Impact of Illuminated Road Signs on Driver’s Perception

Abstract

This study determined citizens’ perceptions and impact factors of illuminated road signs installed to ensure their visibility at night when the risk of traffic accidents is high. An ordered logit model was used to measure illuminated road signs’ impact on drivers’ perception based on the data from the citizens’ survey conducted by the Road Sign Center. According to the results, the internal (optical fiber) type variable had the highest impact, followed by the frequent fog variable and the complex road line variable. This study found that most citizens positively recognized road signs, preferred internal (optical fiber) types, and desired illuminated road signs that considered climate, environment, and road structure types. In Seoul, the importance and improvement of illuminated road signs at points where road structures are complex, such as city streets, were high. Additionally, the illuminated road sign recognition and road type variable were significant in Gyeonggi-do, which reflected the high number of citizens that commute to Seoul from Gyeonggi-do. Concerning local cities and counties and intercity roads highly affected by the climate, the impact was high at points with frequent fog. Fog affects the visibility distance, generates condensation on signs, and significantly degrades visibility. Therefore, an illuminated road sign installation method must be presented based on spatial analysis for regions vulnerable to climate, environment, and road location. Additionally, the road intersection point variable was significant in local cities and counties, which reflects the relatively lagged road infrastructure. Local cities and counties are financially poor and have numerous aged drivers; hence, central government support that considers these aspects is crucial.

1. Introduction

1.1. Research Background and Purpose

Undoubtedly, the fatality or injury risk in a traffic accident is greater at night than during the day. The statistical data from South Korea demonstrate that the frequency of occurrence of traffic accidents from 2017–2019 was greater during the day (6 a.m.–6 p.m.; 130,351 cases) than at night (90,677 cases). Moreover, the death toll per 100 traffic accidents was 1.47 people during the day, while it was 2.05 people at night, approximately 1.4 times higher. To partially prevent such traffic accidents at night and provide night road guidance to drivers, the Ministry of Land, Infrastructure, and Transport (MLIT) established a steering regulation to install lighting on road signs using Road Sign Rules in March 2016. Previously, reflection sheets were attached on letters in road signs to safeguard nighttime visibility; however, considering the development of lighting technology and the autonomous risk response of road management agencies, the permissive range of illuminated road signs has been regulated. Additionally, in 2018, the National Police Agency (NPA) published guidelines stipulating the production, performance, and test standards of illuminated and luminescent traffic safety signs, targeting the locations of frequent nighttime traffic accidents. Subsequently, illuminated road signs were installed on a full scale, which improved nighttime driving safety and convenience and drew positive responses from public institutions and citizens. However, this positive response to the illuminated road signs was a subjective evaluation by the media, such as press articles and policy press releases, and did not include citizens’ feedback. Furthermore, citizens’ satisfaction has not been quantitatively examined until now. As the requirements for illuminated road signs targeting citizens have not been examined, it was difficult to determine the detailed installation requirements, such as the type of illuminated road signs desired by citizens, required roads and locations, and factors affecting drivers’ perception. Additionally, while several studies have analyzed road sign imagery, policy, and traffic signs, research targeting illuminated road signs is scarce. Thus, the Road Sign Center, an institution designated by the MLIT for operating and managing road signs, surveyed citizens on illuminated road signs in June 2022. The Road Sign Center specializes in road signs, and it systemizes illuminated road signs and national road sign information, reviews existing/new road signs, amends regulations, and provides other technological, political, and research support. The present study analyzed the impact factors of illuminated road signs on citizens’ and drivers’ perception that have not been previously examined using the data collected from the citizens’ surveys on illuminated road signs performed by the Road Sign Center. Therefore, the degree of recognition of illuminated road signs, preferences by type, and points requiring sign installation were identified. When analyzing impact factors, implications were derived by classifying them by city size and road type, and future plans for the installation and operation of illuminated road signs were formulated. Section 1 presents the literature review as part of the Introduction. Section 2 presents the methodology for analyzing the impact factors. Section 3 presents an empirical analysis of the drivers’ perception of impact factors. Section 4 and Section 5 present the discussion and implications of the study.

1.2. Research Scope

The target of this study was limited to road signs installed to guide drivers to easily reach their destination and the illuminated road signs equipped with lighting for nighttime visibility. Under the jurisdiction of the MLIT, the sign system in South Korea includes road signs, also known as guide signs, that aim to provide guidance for the destination, direction, and street name [1]. Traffic signs that indicate the speed limit are regulated by the NPA. This study focuses on road signs (Figure 1).

Illuminated road signs are classified as the internal (optical fiber) type, internal (backlit) type, external (spotlight) type, and external (flood light) type (Figure 2). Concerning the characteristics of each type, the internal (optical fiber) type is a method of transferring light from the internal LED to the optical fiber while arranging the end of the optical fiber according to the design of the road sign. As patterns such as letters and arrows emit their own light, they have the highest visibility, and electricity can be generated through solar cells; therefore, electricity inflow is unnecessary. However, the text of the draft design cannot be changed, and its installation cost is relatively higher than other methods. The internal and external (spotlight) types can commonly be installed only with an electric facility. These types differ because the internal (backlit) type has consistent brightness due to the lighting being placed inside the sign; however, the installation costs are high, and repair is difficult. The external (spotlight) type involves lighting installation on the upper part of the road sign; consequently, the installation costs are low, and installation and repair are relatively easy, but the sign’s brightness is not even. Therefore, visibility is compromised. The external (flood light) type is similar to the external (spotlight) type; however, the former involves the use of lights outside the sign, such as on the ground, unlike the external (spotlight) type, which includes lighting in the upper part of the road sign. These are sometimes installed on highways but are not used often due to the issue of degraded visibility.

The urban structure and population density in South Korea differ depending on the urban scale of each region. Accordingly, road sign information, such as road number, destination name, and so on, should differ. Particularly, considering the circumstances at night, methods of illuminated road sign installation that reflect various transportation infrastructures according to the urban scale, including streetlights, traffic lights, and road complexity, must be derived. Therefore, the present study considered the structural characteristics of South Korean urban areas and classified the spatial range using Gyeonggi-do, Seoul (capital), metropolitan cities, and local cities and counties (Figure 3). Seoul is a world-class metropolitan city with a population of 10 million. It has a complex urban structure with high-quality road infrastructure. A metropolitan city is an urban region with a population of more than 1 million. Except for some focus cities, the local counties and districts comprise small cities and rural areas with less than 100,000 people, and the road infrastructure is relatively underdeveloped. Gyeonggi-do has many new cities developed to disperse the expanding population and industry of Seoul, and many residents commute from the new city to Seoul. Furthermore, this study classified the downtown area where survey participants frequently drove and other sections to additionally set the range of analysis according to the road type.

1.3. Literature Review

Illuminated road signs are mainly installed to prevent traffic accidents at night. Nighttime traffic accidents are typically affected by night visibility combined with poor visual guidance, lack of lighting facilities, and complex road structures [2,3,4]. Additionally, previous studies have found that traffic accidents are directly related to weather conditions, such as rainfall, severe cold, fog, and heat waves [5,6]. In a previous study on illuminated road signs, Kim et al. [7] explored the methods of improving the visibility of external (spotlight) type road signs by considering the location of the lighting device installation, suitable quadrant, and reflecting board. For the external (spotlight) type that includes lighting on the upper part of the road sign and the external (flood light) type that includes lighting at the bottom, the location of the lighting installation, suitable quadrant, and reflecting board use were adjusted to present the results of the on-site experiment and economic efficiency analyses. The illuminated road sign developed in the corresponding research reduced the costs required for fabrication, construction, and maintenance by 39.8% compared to the existing ones [7]. However, few previous studies have investigated illuminated road signs for the purpose of night road guidance. Furthermore, these studies targeted the external types and did not consider optical fiber or backlit types. Madlenák et al. [8] analyzed the visibility of road signs during night driving. In the experimental road section, 97% of the 174 billboards were lightless, and the driver was visually drawn to the illuminated billboards with reflective elements. Shin et al. [9] conducted a daytime and nighttime legibility distance measurement study on lightweight variable message signs (VMSs). In the case of single-sided LED VMSs, similar to the subject of this study, it was found that the legibility distance increased by 1.2 times during the day compared to night.

Considering the relevant literature, several studies have examined VMSs, including studies on VMS display preferences and safety effects in construction sites and on optimal VMS display methods through brightness analysis and preference surveys [10,11,12,13,14,15,16,17,18]. Along with VMSs, studies have also been conducted on the validity of vehicle-activated signs (VASs) [19,20,21]. VMSs and VASs can deliver variable information at night with high visibility, but they mainly deliver traffic circumstance and traffic safety (e.g., speed limit and lane information) information. Additionally, VMS-based directional road signs are operated sometimes; however, their installation cost is high. This is functionally different from this study, which only aims to guide destinations based on existing road sign standards.

When considering the trends in the literature relevant to signs, image analysis research cannot be discounted. Representative studies on advanced driver assistance systems (ADASs) of vehicles and autonomous driving have considered future transportation environments and continuity [22,23,24,25,26,27]. However, these studies have targeted traffic signs rather than road signs. In Korea, road sign image analyses and image-based facility safety assessments have been performed with only road signs as subjects. [28,29,30,31]. Notably, these road sign image analysis studies were mainly conducted during the day. However, their reliability at night is also vital for supporting ADASs and autonomous driving. Therefore, efforts are needed to solve the safety problem of driving vehicles at night through additional studies on illuminated road signs.

This section discusses the existing literature on road sign recognition. In South Korea, traffic signs and road signs are distinguished by management and, hence, research that only targets road signs as subjects is conducted by distinguishing them from traffic signs. Studies on road sign design [32,33,34] and road sign confusion and misreading rate analysis [35,36] are some examples. On the contrary, studies conducted overseas have mainly targeted traffic signs. Additionally, experimental research on the recognition of signs, degree of understanding, and visual movement has been performed from the driver’s perspective [37,38,39,40,41]. These studies have academically contributed by presenting directions for sign improvement through sign recognition experiments from the driver’s perspective; however, most studies were conducted during the day and did not consider nighttime circumstances. Research that accounted for nighttime visibility has mostly emphasized the quality of the retro-reflection material of signs [42,43,44,45]. Compared to existing signs, retro-reflection material signs improve visibility but they do not ensure readability and strong visibility as illuminated road signs do. Therefore, it is crucial to conduct research on nighttime recognition of general road signs and illuminated road signs. Additionally, suitable apportionment of retro-reflection material road signs and illuminated road signs is necessary by considering the installation costs and road circumstances.

2. Methodology

The Road Sign Center conducted an online survey of male and female adults aged 18 years or above with driver’s licenses between 15 to 22 June 2022 to investigate their level of satisfaction and perception of illuminated road signs. The survey was conducted at the request of the MLIT to identify driver satisfaction with illuminated road signs and explore ways to expand and install them in the future. The survey data were collected through a professional research company. The samples were extracted based on the allocation proportional to age and sex, considering the current statistics of holders of the National Police Agency’s driver’s license. The number of valid samples excluding insincere responses was 2000. The survey comprised the respondent’s demographic characteristics, driver’s perception characteristics, illuminated road sign perception, and preferences. In case an illuminated road sign type was unfamiliar to the respondents, the characteristics and images of each type were presented, as illustrated in Figure 2, to increase their understanding before proceeding with the survey. Ordered logit regression (OLR) was used for analysis. The analysis proceeded for each classification of urban scale and road type by considering the characteristics of the respondents to derive implications (

Table 1. Analysis outline.

Category	Content
Period (Sample)	15–22 June 2022 (n = 2000)
Sample extraction method	Extraction of allocation proportional to age and sex based on the current state of statistics of the holders of the National Police Agency’s driver’s license
Title	Survey on the perception of illuminated road signs
Survey content	Demographic characteristics, driver’s perception characteristics, illuminated road sign perception, and preference of type
Analysis method	Ordered logit regression
Analysis target	Classified analysis for each urban scale/road type

).

This study analyzes the impact factors of illuminated road signs on drivers’ perception based on the data from the survey performed by the Road Sign Center. Among the questionnaire items, the impact of improving driver perception of illuminated road signs, demographic characteristics, driver’s perception characteristics, recognition of illuminated road signs, and preference questions by type were analyzed. The OLR was used as the analysis method to utilize the five-point Likert-scale survey criteria for the illuminated road signs in improving drivers’ perception among the survey respondents as the dependent variable. OLR is an extended model from the binary logit model. This method is advantageous because it can include ordinal-scale responses as dependent variables and various other responses, such as nominal, ordinal, and ratio scales, as independent variables. Using OLR, the impact factors of illuminated road signs on the driver’s perception were analyzed among the total respondents. The implications were derived through comparative analysis for each urban scale and road type.

The OLR is similar in character to the multinomial logit model. However, unlike the multinomial logit model that includes nominal variables as dependent variables, it is differentiated by calculating the selection probability for the ordered variables [46]. In this survey, the dependent variable is the improvement effect of illuminated road signs on drivers’ perception, and the response criteria contain a five-point scale. The scale standards have the following order in the survey: “very effective”, “effective”, “intermediate”, “not effective”, and “not at all effective” and regulate a certain revealed preference relationship [47]. In this study, the range of the five orders for the extent of road sign utilization was established, and the number of results that the dependent variable could have is g (g ≥ 4). In Equation (1), P(Y ≤ j) refers to the cumulative probability of driving safety as the dependent variable, and it is modeled through a logit transformation, as seen in Equation (2). Here, ${\mu }_j$ refers to the scalar value that determines the range of the latent variable, and ${\displaystyle \sum }_{k=1}^K{\beta }_k{x}_k$ refers to the row vector that represents the independent variables in this study, including demographic characteristics, operational characteristics, road sign satisfaction, and road sign improvement necessities. Consequently, when measured in the order from 1 to g, the used ordered logit model is understood as the odds ratio (OR) for the cumulative probability. Through this, the dependent variable Y, which is within the range of j, is derived [48]:

$$\mathrm{P}\left(Y\le j\right)={\mathsf{\pi }}_1+\cdots +{\mathsf{\pi }}_j,j=1,\cdots ,j$$

(1)

$$\log \left[\frac{\mathrm{P}\left(y\le j\mathrm{lx}\right)}{1-\mathrm{P}\left(y\le j\mathrm{lx}\right)}\right]={\mu }_j-{{\displaystyle \sum }}_{k=1}^K{\beta }_k{x}_k (\mathrm{but}, j = 1, 2, \dots , J - 1)$$

(2)

OR refers to the ratio of the value of odds when the independent variable X increases by one unit to the value of reference odds, where $P\left(y\le j|X\right)$ and $P\left(y>j|X\right)$ represent the probability of the dependent variable Y being equal to or less than category m and the probability of Y being greater than category m, given the values of independent variable X, respectively [49]. The OR of the model for the independent variables is illustrated in Equations (3) and (4):

$$\mathrm{logit} \mathrm{P}\left(Y\le j\right)={\mathsf{\alpha }}_1+\cdots {\beta }_x, j=1,\cdots ,j-1$$

(3)

$$\mathrm{Odds} \mathrm{ratio} \left(\mathrm{OR}\right)=\left[\frac{P\left(Y\le jlX=x_2\right)/P(Y>jlX=x_2)}{P\left(Y\le jlX=x_1\right)/P(Y>jlX=x_{1)}}\right]$$

(4)

Considering the literature on road signs utilizing an OLR analysis, one study has examined the impact factors of road signs for driving safety through surveys [1]. Additionally, another study has derived future road sign utilization methods through OLR for diversification of future road sign function [50]. These studies were significant in that the future directionality for road signs was set while its previous role was diminished due to devices that guided paths, such as navigation devices and smartphones. However, a detailed plan for independent improvements for road structural infrastructure was lacking. The present study is differentiated by presenting improvement methods for road sign infrastructure for improving nighttime drivers’ perception. Other than the abovementioned studies, research on road signs using OLR analysis is non-existent. However, an impact analysis of traffic accidents and driver conformance behavior based on traffic signs and VMSs has been performed [51,52]. Notably, previous studies have examined the impact factors of road traffic accidents [53,54,55,56] and the impact of traffic accidents considering climate factors [6,57,58]. Other studies have analyzed the impact factors of a new commute railroad system, subway, and electric bicycles through OLR [59,60,61]. As such, in transportation sectors, OLR is utilized when the impact factor is analyzed based on the dependent variables composed of ordinal scales such as satisfaction, severity, and importance. However, research that targeted road signs installed to improve driving safety and provide road information for drivers is non-existent, particularly research on illuminated road signs to support nighttime drivers’ perception. In the present study, an OLR analysis was performed based on 2000 samples collected from citizen surveys in 2022 to derive meaningful results.

3. Empirical Analysis

3.1. Variable Composition and Respondent Characteristics

The dependent variables in this study were rated on a five-point scale (5: very effective, 4: effective, 3: intermediate, 2: not effective, and 1: not at all effective) to measure the effect of illuminated road signs on improving drivers’ perception. The expected improvement effect increases the visibility of night road signs to improve road guidance convenience. Therefore, driving safety can be improved to help prevent night traffic accidents. While 26.6% (531 respondents) of responses on illuminated road signs were positive for very effective and 55.8% (1116 respondents) for effective, 1.0% responded that they were not effective (including not at all effective). As most citizens perceive illuminated road signs positively, active installation and use of illuminated road signs will be necessary. The dependent variables, largely dummy variables, were demographic characteristics and the driver’s perception characteristics comprising illuminated road sign perception criteria as continuous variables. Among demographic characteristics, sex was differentiated by male respondents as 0 (reference) and female respondents as 1, and the driver’s license holder allocation proportional to sex was 58.0% (1160) for male respondents and 42.0% (840) for female respondents. Concerning age, the reference variable was set to 41–60, and those 40 or below and 61 or above were set as dummy variables. The driver’s license holder allocation proportional to age was set to 35.6% for age 40 or below, 46.4% for 41–60, and 18.0% for 61 or above. Concerning the driving experience among operational characteristics, 6–20 years (41.3%) was set as the reference variable, and five years or below (26.2%) and 21 years or above (32.5%) were set as dummy variables. In South Korea, roads are divided into express national roads, general national roads, provincial roads, city roads, and so on. However, considering that the target audience was the general public, the types of roads mainly driven on during the survey were simplified to city streets and suburban roads to improve the understanding of the questions. In the analysis, city streets (73.4%) were set as a reference variable and streets outside the city (26.6%) as a dummy variable. The illuminated road sign perception criterion set as a continuous variable was largely classified as illuminated road sign perception, preference for each illuminated road sign type, and necessary points for illuminated road sign installation. The illuminated road sign perception criterion comprised a five-point-scale response for the sufficiency level of illuminated road sign installations. The preference for each illuminated road sign type also comprised a five-point-scale response, and the preference for internal (optical fiber) type, internal (backlit) type, external (spotlight) type, and external (flood light) type was studied. The average value for each response demonstrated that the internal (optical fiber) type had the highest preference at 4.06 points, followed by 3.40 points for the internal (backlit) type, 3.23 points for the external (spotlight) type, and 2.98 points for the external (flood light) type. Concerning the points where illuminated road sign installation was necessary, a seven-point-scale response was prepared to review the relative difference between the criteria in detail. The response criterion with the highest average value was “points with frequent fog (6.02 points)”, followed by “points with complex road lines (6.00 points)” and “intersections of major roads (5.85 points)”. Citizens evaluated the weather conditions and complex road environment as necessary points for illuminated road signs in night situations with poor visibility.

Table 2. Basic statistics of each variable.

Classification		Variable		Frequency	Ratio	Min	Max	Mean	STD
Dependent variable		Driver’s perception improvement effect	1	1	0.1%	1	5	4.08	0.682
			2	17	0.9%
			3	335	16.8%
			4	1116	55.8%
			5	531	26.6%
Independent variable	Demographic characteristics	Sex (dummy variable)	Male (ref.)	1160	58.0%	-	-	-	-
			Female	840	42.0%	-	-	-	-
		Age (dummy variable)	Under 40 years	712	35.6%	-	-	-	-
			41–60 (ref.) years	928	46.4%	-	-	-	-
			Over 61 years	360	18.0%	-	-	-	-
	Driver or user characteristics	Driving experience (dummy variable)	Under 5 years	524	26.2%	-	-	-	-
			6–20 years (ref.)	826	41.3%	-	-	-	-
			Over 21 years	650	32.5%	-	-	-	-
		Frequently used road type (dummy variable)	City street	1468	73.4%	-	-	-	-
			Streets outside the city	532	26.6%	-	-	-	-
	Perception of the illuminated road sign	Recognition of the illuminated road sign	Level of the sufficiency of illuminated road sign installation	-	-	1	5	2.96	1.005
		Preference for each illuminated road sign	Internal (optical fiber) type	-	-	1	5	4.06	0.831
			Internal (backlit) type	-	-	1	5	3.40	1.045
			External (spotlight) type	-	-	1	5	3.23	1.045
			External (flood light) type	-	-	1	5	2.98	1.069
		Necessary points for illuminated road sign installation	Points with frequent fog	-	-	1	7	6.02	0.063
			Points with significant traffic	-	-	1	7	5.57	1.135
			Points with complex road lines	-	-	1	7	6.00	1.022
			Intersection of major roads	-	-	1	7	5.85	0.994
			Vehicle-exclusive roads	-	-	1	7	5.48	1.074
			Points of continuous flow without traffic signals	-	-	1	7	5.71	1.099
			Points with small road curvature radius	-	-	1	7	5.58	1.143
			Points available for electric work for lighting installation	-	-	1	7	5.50	1.128

illustrates the basic statistics of the respondents and variable composition.

3.2. Total Respondent Analysis Results

The impact factors of illuminated road signs for drivers’ perception were analyzed using 2000 samples. Multicollinearity was assessed to examine the correlation between independent variables, wherein the variable inflation factor (VIF) value was 1.061–1.962, and the correlation between variables was not significant. The chi-square value was 575.300 (p < 0.0001), and the model was suitable. Nagelkerke R Z was calculated to be 0.287; thus, the explanation power of the model was good. When the Nagelkerke R Z value is between 0.2 and 0.3 in logistic regression analysis, it is considered a moderate value [62].

For ease of interpretation, the analysis proceeded for the OR as the probability rate for the reference variable of each variable with an exponentiated β value. The demographic characteristics demonstrated that the respondents aged 61 years or above considered that the improvement effect of illuminated road signs on their perception was 1.469 times (p < 0.01) higher than those aged from 41–60 years (reference). This result was attributed to the deteriorated vision associated with biological aging and entering senior years. The variable of sex and variable of age 40 years and below were not significant. The driving characteristics demonstrated that the impact of the improvement effect increased by 1.259 times (p < 0.05) when the frequently used street type was streets other than city streets. This result was attributed to the characteristics of streets outside the city that relatively lack nighttime lighting infrastructure. The driving experience variable was not significant.

The illuminated road sign perception criteria were continuous variables due to the ordinal scale, and the impact size could be compared through the OR of each variable. Concerning the survey results for the sufficiency level of illuminated road sign installation, the positive effect was found to be reduced by 15.5% (p < 0.01) according to the response that illuminated road sign installation was sufficiently increased. The preference for each illuminated road sign type demonstrated that the impact of the internal (optical fiber) type variable was greatest among the significant variables at 1.643 times (p < 0.001). This was attributed to the lighting being applied to the design and having excellent visibility. Additionally, the internal (backlit) type variable was found to have a higher impact by 1.237 times (p < 0.001). This was considered to be the result of the internal (backlit) type road sign being assessed as being preferred over the external (spotlight) type, and external (flood light) types were found to be insignificant. Among the preference criteria for points necessary for illuminated road sign installation, “points with frequent fog” (1.428 times, p < 0.001) was analyzed to have the greatest impact. This was attributed to the inherent value of highly valuing the importance of illuminated road signs under climate conditions that require slow driving by the citizens. The variables “intersection of major roads” (1.255 times, p < 0.001) and “points with complex road lines” (1.214 times, p < 0.01) were also significant. This was attributed to the burden and fatigue of drivers headed toward the destination during nighttime on discontinuous roads. The variable “points available for electric work for lighting installation” (1.118 times, p < 0.05) was significant at a relatively low impact. This was attributed to the existing road signs as fixed infrastructure that do not use electricity and the installation of electricity that is not considered during illuminated road sign installation. The variables “points with significant traffic”, “vehicle-exclusive roads”, “points of continuous flow without traffic signals”, and “points with small road curvature radius” were not significant (

Table 3. Analysis results from all respondents.

Classification		Variable		β	Odds Ratio	Std Error	p-Value
Dependent variable		Driver’s perception improvement effect	Intercept 1	0.452	-	1.075	0.674
			Intercept 2	3.400 ***	-	0.525	0.000
			Intercept 3	7.013 ***	-	0.502	0.000
			Intercept 4	10.223 ***	-	0.535	0.000
Independent variable	Demographic characteristics	Sex (reference: male)	Female	0.102	1.107	0.097	0.297
		Age (reference: 41–60 years)	Under 40 years	−0.049	0.953	0.123	0.692
			Over 61 years	0.385 **	1.469	0.130	0.003
	Driver or user characteristics	Driving experience (reference: 6–20 years)	Under 5 years	0.188	1.207	0.120	0.118
			Over 21 years	−0.150	0.861	0.120	0.213
		Frequently used road type (reference: city streets)	Streets outside the city	0.230 *	1.259	0.105	0.029
	Perception of the illuminated road sign	Recognition of the illuminated road sign	Level of the sufficiency of illuminated road sign installation	−0.169 **	0.845	0.052	0.001
		Preference for each illuminated road sign	Internal (optical fiber) type	0.496 ***	1.643	0.063	0.000
			Internal (backlit) type	0.213 ***	1.237	0.051	0.000
			External (spotlight) type	0.022	1.022	0.055	0.688
			External (flood light) type	0.049	1.050	0.054	0.366
		Necessary points for illuminated road sign installation	Points with frequent fog (sign condensation occurrence)	0.356 ***	1.428	0.057	0.000
			Points with significant traffic	−0.019	0.981	0.050	0.705
			Points with complex road lines	0.194 **	1.214	0.059	0.001
			Intersection of major roads	0.228 ***	1.255	0.061	0.000
			Vehicle-exclusive roads	0.104	1.110	0.054	0.053
			Points of continuous flow without traffic signals	0.052	1.053	0.056	0.357
			Points with small road curvature radius	−0.017	0.983	0.055	0.755
			Points available for electric work for lighting installation	0.111 *	1.118	0.054	0.040

* p < 0.05, ** p < 0.01, *** p < 0.001. Model suitability: chi-Square = 575.300, p = 0.000. Model explanation power: Nagelkerke R Z = 0.287

).

3.3. Analysis Results for Urban Scale

As established in this study’s scope, the analysis was performed by each urban scale by classifying it into Seoul, metropolitan cities, Gyeonggi-do, and local cities and counties after considering the characteristics of the urban structure. The classified urban structure characteristics demonstrate that the population density in Seoul is high as it is one of the world’s largest cities, with nearly 10 million people. Thus, it has the highest number of operated vehicles nationwide. Additionally, the city has complex roads and structure due to the coexistence of the old urban area with irregular roads and the new urban area with grid-shaped roads. Nonetheless, the urban area is concentrated; hence, the nighttime lighting infrastructure is densely structured, and the optimal nationwide transportation system infrastructure is established based on Seoul’s transportation data system and the control center of 21 autonomous districts. In metropolitan cities, a high level of transportation infrastructure is established compared to other local areas. They have a population of one million or above but a smaller population than Seoul. In local cities and counties, the level of transportation infrastructure is relatively low due to the financial circumstances of the local government. They are characterized by a low population density and a predominantly older adult population. Gyeonggi-do is characterized by its large urban-scale belt where the population and cities expand to near Seoul, and the population is about 1.4 times greater than that of Seoul at about 13.6 million. To disperse the concentration phenomenon in Seoul, first and second new city policies were constructed, and numerous citizens commute to Seoul from Gyeonggi-do. The urban structure and road environment are also affected significantly by Seoul because of its adjacency, but Gyeonggi-do and the outskirts include characteristics of the countryside [1]. As such, the characteristics of each urban scale were considered to comparatively analyze the impact factors that the illuminated road signs have for drivers’ perception (

Table 4. Analysis results for each urban scale.

Classification		Variable		Odds Ratio
				Seoul (n = 672)	Metropolitan Cities (n = 328)	Gyeonggi-do (n = 487)	Local Counties and Districts (n = 513)
Dependent variable		Driver’s perception improvement effect		-	-	-	-
Independent variable	Demographic characteristics	Sex (reference: male)	Female	1.410 *	0.937	1.103	0.968
		Age (reference: 41–60 years)	Under 40 years	0.574 *	1.209	1.049	1.365
			Over 61 years	1.657 *	1.124	1.720 *	1.291
	Driver or user characteristics	Driving experience (reference: 6–20 years)	Under 5 years	1.086	1.464	1.494	1.110
			Over 21 years	0.570 **	1.846 *	0.775	0.987
		Frequently used road type (reference: city streets)	Streets outside the city	1.310	0.797	1.854 **	1.154
	Perception of the illuminated road sign	Recognition of the illuminated road sign	Level of the sufficiency of illuminated road sign installation	0.946	0.803	0.726 **	0.859
		Preference for each illuminated road sign	Internal (optical fiber) type	1.702 ***	1.793 ***	1.700 ***	1.540 *
			Internal (backlit) type	1.221 *	1.208	1.414 **	1.193
			External (spotlight) type	1.059	1.044	0.888	1.088
			External (flood light) type	0.978	1.108	1.092	1.049
		Necessary points for illuminated road sign installation	Points with frequent fog (sign condensation occurrence)	1.435 ***	1.821 ***	1.122	1.578 ***
			Points with significant traffic	1.071	0.844	0.929	0.998
			Points with complex road lines	1.197	1.303	1.206	1.137
			Intersection of major roads	1.521 ***	1.024	1.235	1.304 *
			Vehicle-exclusive roads	1.059	1.329 *	1.014	1.194
			Points of continuous flow without traffic signals	1.070	0.907	1.078	1.140
			Points with small road curvature radius	0.918	1.287	0.921	0.922
			Points available for electric work for lighting installation	1.180	0.863	1.378 **	1.060

* p < 0.05, ** p < 0.01, *** p < 0.001.

).

The demographic characteristics demonstrated that the variables of sex and age of 40 years or below were significant in Seoul only. In Seoul, the impact of the improvement effect of illuminated road signs on drivers’ perception was analyzed to be 1.410 times (p < 0.05) higher among women than men (reference). This could be understood as the positive perception of illuminated road signs being relatively higher among women than men from Seoul. The impact decreased by 42.6% for those aged 40 years or below in Seoul compared to those aged from 41–60 years (reference), but the impact increased by 1.657 times for those aged 61 years or above. In Gyeonggi-do, the impact also increased by 1.720 times for those aged 61 years or above. Within the Seoul Metropolitan area, the improvement effect from illuminated road signs was considered to be perceived more positively with increasing age. Metropolitan cities and local cities and counties were found to be not significant. The frequently used road types among the operational characteristics demonstrated that the impact for the respondents that drove on streets outside the city was 1.854 times higher than for those using city streets in Gyeonggi-do. This was understood as the reflection of the high number of citizens who commute to Seoul from Gyeonggi-do using the main road. The main road that connects Seoul and Gyeonggi-do tends to demonstrate a significant number of speeding vehicles at nighttime. Hence, a nighttime speed reduction policy, as well as an illuminated road sign installation method that considers this, will be necessary.

The criteria for the perception of illuminated road signs were as follows. First, in the case of the “level of sufficiency of illuminated road sign installation” variable, this was the only significant criterion with a negative sign in Gyeonggi-do. The impact decreased by 27.4% (p < 0.01) as the level of illuminated road signs was perceived to be sufficient. This could be due to the high number of citizens that commuted to Seoul. The preference for each illuminated road sign demonstrates that the internal (optical fiber) type was significant in all urban types. In particular, the OR in the metropolitan cities was most impactful at 1.793 (p < 0.001). In the case of the internal (backlit) type, Seoul (1.221, p < 0.05) and Gyeonggi-do (1.414, p < 0.01) were significant, and the impact was higher for the internal (optical fiber) type than the internal (backlit) type in both areas. External (spotlight) type and external (flood light) type were not significant in any urban types. Considering the criteria for points necessary for illuminated road sign installation, the OR of the variable of “intersections of major roads” was 1.521 (p < 0.001) and had the highest impact, followed by the variable of “points with frequent fog” at 1.435 (p < 0.001). These results reflect the complex city and road structure of Seoul. Other criteria were not significant. In the metropolitan cities and local cities and counties, the “points with frequent fog” variable (1.821 for the metropolitan cities and 1.578 for local cities and counties, both p < 0.001) had the highest impact. This reflects the necessity of illuminated road signs under severe weather conditions. In the metropolitan cities, the “vehicle-exclusive roads” variable was significant (1.329, p < 0.05), which could reflect the relative abundance of vehicle-exclusive roads compared to local cities and counties and the demand for installation of illuminated road signs. In local cities and counties, the “intersection of major roads” variable (1.304, p < 0.05) was significant.

Along with installing illuminated road signs at intersections of roads in underdeveloped local cities and counties, it is necessary to maintain road guidance systems. In Gyeonggi-do, only the “points available for electric work for lightning installation” variable was significant (1.378, p < 0.01). In the case of the internal (optical fiber) type, electric installation is unnecessary when using solar batteries; thus, considering sign installation based on the internal (optical fiber) type would be a suitable method.

3.4. Analysis Results for Urban Scale Type

The roads mainly used by the survey respondents were classified as city streets and streets outside the city for comparative analysis (

Table 5. Analysis results for each road type.

Classification		Variable		Odds Ratio
				City Street Driving (n = 1468)	Suburban Driving (n = 532)
Dependent variable		Driver’s perception improvement effect		-	-
Independent variable	Demographic characteristics	Sex (reference: male)	Female	1.169	0.949
		Age (reference: 41–60 years)	Under 40 years	0.920	1.049
			Over 61 years	1.345	2.040 **
	Driver or user characteristics	Driving experience (reference: 6–20 years)	Under 5 years	1.185	1.321
			Over 21 years	0.900	0.762
	Perception of illuminated road sign	Recognition of illuminated road sign	Level of the sufficiency of illuminated road sign installation	0.854 **	0.777 *
		Preference for each illuminated road sign	Internal (optical fiber) type	1.594 ***	1.814 ***
			Internal (backlit) type	1.267 ***	1.197
			External (spotlight) type	0.984	1.164
			External (flood light) type	1.067	0.972
		Necessary points for illuminated road sign installation	Points with frequent fog (sign condensation occurrence)	1.312 ***	1.799 ***
			Points with significant traffic	1.035	0.819 *
			Points with complex road lines	1.206 **	1.179
			Intersection of major roads	1.278 ***	1.223
			Vehicle-exclusive roads	1.147 *	1.043
			Points of continuous flow without traffic signals	1.054	1.060
			Points with small road curvature radius	0.965	1.068
			Points available for electric work for lighting installation	1.083	1.249 *

* p < 0.05, ** p < 0.01, *** p < 0.001.

). From demographic and drivers’ perception characteristics, age 61 years or above was the only significant variable for city street driving. Its OR was the greatest at 2.040 (p < 0.01) among the variables. This was attributed to the combined results of the inadequate streetlights on streets other than city streets and the deteriorated vision of aged people. The sex and driving experience variables were not significant.

The “level of sufficiency of illuminated road sign installation” variable was significant in both classifications with negative signs. The positive effects of driving on streets outside the city decreased by 22.3% (p < 0.05) and driving on city streets by 14.6% (p < 0.01) with the higher perception of the sufficiency of the illuminated road sign installation. The preference for each road sign type demonstrated that the internal (optical fiber) type (1.594, p < 0.001) and the internal (backlit) type (1.267, p < 0.001) variables were significant with positive signs when the road lighting infrastructure was better. However, on streets outside the city, only the internal (optical fiber) type demonstrated a high impact at an OR of 1.814 (p < 0.001). This was considered as the preference for internal (optical fiber) type road sign installations with the greatest visibility on streets outside cities where lighting facilities are relatively lacking. Concerning the points necessary for illuminated road sign installation, “points with frequent fog” were most impactful at an OR of 1.312 (p < 0.001) in the case of driving on city streets, followed by “intersection of major roads” (1.278, p < 0.001), “points with complex road lines” (1.206, p < 0.01), and “vehicle-exclusive roads” (1.147, p < 0.05). These were attributed to the climate factors and complex road structures in city streets, which intricately influenced the results. On the contrary, “points with frequent fog” had a strong impact compared to other variables, with an OR of 1.799 (p < 0.001) for driving on streets outside cities. This reflects the characteristics of streets outside cities where climate factors are relatively significant.

4. Discussion

This study found that most citizens perceived illuminated road signs positively, preferred the internal (optical fiber) type, and desired the placement of illuminated road signs that considered the climatic environment and road structures. Although illuminated road signs do not directly prevent traffic accidents, they could lead to improvements such as keeping drivers’ eyes forward for perception, ensuring immediacy during exit–entry, and enhanced visibility during severe weather conditions. The key here is “where to install what type” of illuminated road signs. In this study, the overall preference for the internal (optical fiber) type of lighting was high. Additionally, the demand was high for points significantly affected by climatic factors such as fog or points with complex road structures. Particularly, this necessity for road intersections was high in local cities and counties, where road infrastructure is lacking, and in Seoul, where the urban and road structures are complex. In the case of streets outside cities, the environment for streetlights was weaker than that in city streets. Therefore, the highly intuitive optical fiber type was preferred, and the influence of climatic factors was significant. Additionally, the optical fiber type is effective in terms of sustainability because it does not require electrical installation using solar cells. However, as it is impossible to modify the text, it should be used at a fixed road guidance point. Regionally, it is necessary to improve driving safety and prevent night traffic accidents by strengthening the installation of illuminated road signs on small local roads without streetlights. As such, this study presented the construction direction for illuminated road signs in a broad framework of the classified urban scales and road types. However, a more detailed review of the current state is necessary for making decisions about illuminated road sign installation in reality. The variables not covered in this study, such as road types across the nation or in each local government’s jurisdiction, exits, number of foggy days, and the current state of nighttime traffic accidents, must be considered. In South Korea, the current state of nationwide road signs can be examined through the road sign management system operated by the Road Sign Center [63]. Additionally, meteorological data, such as rainfall/snowfall and the number of foggy days, can be identified through the Portal for Open Meteorological Data by the Ministry of Environment [64]. Furthermore, the current state of traffic accidents can be examined for each period through the traffic accident analysis system based on a geographical information system (GIS) [65]. Thus, it is possible to establish the optimal location for illuminated road sign installation by combining these data.

For example, the traffic accident analysis system statistics will identify the status of accident occurrence for each year and nighttime period related to illuminated road signs. After the amendment of the Road Sign Rules in 2016, illuminated road signs began to be installed on a full scale. Thus, the statistical data from 2016 to 2020 were studied. First, the progress in nighttime collision accidents, directly and indirectly, relevant to drivers’ illuminated road sign perception was reviewed. The sum of the number of nighttime accident occurrences, including crashes related to abrupt stopping, broadside collisions related to sudden lane change, and head-on collisions related to reverse driving, was 56,051 in 2016, 50,995 in 2017, 48,187 in 2018, 48,022 in 2019, and 44,317 in 2020. Thus, the sum continuously decreased at an annual average of 5.7%. The death toll had also decreased by an annual average of 9.9% from 570 in 2016 to 375 in 2020. According to a study by Kim et al. [3], lighting facilities affected the severity of night accidents and low perception of driving conditions were found to be the cause of the accidents [3]. Although installing illuminated road signs does not have an absolute effect on reducing nighttime collisions, it is considered to have a partial impact. Second, the nighttime traffic accidents for each type of intersection were reviewed. The number of traffic accidents on road types other than intersections was 51.693 in 2016, 47,278 in 2017, 45,657 in 2018, 44,945 in 2019, and 40,724 in 2020, and it decreased by an annual average of 5.8%. However, the reduction rate was 0.6% in three-leg intersections, 0.4% in four-leg intersections, and 1.2% in five-leg intersections; the rate of reduction was significantly low compared to road types other than intersections. Particularly, the occurrence rate of traffic accidents in modern roundabouts increased by 11.0%. In South Korea, the installation ratio of roundabouts is not high, and they may be difficult to navigate when driving for the first time. During the day, it is easy to visually recognize roundabouts, reduce speed, and smoothly support traffic flow; however, the rate of accidents at night seems to have increased due to the unfamiliar road structure. According to a study by Kim and Oh [4], places with frequent traffic accidents at night were found to be IC and tunnel sections [4]. This is similar to the road intersection points derived as points of need for illuminated road signs in this study. Illuminated road signs must be installed to reduce the occurrence rate of traffic accidents at intersection points or locations with complex road lines. Third, the current state of traffic accident occurrence for each climate state was assessed. The number of traffic accident occurrences under severe weather conditions, including rain, fog, and snow, was 17,973 in 2016, 12,549 in 2017, 16,490 in 2018, 15,025 in 2019, and 16,528 in 2020. It decreased at an annual average of 2.1%, but the number of cases increased by 1503 in 2020 and, hence, the decreasing trend was not continuous. During severe weather conditions, the visibility of road signs decreases, and issues such as condensation on road signs could occur. Therefore, an installation method for illuminated road signs based on spatial analysis for regions with severe climatic environments and road locations is necessary.

This study examined illuminated road signs. The circumstances for keeping illuminated road signs within a limited budget are realistically difficult. Therefore, installing road signs, including retro-reflection material as reviewed from previous literature, is necessary, and illuminated road signs must be selectively installed by considering the urban scale, road type, road structure, climatic conditions, and current accident occurrences. The standard and location of the posts on the road sign, the driver’s reading distance, and the amount of information recognized should also be considered [30,66,67]. According to the characteristics of the road and traffic, road signs combined with VMSs (or VASs) could also be considered. To suitably allocate various types of road signs, including illuminated road signs, an economic efficiency analysis and the previously emphasized detailed survey of the current state must be performed. The costs for each type must be determined, but a comprehensive review is necessary, considering the simulation of on-site allocation by testing various types, comparing costs invested for each type, and measuring benefits according to the installation of illuminated road signs. Additionally, the maintenance costs must be reviewed. Due to the recent improvement in solar battery efficiency, the internal (optical fiber) type illuminated road sign does not require electricity installation. The internal (optical fiber) type illuminated road sign has higher installation costs than other types, but its advantage in maintenance costs, high visibility, and citizen preference must be considered during selection. Considering the low financial budget of local governments, the support of the central government is crucial. Particularly, small and medium-sized cities outside of the city with weaker financial circumstances have low road traffic infrastructure and numerous old drivers. Therefore, practical measures such as SOC budget support relevant to roads and road sign maintenance project expansion are necessary.

5. Conclusions

This study analyzed the impact factors of illuminated road signs on drivers’ perception through OLR analysis of 2000 samples from a citizens’ survey. Considering the characteristics of South Korea, the analysis was performed for each classification at the urban scale, including Seoul, metropolitan cities, Gyeonggi-do, local cities and counties, city streets, and streets outside of cities with different nighttime road lighting to derive the implications. First, the positive effect for the internal (optical fiber) type variable was greatest among the illuminated road sign types. Particularly, the classified analysis for streets outside of cities demonstrated a 1.814-times higher impact, and this implies that the internal (optical fiber) type illuminated road sign with the greatest visibility must be mainly installed on streets outside of cities with relatively deficient lighting facilities. Second, the importance and improvement impact of illuminated road signs were highest at points where the road structure is complex in Seoul, particularly on city streets. Additionally, the illuminated road sign recognition and road type variables were significant in Gyeonggi-do, which was attributed to the high number of citizens commuting from Gyeonggi-do to Seoul. The significant age variable of the Seoul Metropolitan area, including Seoul and Gyeonggi-do, was also a major characteristic. An illuminated road sign installation method must be presented based on an understanding of the Seoul Metropolitan area’s urban and traffic structure. Third, the impact was high at points with frequent fog in local cities and counties and streets outside cities that are impacted by the climate to a greater extent than the Seoul Metropolitan area. Fog affects the visibility range, induces condensation on signboards, and significantly reduces visibility. In local cities and counties, the road intersection point variable was significant, which reflects the relatively underdeveloped road infrastructure. Local cities and counties are financially inferior and include numerous old drivers. Thus, support from the central government considering the aforementioned aspects is necessary.

Although the role of road signs is being reduced due to navigation devices and smartphones, road signs are essential public goods considering the older adults who have difficulty using devices and communication network failures. This study is significant in that it was the first to analyze the impact of illuminated road signs on drivers’ perception to improve the nighttime safety and convenience of vehicle drivers. This study is anticipated to be utilized as basic data for installing illuminated road signs by local governments and road management institutions. However, there were limitations. As it was an early study on illuminated road signs, the current state of the field, including the specific state of roads, the current state of the climate, and the current state of accidents, could not be analyzed. Furthermore, a complex analysis involving other types of signs, such as existing road signs and VMSs other than illuminated road signs, could not be considered in the analysis. Experiments related to the visibility and readability of illuminated road signs, including other types of road signs, must be carried out for certain locations of on-site placement. Methodologically, when considering the classification analysis by city size, considering South Korean city types, the sample size is rather small, and there is an error in the sample ratio. Repeated illuminated road sign layout simulation, economic analysis, robustness, and maintenance are also necessary. In the future, it is anticipated that illuminated road signs will be optimally placed after subsequent active research, which will allow the provision of safe road guidance services at night.