrian Perception

Measuring human perception and the characteristics of the built environment has long been of interest in a variety of fields. The link between the environment and human perception has been considered for decades in a variety of disciplines and fields, including geography, urban planning, environmental psychology, and neuroscience [40,41,42]. With the development of artificial intelligence and new technologies, there are new ways to diversify perceptual perspectives. In 2013, the MIT Media Lab launched the “Place Pulse” project, a data collection platform for volunteers to participate in urban perceptual rating experiments, which utilizes deep-learning techniques to analyze perceptions of urban appearance through human vision [43]. Yunqin Li and Nobuyoshi Yabuki proposed a deep-learning model. It is based on video image data to study the evaluation of tourist vitality status [44]. Additionally, Zhang et al. established six human perception indicators of streets [45]. They used a deep-learning model to evaluate the built environment features that affect humans’ sense of place. Tianyue Li et al. semantically segmented street images. They collected different perception scores of streets and lanes. They analyzed the spatial patterns of spatial quality perception of urban streets through global Moran’s I and GIS hotspot analysis [46]. Fang Zhiguo et al. conducted an association study of Shanghai streets using crowd visual perception [47]. Chen Zheng, Ren Hongguo, Huang Yihua, and other scholars analyzed individual behavior and crowd perception characteristics using eye-tracking and head-mounted devices to capture image data [48,49,50]. Salesses [51], Naik [52], Zhang [45], and their team used data from the Urban Pulse project for training. They combined street view images (SVIs) with machine-learning techniques to evaluate the aesthetics, safety, and pleasantness of street spaces. Scholars Liu Lun et al. focused on the characteristics and recognizability of street appearance. Using machine-learning techniques and large-scale street view images, they conducted a large-scale automated evaluation of the quality of the urban environment, specifically assessing the construction and maintenance quality of building façades and the continuity of street walls [39]. Noji, Kishimoto, Kanyou Sou, Hiroya Shiokawa, and other scholars shifted from visual perception to psychological perception of facial expressions, providing a new methodology and idea for capturing pedestrian perceptions [53,54].

In summary, research exploring the impact of streets’ spatial environments on pedestrians can be divided into two categories: one focuses on extracting and exploring the factors related to streets’ spatial environment elements, and the other explores the behavioral and perceptual characteristics of pedestrians from various angles. However, most research findings on street environments are centered on urban settings, with a bias toward studying urban street element indicators while neglecting villages, particularly commercialized street spaces in tourism-oriented villages. Moreover, due to the high availability of urban data, many studies utilize street imagery or point-of-interest (POI) data sources [55], but the singularity and the difficulty in obtaining rural street data limit big-data approaches in this context. As a result, there has been less exploration of flexible and varied data research methods targeting rural areas. Additionally, the academic focus has been on street pedestrians’ vitality and behavior, with limited exploration of the psychological perceptions of tourists, especially those with strong tourism motivations. There has also been a lack of research into new AI technologies to study such perceptions. Since the introduction of the spatial gene [8] theory in 2019 and the study of characteristic Chinese villages and towns [9], research on the refinement of elements affecting tourism-oriented street spaces remains insufficient. This gap makes the technical guidance for village space construction and planning inadequate. Therefore, this study builds on the existing literature by focusing on the street space of traditional villages. Unlike previous research that primarily targets urban environments, this study aims to fill the gap in understanding tourists’ perceptions of the quality of street spaces in traditional villages. By employing a deep-learning model, it accurately analyzes tourists’ behaviors and perceptions in the context of traditional village environments. This methodology can identify key spatial features that enhance the tourist experience, providing a comprehensive framework for scientifically preserving and developing traditional village spaces while maintaining their unique cultural and environmental value. This not only broadens the current research perspective but also offers new methodological support for the tourism development and spatial planning of traditional villages.

3. Materials and Methods

3.1. Materials

3.1.1. Case Selection

Yuanjia Village, situated in Liquan County, Xianyang City, Shaanxi Province, was part of the second batch of traditional villages included in China’s traditional village list. It was a natural village with a rich history located in the Guanzhong District of Shaanxi Province (Figure 1). The research value of the Yuanjia Village case study can be highlighted in the following three aspects:

• The village has a long history with a well-preserved spatial layout. In the second year of the Northern Song Dynasty’s Jianlong period (961), the Yuan family relocated here to escape conflict and established a village through clan-based settlements. To this day, it retains many historical buildings, such as Baoning Temple, Bishan Hall, Ronghua Pavilion, and Kangzhuang Archway. Additionally, Yuanjia Village’s site selection and layout centered around its original village base, adhering to the principle of “embracing the yin and welcoming the yang, with mountains at the back and water at the front”, reflecting the traditional landscape pattern with surrounding mountains and flowing waters. Furthermore, the street spaces in Yuanjia Village were arranged in a neat and orderly manner. The alignments were straight and perpendicular. The design accommodated the varying terrain with different elevations, which created a harmonious and structured appearance. The traffic orientation, airflow, drainage, and ventilation in the street space all depended on its unique topography and landform. This ensured that its architectural style and features aligned with the principles of humanization, nature, and ecological sustainability [56]. Consequently, Yuanjia Village successfully maintained and continued the traditional spatial pattern of the Guanzhong region quite well.
• In 2007, Yuanjia Village, building on its existing historical resources and with villagers as the main focus, created a village-themed integrated tourism area centered on Guanzhong folk culture. The plan encompassed six major business types: food and dining, workshops, handicrafts, entertainment and leisure, guesthouses, and e-commerce. This initiative resulted in the establishment of a culturally rich and distinct “Guan Zhong Impression Experience Area”. Within this framework, Yuanjia Village managed to preserve various food-processing workshops and factories. These vividly showcased the working life of Guanzhong villagers, ultimately becoming the most profitable business type. This diverse range of tourism business types significantly contributed to the spatial diversity of the village.
• As of 2023, the village had received over 8 million visitors, with annual tourism revenue exceeding 1.2 billion RMB [57]. This made it the most-visited village in the entire Shaanxi region throughout the year. Given the consistent growth in annual visitor numbers and tourism revenue, Yuanjia Village is considered one of the most successful examples of traditional villages transitioning to tourism. The large number of visitors to Yuanjia Village also provides a high level of data reliability for sampling tourists’ perceptions due to the extensive and representative sampling pool.

Therefore, based on the above characteristics, Yuanjia Village provided ample evidence for the experiment, and the results of the experiment were generally applicable and persuasive.

3.1.2. Data Collection

In view of the characteristics of Yuanjia Village’s economic industry and space, six typical streets were selected for data collection (Figure 2): the Ancestral Hall street (CT = 248 m), Hui Min street (HM = 156 m), Snack street (XC = 428 m), SanQin ancient street (SQ = 704 m), ShuXuan street (SY = 413 m), and KangZhuang old street (KZ = 383 m).

The primary data collected included crowd behavior data, point-of-interest (POI) data, and field survey data. The crowd behavior data included tourist facial expressions, stay duration, and crowd aggregation. Tourist facial expressions and stay duration were analyzed using photo and video data from the TwoStep app. Crowd aggregation was captured through drone video recordings from fixed positions. In addition, commercial facility POI data were obtained from Bigmap Gis Office software in 2023. Based on feedback from field surveys, approximately 230 effective POIs were related to Yuanjia Village’s core scenic area, which was supplemented through additional field research, resulting in a total of 258 confirmed POIs. These 258 POIs were categorized into six types: hotels, food shops, shopping stores, scenic spots, entertainment shops, and public service facilities. This study focuses on the microscale analysis of street spaces, and by conducting field surveys to correct data, the results are more reliable and convincing.

3.2. Methods

This paper categorizes street elements through the composition of a “character scene” [12,58]. In the scene, there are street elevations and cross-sections that control the framework of the scene, which correspond to the street’s scale and form elements, such as street width and aspect ratio. The street’s ground-floor interface is the rich boundary that frames the scene, such as the flat ground-floor elevation and spatial separation. The street furniture and public activity space in the scene play a decorative role, corresponding to the environmental facility elements. The street business feature elements constitute the diverse functional connotations of the scene [32,46,48,56].

The experiment captured the data of tourists walking, staying, and gathering in the street through drone video and mobile cell phone cameras in order to quantify the emotional perceptions of tourists. Annotated by the machine-learning model, the human feelings about the use of the material space environment were transformed into a clear recognition model of artificial intelligence, which can fully simulate the human psychological feelings about the street environment and provide a more comprehensive understanding of the characteristic spatial scene of the street. Secondly, multidimensional statistics of the street and alley material space data, commercial business data, and other data were measured in combination with the tourism characteristics of Yuanjia Village. On the basis of the measurement analysis, we analyzed the distribution law and characteristics of the tourists’ perceptions in the street and alley space, and then we used SPSS to construct the mathematical and theoretical model of perception evaluation and the spatial scene, explain its correlation according to the multiple regression model, and preliminarily condense the featured factors (Figure 3).

3.2.1. Deep-Learning-Based Multidimensional Perception Measurement for Tourists

• Measurement of the degree of tourist aggregation

The collection of data on tourist aggregation utilized drone video data and YOLOv5’s machine-learning algorithm [53,59,60] to identify a large number of drone video photos and extract tourist behavior trajectories [61,62]. First, we wrote and executed a script in Python, using PyTorch and Keras as deep-learning frameworks, OpenCV for video processing, pandas and NumPy for data analysis, YOLOv5 as the object-detection network model, and the DeepSORT algorithm to capture real-time pedestrian trajectory data. Finally, we used ArcGIS 10.0 to create buffer zones for the main street spaces in Yuanjia Village (Figure 4). The degree of tourist aggregation (G) is reflected by the kernel density derived from the trajectory extraction, as indicated in Formula (1):

$$\widehat{G}n(x)=\frac{1}{nh}\sum _{i=1}^{n}K\left(\frac{s_i-x}{h}\right)$$

(1)

where x is any variable from the tourist’s tour trajectory data, X represents the number of data points in the trajectory data, and h is the smoothing bandwidth controlling the degree of smoothing—the larger the h value, the smoother the data, and h > 0. K is the kernel density function, and S_i represents a data point in the trajectory data.

2.

Facial emotion measurement of tourists

During their visit, tourists’ facial expressions, influenced by the microscale street environment, can reflect their psychological perception state. This study utilized the YOLOv5 object-detection model. The YOLO network consists of convolutional layers specifically designed for object-detection tasks, including three key modules shown in Figure 5. The machine-learning training data originated from YOLO’s built-in dataset tools [56,57].

To set up the execution environment for YOLO, we first used Conda. Then, within the environment, the main code file of the labeling Python annotation tool was executed. By recognizing and statistically analyzing four types of emotion symbols—smile, laugh, neutral, and negative emotion—the distribution of facial expressions at different street locations was determined. This allowed for the calculation of the subjective smile rate (E) in various street locations, based on the total recognized facial expressions. See Formula (2) for more details:

E = (∑Smile + ∑Lugh)/(∑Smile + ∑Lugh + ∑Neutral + ∑Negative Emotion)

(2)

3.

Multidimensional evaluation of tourists’ comprehensive perception of the spatial characteristics of streets

The comprehensive characteristic perception index (S) for tourists is influenced by the degree of aggregation (G), the stay time (D), and the smile rate (E). The degree of aggregation was derived from the kernel density of tourist trajectories (see Formula (1)), the stay duration was calculated from video and photo analysis, and the smile rate was determined through facial recognition (see Formula (2)). By combining tourist interviews and extracting the subjective perception weights for these three factors (β1 = 0.52, β2 = 0.31, β3 = 0.17), we obtained the formula for calculating the perception index (see Formula (3)). This index was then used to assign values to the POIs extracted from the street space. The higher the score, the stronger the characteristic perception, indicating a more distinct and engaging experience for tourists as they navigate the village’s streets and shop surroundings.

Subjective = β₁ × D + β₂ × E + β₃ × G

(3)

3.2.2. Indicators of Elements of Street Spatial Scene Based on Spatial Gene Theory

• Selection of street element indicators

In this study, we reference the work of experts such as Wang Kai, Xu Leiqing, Christopher Alexander, Zhang Zhang, Long Ying, and Fang Zhiguo on the street index measurement system. The street elements were classified through the composition of “characteristic scenes”. The framework of a scene includes street façades and cross-sections, corresponding to the streets’ scale and morphological elements, such as street width and aspect ratio. The street-level interface forms a rich boundary for the scene, such as flat ground-floor façades and spatial separations. The street furniture and the historical reproduction of the scene play a crucial role in creating the atmosphere, attracting tourists, and enhancing their experience, corresponding to the environmental facility elements. The business composition in the street forms the functional diversity of the scene.

Based on these references and the premise of data availability, the indicators affecting street characteristic evaluation were divided into four dimensions: (1) street scale and form, including street width, height-to-width ratio, and the relationship between streets and landforms; (2) characterization of environmental amenity elements, including street furniture as well as public space; (3) street floor characteristics, including transitional space, permeability, and curvature; and (4) street business feature elements, including the total number of businesses, density, mixing, and commercial in situ value. Thus, under the premise of the literature base and data availability, combined with the above analysis, the author divided the indicators affecting the evaluation of street scene perception into eleven major categories and twenty-two subcategories (

Table 1. Indicators for measuring the four categories of street elements: scale and form, environmental amenities, underlying character, and business.

Primary Indicator	Secondary Indicators	Description	Labeled Name	Assignment Description	Value Assignment
					0	1	Absolute Value
Street scale and form	Scale of streets	Width	KD	Street width			From Measurement data
		Ratio of street height to width	GKB	Average building height on both sides/street width
	Street morphology	Relationship of streets to landforms	JX	Responding to the terrain	No	Yes
Characterization of environmental amenity elements	Public space	Iconic space	BZXKJ	Responding to the terrain	No	Yes
		Space for daily activities	ECHDKJ	Theater historic buildings	No	Yes
	Street furniture	Commercial characteristic processing tools	SYTSJGGJ	Streets small plazas	No	Yes
		Regional characteristic furnishings	DYTSBS		No	Yes
		Tables, chairs, and stools	ZYBD		No	Yes
		Umbrellas/awnings	ZYS	Local materials	No	Yes
		Landscape water system	JGSX	Local materials	No	Yes
		Greenery	LZ		No	Yes
Street floor characteristics	Transition space	Ground-level elevation	DMTS	Local materials	No	Yes
		Space separation	KJFG	Significant height difference between transitional space and street space	No	Yes
		Façade setbacks/extensions	LMHT/WY	Separation of spaces from the street through parapets, low walls, planters	No	Yes
	Permeability	Longitudinal subgrade	ZSDM	Ground-floor openings/extension to interior	No	Yes
		Longitudinal cross-roads	ZXJCDL	Vertical alleyway entrance at ground floor	No	Yes
	Curvature	Straight bottom elevation	PZDCLM	Vertical projection of ground-floor elevation is straight	No	Yes
Street floor characteristics	Curvature	Curvature curvilinear ground floor	QZDCLM	Ground-floor façade is clearly angled, projecting, or recessed	No	Yes
Street business feature elements	Type of business	Six business types: hotel, food, shopping, attractions, entertainment, and public services	F	Assign values 1–6			From POI data
	Business density	Density of commercial POI points per 10 m street buffer zone	FD	Formula (4)			From POI data
	Business mixing degree	Street mix (diversity) is the mix of vitality-related POIs in the street buffer.	FH	Formula (5)			From POI data
	Commercial in situ value	Percentage of street and street business densities that are self-producing and self-selling	FL	Formula (6)			From POI data

).

2.

Measurement and statistics of factor indicators

(1)

Elements of street scale and morphology: The data analysis included street scale, focusing on street width and street height, which were measured using infrared rangefinders. Because the case area belongs to the Loess Plateau terrace and there are variations in the topography, the relationship between street morphology and topography was also taken into account as one of the elements (Figure 6).

(2)

Environmental facility elements: First, public space. Through the on-site research, the public space was divided into iconic space and space for daily activities, and the theater, historical buildings, squares, and streets within the space were taken into account as evaluation elements. Second, street facilities. The clear transition space was matched with diverse street furniture arrangements, and according to different functional attributes there were commercial characteristic processing tools, regional characteristic furnishings, tables, chairs, benches, umbrellas, landscaped water systems, and greenery on the street, which can be used as evaluation elements (Figure 6).

(3)

Elements of the ground floor along the street: First, transitional space perception. Changes in terrain undulation bring about changes in the height difference between the transition space and the street space, and the perception of the transition space between the street and the building can be strengthened through the height difference between the street and the building, while the indicator of the ground elevation can show whether or not the change in height difference is significant. In terms of spatial separation, we considered whether there was spatial separation from the street through parapets, low walls, flower pools, etc. In terms of building elevation, we observed whether the building elevation or the ground floor had been set back and extended, thereby expanding the space. Second, permeability. From the vertical direction of the street space, we analyzed and identified whether the ground floor of the building was open or extended to the interior as well as whether there were intersections and vertical street entrances in the road, and we focused on observing whether the building façade along the street penetrated into the interior space of the featured scene. Third, the degree of curvature and straightness. The straightness and curvature of the ground-floor façade refers to whether or not the vertical projection of the ground-floor façade is straight, with or without obvious folded corners, protrusions, or depressions (Figure 6).

(4)

Street business feature elements: The “street business” in this paper refers to the diversified business forms formed to meet different consumer needs, i.e., the proportion of different types of stores, production methods, and processing methods. Taking into account the characteristics of Yuanjia Village’s tourism-oriented streets, we constructed a pattern measurement system with indicators including the number of patterns, pattern density [63], pattern localization, and the mixture of pattern production types [45]. (The POI data of street and street business feature elements were crawled from BIGEMAP map in 2023).

Formation type (F): This included six types of formations in the street and its industry—hotel, food, shopping, attractions, entertainment, and public services (assigning values according to the type, corresponding to the numbers 1–6).

Formal density (FD): Due to the short length of the streets in the case study object, the density of commercial POIs in the buffer zone of every 10 m street was used as a measure of formal density; see Formula (4):

FD = ∑POI num/Road length

(4)

Business mixing degree (FH): Street business mix (diversity) is the mix of vitality-related POIs in the street buffer, which can be calculated using the information entropy value; see Formula (5):

FH(X) = −∑P_(xi)logP_(xi) (i = 1,2,…n)

(5)

where X denotes a random certain type of business POI, P_(xi) denotes the relative ratio of the total number of street segments where a certain business type is located, and n denotes the type of the street POI. The higher the information entropy, the greater the mixing degree and the higher the diversity.

The degree of localization of industry (FL): Combined with the research process of characteristic industries in Yuanjia Village, a part of the industry is secondary processing of the crops around the village through village-run factories and then selling them in stores, so this paper puts forward a method for calculating the proportion of this type of business. That is, the percentage of the total number of industry types that utilize local resources for self-production and self-sale in every 10 m street buffer is shown in Formula (6):

FL = ∑POI_local/(∑POI_num × Road_length)

(6)

4. Results

4.1. Visitor Characteristics Perception Analysis

Tourists’ multidimensional perception encompassed aggregation degree, stay duration, and emotional feedback. This study used a deep-learning-based method to measure tourists’ multidimensional perception and calculated perception scores. The characteristic perception scores were then normalized, showing that only 5.1% of the street spaces had scores above 0.8, 9.3% between 0.5 and 0.8, 28% between 0.5 and 0.3, and 57.6% below 0.3. The mean score was 0.3011, the median was 0.2474, and the standard deviation was 0.2424. This suggests that, overall, tourists’ perception of street scene features was generally acceptable. Using streets as the visualization unit, the ArcGIS software connected image points and their perception scores to the street, resulting in the characteristic perception map for Yuanjia Village’s street scenes in Figure 7.

Spatially, XC and HM scored higher (Figure 7 (4)), while KZ was more distinctive than CT and SY (Figure 7 (3)). In terms of street function, the streets’ function is primarily centered on local food in Yuanjia Village. Streets such as XC-1, XC-2, XC-3, and XC-4 had characteristic perception scores of 0.478, 0.647, 0.484, and 0.543, respectively. This indicates that tourists tended to prefer street experiences that offered unique culinary delights. Although CT also features Shaanxi’s style of food, it did not achieve the same level of success. Furthermore, SY-3 and SQ-2 had the lowest characteristic perception scores, with averages of 0.091 and 0.122, respectively (Figure 7 (1)). SY-3 is Yuanjia Village’s bar street, with limited operating hours, which contributed to a decrease in tourist attraction. In Figure 7 (6), SQ-2 represents newly constructed commercial spaces in Yuanjia Village. Some sections feature large-scale structures, while certain business types, such as retail, alcohol, tobacco, and imitation handicrafts, lack distinctive characteristics. This contributed to reduced tourist appeal and lower overall perception.

4.2. Correlation Analysis of Behavioral Perception and Street Elements

4.2.1. Multiple Linear Regression Analysis

Table A1. The correlation coefficients for the 21 independent variables.

Correlation
		GKB	JX	BZXKJ	CRHDKJ	DMTS	KJFG	LMHT/WY	SYTSJGGJ	DYTSBS	ZYIBD	ZYS	JGSX	LZ	ZSDM	ZXJCDL	PZDCLM	QZDCLM	F	FD	FH	FL
GKB	Pearson correlation	1.00	0.05	0.307 **	0.02	0.05	−0.230 **	0.04	0.10	0.329 **	0.144 *	0.224 **	0.125 *	00.07	0.185 **	−0.07	−0.159 *	−0.218 **	0.01	0.133 *	−0.142 *	0.150 *
	Significance (two-tailed)		0.43	0.00	0.78	0.42	0.00	0.49	0.10	0.00	0.02	0.00	0.04	0.26	0.00	0.24	0.01	0.00	0.86	0.03	0.02	0.02
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
JX	Pearson correlation	0.05	1.00	−0.171 **	−0.01	0.321 **	0.11	0.195 **	0.01	0.149 *	0.448 **	0.165 **	0.474 **	0.409 **	−0.141 *	0.384 **	−0.12	−0.08	0.02	−0.161 **	−0.316 **	−0.253 **
	Significance (two-tailed)	0.43		0.01	0.90	0.00	0.07	0.00	0.93	0.02	0.00	0.01	0.00	0.00	0.02	0.00	0.05	0.19	0.72	0.01	0.00	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
BZXKJ	Pearson correlation	0.307 **	−0.171 **	1.00	−0.135 *	−0.11	−0.08	0.05	0.157 *	0.486 **	0.261 **	0.364 **	−0.05	0.00	−0.05	−0.10	−0.209 **	−0.225 **	−0.02	0.335 **	−0.12	0.256 **
	Significance (two-tailed)	0.00	0.01		0.03	0.08	0.18	0.43	0.01	0.00	0.00	0.00	0.42	0.94	0.42	0.12	0.00	0.00	0.80	0.00	0.05	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
CRHDKJ	Pearson correlation	0.02	−0.01	−0.135 *	1.00	0.03	−0.144 *	−0.03	−0.216 **	−0.09	0.03	0.05	−0.04	−0.02	0.10	0.03	0.137 *	0.148 *	−0.05	−0.180 **	0.241 **	−0.213 **
	Significance (two-tailed)	0.78	0.90	0.03		0.69	0.02	0.66	0.00	0.17	0.68	0.41	0.49	0.72	0.13	0.67	0.03	0.02	0.41	0.00	0.00	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
DMTS	Pearson correlation	0.05	0.321 **	−0.11	0.03	1.00	−0.11	0.227 **	0.00	0.06	0.163 **	0.229 **	0.260 **	0.271 **	0.01	0.135 *	0.02	0.01	0.173 **	−0.234 **	0.06	−0.08
	Significance (two-tailed)	0.42	0.00	0.08	0.69		0.09	0.00	0.96	0.38	0.01	0.00	0.00	0.00	0.81	0.03	0.71	0.90	0.01	0.00	0.30	0.21
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
KJFG	Pearson correlation	−0.230 **	0.11	−0.08	−0.144 *	−0.11	1.00	−0.09	−0.397 **	−0.244 **	−0.05	−0.331 **	0.369 **	0.353 **	0.11	0.312 **	0.172 **	0.443 **	0.06	−0.433 **	0.173 **	−0.469 **
	Significance (two-tailed)	0.00	0.07	0.18	0.02	0.09		0.13	0.00	0.00	0.40	0.00	0.00	0.00	0.07	0.00	0.01	0.00	0.30	0.00	0.01	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
LMHT/WY	Pearson correlation	0.04	0.195 **	0.05	−0.03	0.227 **	−0.09	1.00	0.09	0.221 **	0.255 **	0.288 **	0.00	−0.02	0.01	0.135 *	−0.229 **	−0.193 **	0.11	−0.11	0.149 *	−0.08
	Significance (two-tailed)	0.49	0.00	0.43	0.66	0.00	0.13		0.13	0.00	0.00	0.00	0.94	0.73	0.82	0.03	0.00	0.00	0.07	0.09	0.02	0.22
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
SYTSJGGJ	Pearson correlation	0.10	0.01	0.157 *	−0.216 **	0.00	−0.397 **	0.09	1.00	0.467 **	−0.08	0.08	−0.12	−0.198 **	−0.308 **	−0.247 **	−0.305 **	−0.631 **	−0.12	0.706 **	−0.626 **	0.674 **
	Significance (two-tailed)	0.10	0.93	0.01	0.00	0.96	0.00	0.13		0.00	0.18	0.21	0.06	0.00	0.00	0.00	0.00	0.00	0.05	0.00	0.00	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
DYTSBS	Pearson correlation	0.329 **	0.149 *	0.486 **	−0.09	0.06	−0.244 **	0.221 **	0.467 **	1.00	0.326 **	0.411 **	0.209 **	0.238 **	−0.03	0.09	−0.275 **	−0.328 **	−0.04	0.380 **	−0.424 **	0.355 **
	Significance (two-tailed)	0.00	0.02	0.00	0.17	0.38	0.00	0.00	0.00		0.00	0.00	0.00	0.00	0.59	0.17	0.00	0.00	0.52	0.00	0.00	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
ZYIBD	Pearson correlation	0.144 *	0.448 **	0.261 **	0.03	0.163 **	−0.05	0.255 **	−0.08	0.326 **	1.00	0.810 **	0.238 **	0.187 **	−0.323 **	0.418 **	0.00	0.10	−0.04	0.00	−0.08	−0.11
	Significance (two-tailed)	0.02	0.00	0.00	0.68	0.01	0.40	0.00	0.18	0.00		0.00	0.00	0.00	0.00	0.00	0.96	0.11	0.58	0.97	0.21	0.08
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
ZYS	Pearson correlation	0.224 **	0.165 **	0.364 **	0.05	0.229 **	−0.331 **	0.288 **	0.08	0.411 **	0.810 **	1.00	0.05	0.01	−0.279 **	0.290 **	−0.04	−0.07	0.03	0.132 *	−0.03	0.06
	Significance (two-tailed)	0.00	0.01	0.00	0.41	0.00	0.00	0.00	0.21	0.00	0.00		0.46	0.82	0.00	0.00	0.48	0.27	0.62	0.03	0.67	0.30
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
JGSX	Pearson correlation	0.125 *	0.474 **	−0.05	−0.04	0.260 **	0.369 **	0.00	−0.12	0.209 **	0.238 **	0.05	1.00	0.902 **	0.12	0.441 **	0.00	0.12	0.137 *	−0.261 **	−0.342 **	−0.250 **
	Significance (two-tailed)	0.04	0.00	0.42	0.49	0.00	0.00	0.94	0.06	0.00	0.00	0.46		0.00	0.05	0.00	0.96	0.06	0.03	0.00	0.00	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
LZ	Pearson correlation	0.07	0.409 **	0.00	−0.02	0.271 **	0.353 **	−0.02	−0.198 **	0.238 **	0.187 **	0.01	0.902 **	1.00	0.183 **	0.404 **	−0.01	0.170 **	0.11	−0.324 **	−0.253 **	−0.323 **
	Significance (two-tailed)	0.26	0.00	0.94	0.72	0.00	0.00	0.73	0.00	0.00	0.00	0.82	0.00		0.00	0.00	0.82	0.01	0.07	0.00	0.00	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
ZSDM	Pearson correlation	0.185 **	−0.141 *	−0.05	0.10	0.01	0.11	0.01	−0.308 **	−0.03	−0.323 **	−0.279 **	0.12	0.183 **	1.00	−0.137 *	−0.01	0.12	0.125 *	−0.247 **	0.179 **	−0.10
	Significance (two-tailed)	0.00	0.02	0.42	0.13	0.81	0.07	0.82	0.00	0.59	0.00	0.00	0.05	0.00		0.03	0.87	0.06	0.04	0.00	0.00	0.12
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
ZXJCDL	Pearson correlation	−0.07	0.384 **	−0.10	0.03	0.135 *	0.312 **	0.135 *	−0.247 **	0.09	0.418 **	0.290 **	0.441 **	0.404 **	−0.137 *	1.00	0.05	0.269 **	0.02	−0.349 **	0.07	−0.383 **
	Significance (two-tailed)	0.24	0.00	0.12	0.67	0.03	0.00	0.03	0.00	0.17	0.00	0.00	0.00	0.00	0.03		0.47	0.00	0.80	0.00	0.27	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
PZDCLM	Pearson correlation	−0.159 *	−0.12	−0.209 **	0.137 *	0.02	0.172 **	−0.229 **	−0.305 **	−0.275 **	0.00	−0.04	0.00	−0.01	−0.01	0.05	1.00	0.587 **	0.01	−0.236 **	0.230 **	−0.165 **
	Significance (two-tailed)	0.01	0.05	0.00	0.03	0.71	0.01	0.00	0.00	0.00	0.96	0.48	0.96	0.82	0.87	0.47		0.00	0.85	0.00	0.00	0.01
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
QZDCLM	Pearson correlation	−0.218 **	−0.08	−0.225 **	0.148 *	0.01	0.443 **	−0.193 **	−0.631 **	−0.328 **	0.10	−0.07	0.12	0.170 **	0.12	0.269 **	0.587 **	1.00	0.08	−0.599 **	0.493 **	−0.498 **
	Significance (two-tailed)	0.00	0.19	0.00	0.02	0.90	0.00	0.00	0.00	0.00	0.11	0.27	0.06	0.01	0.06	0.00	0.00		0.18	0.00	0.00	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
F	Pearson correlation	0.01	0.02	−0.02	−0.05	0.173 **	0.06	0.11	−0.12	−0.04	−0.04	0.03	0.137 *	0.11	0.125 *	0.02	0.01	0.08	1.00	−0.215 **	0.08	−0.136 *
	Significance (two-tailed)	0.86	0.72	0.80	0.41	0.01	0.30	0.07	0.05	0.52	0.58	0.62	0.03	0.07	0.04	0.80	0.85	0.18		0.00	0.19	0.03
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
FD	Pearson correlation	0.133 *	−0.161 **	0.335 **	−0.180 **	−0.234 **	−0.433 **	−0.11	0.706 **	0.380 **	0.00	0.132 *	−0.261 **	−0.324 **	−0.247 **	−0.349 **	−0.236 **	−0.599 **	−0.215 **	1.00	−0.661 **	0.818 **
	Significance (two-tailed)	0.03	0.01	0.00	0.00	0.00	0.00	0.09	0.00	0.00	0.97	0.03	0.00	0.00	0.00	0.00	0.00	0.00	0.00		0.00	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
FH	Pearson correlation	−0.142 *	−0.316 **	−0.12	0.241 **	0.06	0.173 **	0.149 *	−0.626 **	−0.424 **	−0.08	−0.03	−0.342 **	−0.253 **	0.179 **	0.07	0.230 **	0.493 **	0.08	−0.661 **	1.00	−0.494 **
	Significance (two-tailed)	0.02	0.00	0.05	0.00	0.30	0.01	0.02	0.00	0.00	0.21	0.67	0.00	0.00	0.00	0.27	0.00	0.00	0.19	0.00		0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00
FL	Pearson correlation	0.150 *	−0.253 **	0.256 **	−0.213 **	−0.08	−0.469 **	−0.08	0.674 **	0.355 **	−0.11	0.06	−0.250 **	−0.323 **	−0.10	−0.383 **	−0.165 **	−0.498 **	−0.136 *	0.818 **	−0.494 **	1.00
	Significance (two-tailed)	0.02	0.00	0.00	0.00	0.21	0.00	0.22	0.00	0.00	0.08	0.30	0.00	0.00	0.12	0.00	0.01	0.00	0.03	0.00	0.00
	N	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00	258.00

** Significant at the 0.01 level (two-tailed) for correlation. * Significant at the 0.05 level (two-tailed) for correlation.

in Appendix A displays the Pearson correlation coefficients among 21 independent variables. Generally, a correlation coefficient greater than 0.7 indicates potential multicollinearity between variables. In this case, the results show that the correlation coefficients among the independent variables are all less than 0.7, indicating no multicollinearity. The 21 indicators for this street were treated as independent variables, while the dependent variable was the subjective evaluation score for street characteristic perception. A multiple linear regression analysis was used to determine the impact of street indicators on street characteristic perception evaluation. The formula is as follows:

Y(Subjective) = β₀ + β₁ × GKB + β₂ × JX + β₃ × BZXKJ + β₄ × RCHDKJ + β₅ ×
DMTS + β₆ × KJFG + β₇ × LMHT/WY + β₈ × SYTSJGGJ + β₉ × DYTSBS + β₁₀ ×
ZYIBD + …. + β₂₁ × FL + e

(7)

where β₁–β₂₁ represent the regression coefficients, β₀ is the constant term, and e is the error term. The post-regression tolerance value = 1/VIF, and all tolerance values were greater than 0.1, indicating no multicollinearity. Model D-W = 1.606, which indicates that the data satisfy the independence (

Table 2. Model summary.

Model	R	R2	Adjusted R2	Standard Skewness Error	Durbin–Watson
1	0.848 a	0.719	0.693	0.08892682	1.606

^a Predicted value: (constant), FL, ZYS, F, PZDCLM, CRHDKJ, DMTS, GKB, JGSX, LMHT_WY, ZSDM, BZXKJ, JX, ZXJCDL, DYTSBS, KJFG, QZDCLM, SYTSJGGJ, FH, ZYIBD, LZ, FD. Note: Dependent variable: S.

). In the ANOVA analysis, F = 28.576 and p = 0.000 < 0.05, which is statistically significant. The regression model passed the hypothesis test, R² = 0.642 > 0.5, indicating that the regression model was well fitted (

Table 3. Analysis of variance (ANOVA) ^a.

Model		Sum of Squares	df	Mean Square (MS)	F	Significance
1	Regression	4.745	21	0.226	28.576	0.000 b
	Residuals	1.858	235	0.008
	Statistics	6.604	256

^a Predicted value: (constant), FL, ZYS, F, PZDCLM, CRHDKJ, DMTS, GKB, JGSX, LMHT_WY, ZSDM, BZXKJ, JX, ZXJCDL, DYTSBS, KJFG, QZDCLM, SYTSJGGJ, FH, ZYIBD, LZ, FD; ^b Dependent variable: S.

). Thus, a meaningful statistical model was constructed with a good degree of explanation.

• Street Scale and Form

The correlation between tourist perception evaluation and the factors related to street scale and morphology was found to be relatively weak. In general, street width and aspect ratio are significant in affecting street enclosure, but in the case of Yuanjia Village, this enclosure did not appear to influence individuals’ characteristic evaluation. Upon analyzing the reasons, it was observed that the historical streets and lanes in Yuanjia Village, which were built early on, have a D/H ratio of about 1. The later developments, such as Sanqin Ancient Street and Shuyuan Street, also adhere to a D/H ratio of around 1. Even though Sanqin Ancient Street is about four stories high (approximately 12 m), its width generally varies between 10 and 15 m. On the other hand, the standardized coefficient for the JX‘s Beat was negative (

Table 4. Results of linear regression analysis between tourists’ subjective feature perception and objective spatial elements.

Model		Unstandardized Coefficients		Standardized Coefficients	T	Significance	Collinearity Statistics
		B	Standard Error	Beta			Tolerance	VIF
1	(constant)	0.316	0.094		3.364	0.001
	GKB	−0.001	0.005	−0.001	−0.016	0.987	0.707	1.414
	JX	−0.020	0.020	−0.059	−0.983	0.327	0.334	2.994
	BZXKJ	−0.003	0.020	−0.008	−0.156	0.876	0.500	1.998
	CRHDKJ	0.019	0.017	0.043	1.117	0.265	0.797	1.254
	DMTS	0.008	0.018	0.021	0.475	0.635	0.608	1.646
	KJFG	−0.018	0.020	−0.048	−0.887	0.376	0.413	2.422
	LMHT_WY	0.026	0.018	0.062	1.427	0.155	0.642	1.557
	SYTSJGGJ	0.016	0.023	0.048	0.685	0.494	0.242	4.125
	DYTSBS	−0.024	0.023	−0.060	−1.061	0.290	0.375	2.669
	ZYIBD	−0.045	0.029	−0.140	−1.594	0.112	0.154	6.484
	ZYS	0.028	0.027	0.087	1.043	0.298	0.171	5.842
	JGSX	0.047	0.032	0.147	1.499	0.135	0.125	8.011
	LZ	−0.005	0.031	−0.015	−0.159	0.874	0.134	7.443
	ZSDM	−0.058	0.016	−0.162	−3.564	0.000	0.580	1.723
	ZXJCDL	−0.045	0.015	−0.140	−2.934	0.004	0.526	1.901
	PZDCLM	0.041	0.015	0.128	2.827	0.005	0.581	1.720
	QZDCLM	−0.056	0.020	−0.175	−2.755	0.006	0.298	3.353
	F	−0.005	0.006	−0.030	−0.807	0.420	0.860	1.163
	FD	0.146	0.164	0.090	0.890	0.374	0.118	8.507
	FH	−0.258	0.162	−0.137	−1.589	0.113	0.160	6.254
	FL	0.258	0.040	0.470	6.463	0.000	0.226	4.423

Note: Dependent variable: S.

). We speculate that the possible reason is the terrain variation causing undulations in the Yuanjia Village street space, especially on SY Street, where the elevation difference is about 5–6 m. The undulating terrain likely made walking tiring for tourists, resulting in fewer smiles and weaker crowd density, which led to lower actual calculated scores.

2.

Characterization of environmental amenity elements

Among environmental facility factors, street furniture was strongly correlated with tourist perception evaluation. However, landmark spaces and historical nodes (BZXKJ) did not show significant attraction in terms of tourist perception and aggregation. The presence of landscape water features (JGSX) in street furniture demonstrated a strong positive correlation with tourist experience evaluation. According to Yoshinobu Ashihara’s street landscape art theory, urban greenery contributes to a sense of calm and tranquility. Consistent with this theory, “green” and “natural” water landscape features help reduce user stress in an environment, significantly impacting street characteristic perception. The presence of tables and chairs (ZYIBD) in the streets, while strongly correlated with tourist evaluation, had a negative correlation. We speculate the reason: although tables and chairs (ZYIBD) on HM and XC streets can enhance tourists’ food purchases and provide places to rest, these streets are overly crowded. This might lead to lower emotional perceptions among tourists. As a result, the ZYIBD element was negatively correlated with tourists’ perceptions.

3.

Street Floor Characteristics

The characteristics of the ground floor along the street were strongly correlated with tourist perception evaluation, particularly in terms of street permeability and degree of zigzagging. The factors related to street curvature (QZDCLM) and ground-floor façade were significantly correlated with tourist perception evaluation, but they showed a negative correlation. Streets with noticeable corners, protrusions, or recesses had a negative impact on tourists’ perception. The more corners there were, the lower the perception rating. Additionally, transitional space factors were also strongly correlated with tourist perception evaluation. Ground elevation (DMTS), building façade setbacks (LMHT/WY), and spatial separation (KJFG) on the ground influenced tourist perception. The creation of gray space through ground-floor façade setbacks enhanced the perception of transitional spaces. The height differences between the street and building interfaces also contributed to this effect. However, spatial separation (KJFG) showed a negative correlation with tourist perception evaluation. When the transitional space was too clearly defined, it could negatively affect the tourists’ visiting experience.

4.

Street business feature elements

Street business features were found to be the most significant factors among the four aspects that influenced tourists’ perception of street scenes. The commercial in situ value had the greatest positive impact on tourist ratings. However, business diversity had a significant negative correlation with tourist perception scores. This suggests that the more diverse the business types in Yuanjia Village, the lower the tourist perception. This might be related to the purpose of the visit. Yuanjia Village attracts many visitors from outside who seek to experience local food, landscapes, and folk culture. However, some streets in Yuanjia Village are designed to showcase food and culture from across the country, resulting in high business diversity but reduced regional uniqueness, ultimately leading to a negative impact on tourist perception.

4.2.2. Screening of Street Characterization Factors

Through statistical analysis using multiple linear regression, we identified the factors influencing tourists’ behavioral perceptions of street and lane space scenes. The regression coefficients in Table 4 indicate the degree to which the 21 indicators simultaneously affect tourists’ characteristic perceptions. By ranking the influence of the factors (

Table 5. Analysis of impact coefficients of street space scenarios.

Subdivision Indicators	Labeled Name	Impact Factor
Commercial in situ value	FL	0.47
Landscaped water system	JGSX	0.147
Flat ground-floor elevations	PZDCLM	0.128
Density of commercial POI points per 10 m street buffer	FD	0.09
Umbrellas/awnings	ZYS	0.087
Elevation setbacks/extensions	LMHT/WY	0.062
Commercial character processing tools	SYTSJGGJ	0.048
Daily activity space	RCHDKJ	0.043
Ground-floor elevation	DMTS	0.021
Width	KD	--
Aspect ratio	GKB	−0.001
Iconic space	BZXKJ	−0.008
Greenery	LZ	−0.015
Type of business	F	−0.03
Spatial separation	LJFG	−0.048
Relationship of the street to the landform	JX	−0.059
Regional character furnishings	XYTSBS	−0.06
Street business mix (diversity) is the mix of POIs related to vitality within the street buffer.	FH	−0.137
Tables, chairs, and benches	ZYBD	−0.14
Longitudinal crosswalks	ZXJCDL	−0.14
Longitudinal base elevation	ZSDM	−0.162
Curved ground-floor elevations	QZDXLM	−0.175

), we were able to identify the elements with a positive correlation: FL, JGSX, PZDCLM, FD, ZYS, LMHT/WY, SYTSJGGJ, RCHDKJ, DMTS, and KD, for a total of ten elements. The street scale category included two elements, the characterization of environmental amenity elements included four elements, the street floor characteristics contained three elements, and the street business features contained three elements. It is worth mentioning that the factors of street width and aspect ratio encountered multicollinearity issues during model building. Thus, these elements required special handling during the analysis to address the collinearity problem. The influence coefficient of the street mixing degree was negative, indicating that lower diversity leads to superior tourists’ perceptions. Thus, GKB and FH are indicators that positively correlate with tourists’ characteristic perceptions. In summary, these 12 elements form key components of distinctive spatial scenes.

5. Discussion

This study examines the multidimensional behavior and perceptions of tourists under different street spatial environments. The findings confirm that these environmental influences do indeed lead to significant variations. Drone-based analysis of tourist behavior trajectories revealed a notable increase in tourist aggregation on XC and HM Streets. Using the YOLOv5 model to identify tourists’ facial emotions, it was observed that joy perception increased on XC Street, while negative emotions were more prominent on SY3 and SQ3 Streets. Regression analysis of perception and of street and lane elements identified 12 indicator elements that were positively correlated with tourists’ perceptions, while other indicators had a negative impact on tourists’ characteristic perceptions. These results build on the work of experts such as Wang Kai, Zhao Wanmin [64], and Cheng Junjie in extracting microlevel street space elements.

In the multiple regression analysis (Table 4), the street business feature elements, represented by commercial in situ value, business density, and business mixing degree, emerged as key factors influencing tourists’ characteristic perceptions. Additionally, landscape water features, umbrellas, and commercial processing tools within environmental facilities were found to be critical in influencing tourists. This analysis indicates that street spaces’ environmental facilities are strongly appealing to people. The facility elements such as landscape water systems, sunshades, and special processing tools had a significant impact on the perception of distinctiveness. This emphasizes the importance of these elements more than previous studies [36,65].

However, the street business mixing degree in Yuanjia Village shows a different or even opposite conclusion compared with previous studies [55,66]. The underlying reason for this might be related to the unique operational model in Yuanjia Village, where village-run factories process and reprocess agricultural raw materials and sell the products through their own shops. Notably, the shops on XC Street are extensions of various village factories (such as grain and oil mills, tofu factories, and flour mills), making the street and lane space composition quite complex. This complexity results in a diverse range of tourist behaviors, including significant aggregation, extended stay durations, and generally positive emotional responses. These distinctive characteristics differ from other street spaces within Yuanjia Village, where such behaviors are less pronounced. The unique setting of XC Street offers an ideal environment to capture and identify specific behaviors through the multidimensional approach provided in this study. These findings further extend the existing research conclusions by demonstrating how varying degrees of business mixing can create distinctive behavioral outcomes and corresponding emotional responses. This discrepancy from prior studies indicates that context-specific elements and the unique business models in Yuanjia Village play a significant role in shaping tourists’ perceptions and behaviors.

Furthermore, the correlation between tourist perceptions and the presence of landmark public spaces in street and lane areas is relatively weak. This result differs from those of previous studies on public space elements [10,11,12]. The likely reason could be that Yuanjia Village emphasizes the creation of snacks and food entertainment while neglecting activities and cultural exploration in landmark public spaces. This leads to tourists spending more time in everyday activity spaces, thereby causing them to be more influenced by these everyday spaces. Additionally, the distinction between the tortuosity and straightness of street-level façades significantly impacts tourists’ perceptions. Specifically, tourists tend to perceive the characteristic features of straight elements more clearly.

This finding is consistent with Gehl’s research but differs from Zhang’s MiBE variable system [36,67]. The reason for this discrepancy could be that the street space in Yuanjia Village is affected by the topography and elevation changes within the village, leading to variations in street interfaces between tortuous and straight configurations, resulting in differences in tourists’ perceptions. Furthermore, in their study, Xu and Zhang chose Wudaoying Hutong, a long, straight street, as a case study, which differs significantly from the various street shapes in Yuanjia Village, contributing to the divergence in conclusions.

This paper, based on the theoretical background of tourists’ environmental behavior perception and the exploration of streets’ and lanes’ characteristic spatial genes, focuses on analyzing the characteristic elements of street space in Yuanjia Village from a multidimensional perception perspective. This approach aims to refine the characteristic factor extraction method in street space, grounded in spatial gene theory. Additionally, through a combination of environmental behavior methods—specifically, tourist walking aggregation, stay behavior, and emotional perception—this study scientifically captures the unique characteristics of this type of village space. Firstly, by establishing a multidimensional perception analysis from tourist behavior to psychological perception, this study demonstrates that it is possible to create an accurate and efficient tourist characteristic perception map. Secondly, in terms of extracting street feature factors, a pattern of multiple associations between architectural environmental elements and tourists’ aggregation, stay, and emotional perception was identified.

This study could be improved in the following ways in the future:

• Research methodology: The approach presented in this study, which identifies street space characteristics through multidimensional tourist perception, is in its early stages. Changes in tourists’ emotional perceptions under different spatial environment factors can vary, influenced by personal moods, conversations, and other elements. Additionally, age and cultural differences among tourists can affect their perception of spatial environments, making this a complex problem with multiple dimensions. Secondly, the numerical values for the weights in Equation (3) were derived from a statistical analysis based on interviews with some tourists. However, due to the limitations in interview duration and the number of interviews conducted, the assigned weights had a certain degree of subjectivity. Future research should involve experiments with people from diverse backgrounds to increase the scientific validity and diversity of the sample.
• The research case: This study used spatial gene theory to analyze and explore the unique features of Yuanjia Village, providing a reference for villages transitioning to tourism. However, this study may not cover the various spatial environmental factors of different types of villages. Future research should investigate different villages to collect more diverse samples.
• Data collection: The tourist behavior data collection was conducted during holidays, which could have affected perceptions, as local tourists and out-of-province tourists may have different understandings of the cultural characteristics. For example, HM Street in Yuanjia Village resembles Xi’an’s Hui People Street in terms of businesses and architectural style. Local tourists might find it less distinctive, while out-of-province tourists still find it unique. This could lead to biases in our research conclusions regarding feature identification and spatial factor correlation. To address this, future research should ensure that tourist types are adequately differentiated to maintain the relevance of the study conclusions.
• Research feedback: This study lacked empirical evidence to support the extraction of village characteristic factors based on multidimensional tourist behavior perception and the spatial environment elements of streets. Establishing such evidence might have been costly. First, future evaluations of improved street spaces could use dynamic simulations or other methods for secondary assessment. Second, this study’s starting point was based on tourist perception extraction and lacked feedback from original residents in traditional villages. Future research could compare the perception differences between tourists and original residents to provide feedback on characteristic perceptions from different groups, thereby improving the work of extracting streets’ spatial characteristic factors.
• Selection of factors: This study employed a multiple regression model to determine the influence of factors on tourist perception. Due to collinearity issues with KD and KGB, the data analysis for KD was excluded from the analysis and modeling process. Consequently, the impact weight for KD is left blank in Table 5. However, the analysis of individual elements and perception shows a strong positive correlation between KGB and KD. Therefore, future research can improve and optimize the model based on this study.

6. Conclusions

In this study, we constructed an objective evaluation table of street spaces perceived by tourists in multiple dimensions, and we identified and extracted the street environment elements in order to mine the characteristic street space factors of traditional villages and apply them to Yuanjia Village in Shaanxi. The main conclusions are as follows: (1) The multidimensional visitor perception map shows that XC Street had the highest evaluation of visitor perception, and the degree of daily gathering was generally high; the weakest perceptions were for SY Street and SQ Street. (2) Through the regression analysis of the streets’ spatial elements, the street business function was found to be the key factor affecting visitors’ perceptions, where the commercial in situ value had the largest influence coefficient. (3) Among the street space elements extracted from tourist perceptions, 12 factor elements showed a positive correlation with tourists’ characteristic perception, while the rest had a negative correlation. However, due to the significant impact of the commercial in situ value in Yuanjia Village, the influence coefficients of other factors were relatively low. This finding supports the original hypothesis, indicating that the methodology of determining key spatial factors through tourist perception is effective, thereby providing a validated approach for extracting factors for other unique villages. (4) The correlation analysis between tourists’ multidimensional perception and street elements revealed the relationship between tourists’ psychological perceptions and the streets’ spatial environments. Positive emotions and varied walking behaviors showed positive correlations with street business, street furniture, and street-level interface tortuosity elements. They also showed a negative correlation with street elements such as depth, longitudinal cross-streets, and other permeable street features. Based on the results of the above analysis, we found that for the upgrading and reconstruction of streets’ spatial characteristics, it is necessary to consider not only the height and width ratios related to morphology and the ground-floor interface of the street that conforms to the topography, but also the function of the businesses on the street, which has regional characteristics, as well as the street furniture that affects the tourists’ walking tours. Among these factors, the production and processing tools with local characteristics, such as grain and oil workshops, vinegar workshops, chili workshops, and a series of village-run factories, constitute the “Front Store and Back Factory” spatial combination mode, which can be one of the main factors affecting the characteristics. Therefore, for the planning and design of tourist street spaces in characteristic villages and towns, it is necessary to fully respect the production and lifestyle of regional cultural characteristics as well as the old production and processing objects, etc., which can enrich the cultural symbols of street tours. In terms of street layout and form, attention should also be paid to the street interface formed by different topographic elements; Yuanjia Village conforms to the topographic elements formed in line with the Guanzhong regional culture of flat and straight streets, which are more popular among tourists. However, the street patterns and architectural and housing shapes of different regions and cultural backgrounds are also different, so “homogenization” can be avoided only by adapting to local conditions.

The uniqueness of this study is manifested in several aspects. (1) Multidimensional perception method: This study adopted a multidimensional approach to analyzing tourists’ perceptions of traditional village street spaces, including factors such as stay time, aggregation degree, and facial expressions. This method enabled a more comprehensive understanding of tourists’ perception changes in street spaces under different psychological and behavioral states, thus providing a scientific basis for optimizing street spaces. (2) Application of space genetics theory: The study was based on the theory of space genetics, starting from the perspectives of tourists’ behavior and perception, to explore and identify the characteristic factors of traditional village street spaces. Through the analysis of the correlation between street space elements and tourist perception, key factors affecting tourist perception were refined, providing new ideas and methods for village spatial construction and planning. (3) Supplementation and improvement of spatial feature identification framework: This study supplemented and improved the multidimensional population feature identification framework of street space, providing a systematic method for identifying characteristic factors in the spatial genetics of Chinese characteristic towns. This method could more accurately identify and extract the characteristic factors of street space, helping to preserve the unique spatial features of villages and enhance tourist experiences. (4) Empirical research and case analysis: This study combined empirical research and case analysis. Through investigations and analyses of villages such as Yuanjia Village, the feasibility and effectiveness of the method of extracting characteristic factors were verified. This method of combining empirical research enhances the credibility and operability of the research results.