Sensors, Volume 23, Issue 16 (August-2 2023) – 295 articles

Infrared ship target detection is crucial technology in marine scenarios. Ship targets vary in scale throughout navigation because the distance between the ship and the infrared camera is constantly changing. Furthermore, complex backgrounds, such as sea clutter, can cause significant interference during detection tasks. In this paper, multiscale morphological reconstruction-based saliency mapping, combined with a two-branch compensation strategy (MMRSM-TBC) algorithm, is proposed for the detection of ship targets of various sizes and against complex backgrounds. First, a multiscale morphological reconstruction method is proposed to enhance the ship targets in the infrared image and suppress any irrelevant background. Then, by introducing a structure tensor with two feature-based filter templates, we utilize the contour information of the ship targets and further improve their intensities in the saliency map. After that, a two-branch compensation strategy is proposed, due to the uneven distribution of image grayscale. Finally, the target is extracted using an adaptive threshold. The experimental results fully show that our proposed algorithm achieves strong performance in the detection of different-sized ship targets and has a higher accuracy than other existing methods. Full article

This paper proposes a novel design of shielded two-turn near-field probe with focus on high sensitivity and high electric field suppression. A comparison of different two-turn loop topologies and their influence on the probe sensitivity in the frequency range up to 3 GHz is presented. Furthermore, a comparison between a single loop probe and a two-turn probe is given and different topologies of the two-turn probe are analyzed and evaluated. The proposed probes were simulated using Ansys HFSS and manufactured on a standard FR4 substrate four-layer printed circuit board (PCB). A measurement setup for determining probe sensitivity and electric field suppression ratio using an in-house made PCB probe stand, vector network analyzer, microstrip line (MSL) and the manufactured probe is presented. It is shown that using a two-turn probe design it is possible to increase the probe sensitivity while minimizing the influence on the probe spatial resolution. The average sensitivity of the proposed two-turn probe compared to the conventional design is increased by 10.1 dB in the frequency range from 10 MHz up to 1 GHz. Full article

Bone drilling is a common procedure in orthopedic surgery and is frequently attempted using robot-assisted techniques. However, drilling on rigid, slippery, and steep cortical surfaces, which are frequently encountered in robot-assisted operations due to limited workspace, can lead to tool path deviation. Path deviation can have significant impacts on positioning accuracy, hole quality, and surgical safety. In this paper, we consider the deformation of the tool and the robot as the main factors contributing to path deviation. To address this issue, we establish a multi-stage mechanistic model of tool–bone interaction and develop a stiffness model of the robot. Additionally, a joint stiffness identification method is proposed. To compensate for path deviation in robot-assisted bone drilling, a force-position hybrid compensation control framework is proposed based on the derived models and a compensation strategy of path prediction. Our experimental results validate the effectiveness of the proposed compensation control method. Specifically, the path deviation is significantly reduced by 56.6%, the force of the tool is reduced by 38.5%, and the hole quality is substantially improved. The proposed compensation control method based on a multi-stage mechanistic model and joint stiffness identification method can significantly improve the accuracy and safety of robot-assisted bone drilling. Full article

Photographs taken under harsh ambient lighting can suffer from a number of image quality degradation phenomena due to insufficient exposure. These include reduced brightness, loss of transfer information, noise, and color distortion. In order to solve the above problems, researchers have proposed many deep learning-based methods to improve the illumination of images. However, most existing methods face the problem of difficulty in obtaining paired training data. In this context, a zero-reference image enhancement network for low light conditions is proposed in this paper. First, the improved Encoder-Decoder structure is used to extract image features to generate feature maps and generate the parameter matrix of the enhancement factor from the feature maps. Then, the enhancement curve is constructed using the parameter matrix. The image is iteratively enhanced using the enhancement curve and the enhancement parameters. Second, the unsupervised algorithm needs to design an image non-reference loss function in training. Four non-reference loss functions are introduced to train the parameter estimation network. Experiments on several datasets with only low-light images show that the proposed network has improved performance compared with other methods in NIQE, PIQE, and BRISQUE non-reference evaluation index, and ablation experiments are carried out for key parts, which proves the effectiveness of this method. At the same time, the performance data of the method on PC devices and mobile devices are investigated, and the experimental analysis is given. This proves the feasibility of the method in this paper in practical application. Full article

Unmanned vehicles frequently encounter the challenge of navigating through complex mountainous terrains, which are characterized by numerous unknown continuous curves. Drones, with their wide field of view and ability to vertically displace, offer a potential solution to compensate for the limited field of view of ground vehicles. However, the conventional approach of path extraction solely provides pixel-level positional information. Consequently, when drones guide ground unmanned vehicles using visual cues, the road fitting accuracy is compromised, resulting in reduced speed. Addressing these limitations with existing methods has proven to be a formidable task. In this study, we propose an innovative approach for guiding the visual movement of unmanned ground vehicles using an air–ground collaborative vectorized curved road representation and trajectory planning method. Our method offers several advantages over traditional road fitting techniques. Firstly, it incorporates a road star points ordering method based on the K-Means clustering algorithm, which simplifies the complex process of road fitting. Additionally, we introduce a road vectorization model based on the piecewise GA-Bézier algorithm, enabling the identification of the optimal frame from the initial frame to the current frame in the video stream. This significantly improves the road fitting effect ($EV$) and reduces the model running time (T—$model$). Furthermore, we employ smooth trajectory planning along the “route-plane” to maximize speed at turning points, thereby minimizing travel time (T—$travel$). To validate the efficiency and accuracy of our proposed method, we conducted extensive simulation experiments and performed actual comparison experiments. The results demonstrate the superior performance of our approach in terms of both efficiency and accuracy. Full article

Aging of the population and the declining birthrate in Japan have produced severe human resource shortages in the medical and long-term care industries. Reportedly, falls account for more than 50% of all accidents in nursing homes. Recently, various bed-release sensors have become commercially available. In fact, clip sensors, mat sensors, and infrared sensors are used widely in hospitals and nursing care facilities. We propose a simple and inexpensive monitoring system for elderly people as a technology capable of detecting bed activity, aimed particularly at preventing accidents involving falls. Based on findings obtained using that system, we aim at realizing a simple and inexpensive bed-monitoring system that improves quality of life. For this study, we developed a bed-monitoring system for detecting bed activity. It can predict bed release using RFID, which can achieve contactless measurements. The proposed bed-monitoring system incorporates an RFID antenna and tags, with a method for classifying postures based on the RFID communication status. Experimentation confirmed that three postures can be classified with two tags, seven postures with four tags, and nine postures with six tags. The detection rates were 90% for two tags, 75% for four tags, and more than 50% for six tags. Full article

Autonomous robots heavily rely on simultaneous localization and mapping (SLAM) techniques and sensor data to create accurate maps of their surroundings. When multiple robots are employed to expedite exploration, the resulting maps often have varying coordinates and scales. To achieve a comprehensive global view, the utilization of map merging techniques becomes necessary. Previous studies have typically depended on extracting image features from maps to establish connections. However, it is important to note that maps of the same location can exhibit inconsistencies due to sensing errors. Additionally, robot-generated maps are commonly represented in an occupancy grid format, which limits the availability of features for extraction and matching. Therefore, feature extraction and matching play crucial roles in map merging, particularly when dealing with uncertain sensing data. In this study, we introduce a novel method that addresses image noise resulting from sensing errors and applies additional corrections before performing feature extraction. This approach allows for the collection of features from corresponding locations in different maps, facilitating the establishment of connections between different coordinate systems and enabling effective map merging. Evaluation results demonstrate the significant reduction of sensing errors during the image stitching process, thanks to the proposed image pre-processing technique. Full article

Federated learning has attracted much attention in fault diagnosis since it can effectively protect data privacy. However, efficient fault diagnosis performance relies on the uninterrupted training of model parameters with massive amounts of perfect data. To solve the problems of model training difficulty and parameter negative transfer caused by data corruption, a novel cross-device fault diagnosis method based on repaired data is proposed. Specifically, the local model training link in each source client performs random forest regression fitting on the fault samples with missing fragments, and then the repaired data is used for network training. To avoid inpainting fragments to produce the wrong characteristics of faulty samples, joint domain discrepancy loss is introduced to correct the phenomenon of parameter bias during local model training. Considering the randomness of the overall performance change brought about by the local model update, an adaptive update is proposed for each round of global model download and local model update. Finally, the experimental verification was carried out in various industrial scenarios established by three sets of bearing data sets, and the effectiveness of the proposed method in terms of fault diagnosis performance and data privacy protection was verified by comparison with various currently popular federated transfer learning methods. Full article

Fourier ptychographic microscopy (FPM) is a novel technique for computing microimaging that allows imaging of samples such as pathology sections. However, due to the influence of systematic errors and noise, the quality of reconstructed images using FPM is often poor, and the reconstruction efficiency is low. In this paper, a hybrid attention network that combines spatial attention mechanisms with channel attention mechanisms into FPM reconstruction is introduced. Spatial attention can extract fine spatial features and reduce redundant features while, combined with residual channel attention, it adaptively readjusts the hierarchical features to achieve the conversion of low-resolution complex amplitude images to high-resolution ones. The high-resolution images generated by this method can be applied to medical cell recognition, segmentation, classification, and other related studies, providing a better foundation for relevant research. Full article

The accurate identification of highly similar sheet metal parts remains a challenging issue in sheet metal production. To solve this problem, this paper proposes an effective mean square differences (EMSD) algorithm that can effectively distinguish highly similar parts with high accuracy. First, multi-level downsampling and rotation searching are adopted to construct an image pyramid. Then, non-maximum suppression is utilised to determine the optimal rotation for each layer. In the matching, by re-evaluating the contribution of the difference between the corresponding pixels, the matching weight is determined according to the correlation between the grey value information of the matching pixels, and then the effective matching coefficient is determined. Finally, the proposed effective matching coefficient is adopted to obtain the final matching result. The results illustrate that this algorithm exhibits a strong discriminative ability for highly similar parts, with an accuracy of 97.1%, which is 11.5% higher than that of the traditional methods. It has excellent potential for application and can significantly improve sheet metal production efficiency. Full article

In this paper, we propose a combined filtering method rooted in the application of the Sage–Husa Adaptive Kalman filtering, designed specifically to process wave sensor data. This methodology aims to boost the measurement precision and real-time performance of wave parameters. (1) This study delineates the basic principles of the Kalman filter. (2) We discuss in detail the methodology for analyzing wave parameters from the collected wave acceleration data, and deeply study the key issues that may arise during this process. (3) To evaluate the efficacy of the Kalman filter, we have designed a simulation comparison encompassing various filtering algorithms. The results show that the Sage–Husa Adaptive Kalman Composite filter demonstrates superior performance in processing wave sensor data. (4) Additionally, in Chapter 5, we designed a turntable experiment capable of simulating the sinusoidal motion of waves and carried out a detailed errors analysis associated with the Kalman filter, to facilitate a deep understanding of potential problems that may be encountered in practical application, and their solutions. (5) Finally, the results reveal that the Sage–Husa Adaptive Kalman Composite filter improved the accuracy of effective wave height by 48.72% and the precision of effective wave period by 23.33% compared to traditional bandpass filter results. Full article

Motivation: The advancement of preventive medicine and, subsequently, telemedicine drives the need for noninvasive and remote measurements in patients’ natural environments. Heart rate (HR) measurements are particularly promising and extensively researched due to their quick assessment and comprehensive representation of patients’ conditions. However, in scenarios such as endurance training or emergencies, where HR measurement was not anticipated and direct access to victims is limited, no method enables obtaining HR results that are suitable even for triage. Methods: This paper presents the possibility of remotely measuring of human HR from a series of in-flight videos using videoplethysmography (VPG) along with skin detection, human pose estimation and image stabilization methods. An unmanned aerial vehicle (UAV) equipped with a camera captured ten segments of video footage featuring volunteers engaged in free walking and running activities in natural sunlight. The human pose was determined using the OpenPose algorithm, and subsequently, skin areas on the face and forearms were identified and tracked in consecutive frames. Ultimately, HR was estimated using several VPG methods: the green channel (G), green-red difference (GR), excess green (ExG), independent component analysis (ICA), and a plane orthogonal to the skin (POS). Results: When compared to simultaneous readings from a reference ECG-based wearable recorder, the root-mean-squared error ranged from 17.7 (G) to 27.7 (POS), with errors of less than 3.5 bpm achieved for the G and GR methods. Conclusions: These results demonstrate the acceptable accuracy of touchless human pulse measurement with the accompanying UAV-mounted camera. The method bridges the gap between HR-transmitting wearables and emergency HR recorders, and it has the potential to be advantageous in training or rescue scenarios in mountain, water, disaster, or battlefield settings. Full article

Analyzing the photomicrographs of coal and conducting maceral analysis are essential steps in understanding the coal’s characteristics, quality, and potential uses. However, due to limitations of equipment and technology, the obtained coal photomicrographs may have low resolution, failing to show clear details. In this study, we introduce a novel Generative Adversarial Network (GAN) to restore high-definition coal photomicrographs. Compared to traditional image restoration methods, the lightweight GAN-based network generates more explicit and realistic results. In particular, we employ the Wide Residual Block to eliminate the influence of artifacts and improve non-linear fitting ability. Moreover, we adopt a multi-scale attention block embedded in the generator network to capture long-range feature correlations across multiple scales. Experimental results on 468 photomicrographs demonstrate that the proposed method achieves a peak signal-to-noise ratio of 31.12 dB and a structural similarity index of 0.906, significantly higher than state-of-the-art super-resolution reconstruction approaches. Full article

This study presents an enhanced deep learning approach for the accurate detection of eczema and psoriasis skin conditions. Eczema and psoriasis are significant public health concerns that profoundly impact individuals’ quality of life. Early detection and diagnosis play a crucial role in improving treatment outcomes and reducing healthcare costs. Leveraging the potential of deep learning techniques, our proposed model, named “Derma Care,” addresses challenges faced by previous methods, including limited datasets and the need for the simultaneous detection of multiple skin diseases. We extensively evaluated “Derma Care” using a large and diverse dataset of skin images. Our approach achieves remarkable results with an accuracy of 96.20%, precision of 96%, recall of 95.70%, and F1-score of 95.80%. These outcomes outperform existing state-of-the-art methods, underscoring the effectiveness of our novel deep learning approach. Furthermore, our model demonstrates the capability to detect multiple skin diseases simultaneously, enhancing the efficiency and accuracy of dermatological diagnosis. To facilitate practical usage, we present a user-friendly mobile phone application based on our model. The findings of this study hold significant implications for dermatological diagnosis and the early detection of skin diseases, contributing to improved healthcare outcomes for individuals affected by eczema and psoriasis. Full article

Hybrid beamforming is a viable method for lowering the complexity and expense of massive multiple-input multiple-output systems while achieving high data rates on track with digital beamforming. To this end, the purpose of the research reported in this paper is to assess the effectiveness of the three architectural beamforming techniques (Analog, Digital, and Hybrid beamforming) in massive multiple-input multiple-output systems, especially hybrid beamforming. In hybrid beamforming, the antennas are connected to a single radio frequency chain, unlike digital beamforming, where each antenna has a separate radio frequency chain. The beam formation toward a particular angle depends on the channel state information. Further, massive multiple-input multiple-output is discussed in detail along with the performance parameters like bit error rate, signal-to-noise ratio, achievable sum rate, power consumption in massive multiple-input multiple-output, and energy efficiency. Finally, a comparison has been established between the three beamforming techniques. Full article

Soft tactile sensors based on piezoresistive materials have large-area sensing applications. However, their accuracy is often affected by hysteresis which poses a significant challenge during operation. This paper introduces a novel approach that employs a backpropagation (BP) neural network to address the hysteresis nonlinearity in conductive fiber-based tactile sensors. To assess the effectiveness of the proposed method, four sensor units were designed. These sensor units underwent force sequences to collect corresponding output resistance. A backpropagation network was trained using these sequences, thereby correcting the resistance values. The training process exhibited excellent convergence, effectively adjusting the network’s parameters to minimize the error between predicted and actual resistance values. As a result, the trained BP network accurately predicted the output resistances. Several validation experiments were conducted to highlight the primary contribution of this research. The proposed method reduced the maximum hysteresis error from 24.2% of the sensor’s full-scale output to 13.5%. This improvement established the approach as a promising solution for enhancing the accuracy of soft tactile sensors based on piezoresistive materials. By effectively mitigating hysteresis nonlinearity, the capabilities of soft tactile sensors in various applications can be enhanced. These sensors become more reliable and more efficient tools for the measurement and control of force, particularly in the fields of soft robotics and wearable technology. Consequently, their widespread applications extend to robotics, medical devices, consumer electronics, and gaming. Though the complete elimination of hysteresis in tactile sensors may not be feasible, the proposed method effectively modifies the hysteresis nonlinearity, leading to improved sensor output accuracy. Full article

Wi-Fi signals are ubiquitous and provide a convenient, covert, and non-invasive means of recognizing human activity, which is particularly useful for healthcare monitoring. In this study, we investigate a score-level fusion structure for human activity recognition using the Wi-Fi channel state information (CSI) signals. The raw CSI signals undergo an important preprocessing stage before being classified using conventional classifiers at the first level. The output scores of two conventional classifiers are then fused via an analytic network that does not require iterative search for learning. Our experimental results show that the fusion provides good generalization and a shorter learning processing time compared with state-of-the-art networks. Full article

Zirconium sheet has been widely used in various fields, e.g., chemistry and aerospace. The surface scratches on the zirconium sheets caused by complex processing environment have a negative impact on the performance, e.g., working life and fatigue fracture resistance. Therefore, it is necessary to detect the defect of zirconium sheets. However, it is difficult to detect such scratch images due to lots of scattered additive noise and complex interlaced structural texture. Hence, we propose a framework for adaptively detecting scratches on the surface images of zirconium sheets, including noise removing and texture suppressing. First, the noise removal algorithm, i.e., an optimized threshold function based on dual-tree complex wavelet transform, uses selected parameters to remove scattered and numerous noise. Second, the texture suppression algorithm, i.e., an optimized relative total variation enhancement model, employs selected parameters to suppress interlaced texture. Finally, by connecting disconnection based on two types of connection algorithms and replacing the Gaussian filter in the standard Canny edge detection algorithm with our proposed framework, we can more robustly detect the scratches. The experimental results show that the proposed framework is of higher accuracy. Full article

Ultrasound has a deep penetrating ability with minimal or no tissue injury, while cancer-mediated complications during diagnosis, therapy, and surgery have become a serious challenge for clinicians and lead to the severity of the primary condition (cancer). The current study highlights the importance of ultrasound imaging and focused ultrasound therapy during cancer diagnosis, pain reduction, guidance for surgical resection of cancer, and the effectiveness of chemotherapy. We performed the bibliometric analysis on research domains involving ultrasound, cancer management, pain, and other challenges (chemotherapy, surgical guidance, and postoperative care), to observe the trend by which the research field has grown over the years and propose a possible future trend. The data was obtained from the Web of Science, processed, and exported as plain text files for analysis in the Bibliometrix R web interface using the Biblioshiny package. A total of 3248 documents were identified from 1100 journal sources. A total of 390 articles were published in 2022, with almost a 100% growth rate from previous years. Based on the various network analysis, we conclude that the outcome of the constant research in this domain will result in better patient care during the management of various diseases, including cancer and other co-morbidities. Full article

A thyroid nodule, a common abnormal growth within the thyroid gland, is often identified through ultrasound imaging of the neck. These growths may be solid- or fluid-filled, and their treatment is influenced by factors such as size and location. The Thyroid Imaging Reporting and Data System (TI-RADS) is a classification method that categorizes thyroid nodules into risk levels based on features such as size, echogenicity, margin, shape, and calcification. It guides clinicians in deciding whether a biopsy or other further evaluation is needed. Machine learning (ML) can complement TI-RADS classification, thereby improving the detection of malignant tumors. When combined with expert rules (TI-RADS) and explanations, ML models may uncover elements that TI-RADS misses, especially when TI-RADS training data are scarce. In this paper, we present an automated system for classifying thyroid nodules according to TI-RADS and assessing malignancy effectively. We use ResNet-101 and DenseNet-201 models to classify thyroid nodules according to TI-RADS and malignancy. By analyzing the models’ last layer using the Grad-CAM algorithm, we demonstrate that these models can identify risk areas and detect nodule features relevant to the TI-RADS score. By integrating Grad-CAM results with feature probability calculations, we provide a precise heat map, visualizing specific features within the nodule and potentially assisting doctors in their assessments. Our experiments show that the utilization of ResNet-101 and DenseNet-201 models, in conjunction with Grad-CAM visualization analysis, improves TI-RADS classification accuracy by up to 10%. This enhancement, achieved through iterative analysis and re-training, underscores the potential of machine learning in advancing thyroid nodule diagnosis, offering a promising direction for further exploration and clinical application. Full article

For many years, energy monitoring at the most disaggregate level has been mainly sought through the idea of Non-Intrusive Load Monitoring (NILM). Developing a practical application of this concept in the residential sector can be impeded by the technical characteristics of case studies. Accordingly, several databases, mainly from Europe and the US, have been publicly released to enable basic research to address NILM issues raised by their challenging features. Nevertheless, the resultant enhancements are limited to the properties of these datasets. Such a restriction has caused NILM studies to overlook residential scenarios related to geographically-specific regions and existent practices to face unexplored situations. This paper presents applied research on NILM in Quebec residences to reveal its barriers to feasible implementations. It commences with a concise discussion about a successful NILM idea to highlight its essential requirements. Afterward, it provides a comparative statistical analysis to represent the specificity of the case study by exploiting real data. Subsequently, this study proposes a combinatory approach to load identification that utilizes the promise of sub-meter smart technologies and integrates the intrusive aspect of load monitoring with the non-intrusive one to alleviate NILM difficulties in Quebec residences. A load disaggregation technique is suggested to manifest these complications based on supervised and unsupervised machine learning designs. The former is aimed at extracting overall heating demand from the aggregate one while the latter is designed for disaggregating the residual load. The results demonstrate that geographically-dependent cases create electricity consumption scenarios that can deteriorate the performance of existing NILM methods. From a realistic standpoint, this research elaborates on critical remarks to realize viable NILM systems, particularly in Quebec houses. Full article

Bare board AudioMoth recorders offer a low-cost, open-source solution to passive acoustic monitoring (PAM) but need protecting in an enclosure. We were concerned that the choice of enclosure may alter the spectral characteristics of recordings. We focus on polythene bags as the simplest enclosure and assess how their use affects acoustic metrics. Using an anechoic chamber, a series of pure sinusoidal tones from 100 Hz to 20 kHz were recorded on 10 AudioMoth devices and a calibrated Class 1 sound level meter. The recordings were made on bare board AudioMoth devices, as well as after covering them with different bags. Linear phase finite impulse response filters were designed to replicate the frequency response functions between the incident pressure wave and the recorded signals. We applied these filters to ~1000 sound recordings to assess the effects of the AudioMoth and the bags on 19 acoustic metrics. While bare board AudioMoth showed very consistent spectral responses with accentuation in the higher frequencies, bag enclosures led to significant and erratic attenuation inconsistent between frequencies. Few acoustic metrics were insensitive to this uncertainty, rendering index comparisons unreliable. Biases due to enclosures on PAM devices may need to be considered when choosing appropriate acoustic indices for ecological studies. Archived recordings without adequate metadata may potentially produce biased acoustic index values and should be treated cautiously. Full article

In the realm of the Internet of Things (IoT), a network of sensors and actuators collaborates to fulfill specific tasks. As the demand for IoT networks continues to rise, it becomes crucial to ensure the stability of this technology and adapt it for further expansion. Through an analysis of related works, including the feedback-based optimized fuzzy scheduling approach (FOFSA) algorithm, the adaptive task allocation technique (ATAT), and the osmosis load balancing algorithm (OLB), we identify their limitations in achieving optimal energy efficiency and fast decision making. To address these limitations, this research introduces a novel approach to enhance the processing time and energy efficiency of IoT networks. The proposed approach achieves this by efficiently allocating IoT data resources in the Mist layer during the early stages. We apply the approach to our proposed system known as the Mist-based fuzzy healthcare system (MFHS) that demonstrates promising potential to overcome the existing challenges and pave the way for the efficient industrial Internet of healthcare things (IIoHT) of the future. Full article

Vision-based object detection is essential for safe and efficient field operation for autonomous agricultural vehicles. However, one of the challenges in transferring state-of-the-art object detectors to the agricultural domain is the limited availability of labeled datasets. This paper seeks to address this challenge by utilizing two object detection models based on YOLOv5, one pre-trained on a large-scale dataset for detecting general classes of objects and one trained to detect a smaller number of agriculture-specific classes. To combine the detections of the models at inference, we propose an ensemble module based on a hierarchical structure of classes. Results show that applying the proposed ensemble module increases [email protected] from $0.575$ to $0.65$ on the test dataset and reduces the misclassification of similar classes detected by different models. Furthermore, by translating detections from base classes to a higher level in the class hierarchy, we can increase the overall [email protected] to $0.701$ at the cost of reducing class granularity. Full article

In recent years, integrating structured light with deep learning has gained considerable attention in three-dimensional (3D) shape reconstruction due to its high precision and suitability for dynamic applications. While previous techniques primarily focus on processing in the spatial domain, this paper proposes a novel time-distributed approach for temporal structured-light 3D shape reconstruction using deep learning. The proposed approach utilizes an autoencoder network and time-distributed wrapper to convert multiple temporal fringe patterns into their corresponding numerators and denominators of the arctangent functions. Fringe projection profilometry (FPP), a well-known temporal structured-light technique, is employed to prepare high-quality ground truth and depict the 3D reconstruction process. Our experimental findings show that the time-distributed 3D reconstruction technique achieves comparable outcomes with the dual-frequency dataset ($\mathit{p}$ = 0.014) and higher accuracy than the triple-frequency dataset ($\mathit{p}$ = 1.029 × ${10}^{-9}$), according to non-parametric statistical tests. Moreover, the proposed approach’s straightforward implementation of a single training network for multiple converters makes it more practical for scientific research and industrial applications. Full article

In recent years, deep learning-based speech synthesis has attracted a lot of attention from the machine learning and speech communities. In this paper, we propose Mixture-TTS, a non-autoregressive speech synthesis model based on mixture alignment mechanism. Mixture-TTS aims to optimize the alignment information between text sequences and mel-spectrogram. Mixture-TTS uses a linguistic encoder based on soft phoneme-level alignment and hard word-level alignment approaches, which explicitly extract word-level semantic information, and introduce pitch and energy predictors to optimally predict the rhythmic information of the audio. Specifically, Mixture-TTS introduces a post-net based on a five-layer 1D convolution network to optimize the reconfiguration capability of the mel-spectrogram. We connect the output of the decoder to the post-net through the residual network. The mel-spectrogram is converted into the final audio by the HiFi-GAN vocoder. We evaluate the performance of the Mixture-TTS on the AISHELL3 and LJSpeech datasets. Experimental results show that Mixture-TTS is somewhat better in alignment information between the text sequences and mel-spectrogram, and is able to achieve high-quality audio. The ablation studies demonstrate that the structure of Mixture-TTS is effective. Full article

Social media is a real-time social sensor to sense and collect diverse information, which can be combined with sentiment analysis to help IoT sensors provide user-demanded favorable data in smart systems. In the case of insufficient data labels, cross-domain sentiment analysis aims to transfer knowledge from the source domain with rich labels to the target domain that lacks labels. Most domain adaptation sentiment analysis methods achieve transfer learning by reducing the domain differences between the source and target domains, but little attention is paid to the negative transfer problem caused by invalid source domains. To address these problems, this paper proposes a cross-domain sentiment analysis method based on feature projection and multi-source attention (FPMA), which not only alleviates the effect of negative transfer through a multi-source selection strategy but also improves the classification performance in terms of feature representation. Specifically, two feature extractors and a domain discriminator are employed to extract shared and private features through adversarial training. The extracted features are optimized by orthogonal projection to help train the classification in multi-source domains. Finally, each text in the target domain is fed into the trained module. The sentiment tendency is predicted in the weighted form of the attention mechanism based on the classification results from the multi-source domains. The experimental results on two commonly used datasets showed that FPMA outperformed baseline models. Full article

In recent years, neural network algorithms have demonstrated tremendous potential for modulation classification. Deep learning methods typically take raw signals or convert signals into time–frequency images as inputs to convolutional neural networks (CNNs) or recurrent neural networks (RNNs). However, with the advancement of graph neural networks (GNNs), a new approach has been introduced involving transforming time series data into graph structures. In this study, we propose a CNN-transformer graph neural network (CTGNet) for modulation classification, to uncover complex representations in signal data. First, we apply sliding window processing to the original signals, obtaining signal subsequences and reorganizing them into a signal subsequence matrix. Subsequently, we employ CTGNet, which adaptively maps the preprocessed signal matrices into graph structures, and utilize a graph neural network based on GraphSAGE and DMoNPool for classification. Extensive experiments demonstrated that our method outperformed advanced deep learning techniques, achieving the highest recognition accuracy. This underscores CTGNet’s significant advantage in capturing key features in signal data and providing an effective solution for modulation classification tasks. Full article

Voice spoofing attempts to break into a specific automatic speaker verification (ASV) system by forging the user’s voice and can be used through methods such as text-to-speech (TTS), voice conversion (VC), and replay attacks. Recently, deep learning-based voice spoofing countermeasures have been developed. However, the problem with replay is that it is difficult to construct a large number of datasets because it requires a physical recording process. To overcome these problems, this study proposes a pre-training framework based on multi-order acoustic simulation for replay voice spoofing detection. Multi-order acoustic simulation utilizes existing clean signal and room impulse response (RIR) datasets to generate audios, which simulate the various acoustic configurations of the original and replayed audios. The acoustic configuration refers to factors such as the microphone type, reverberation, time delay, and noise that may occur between a speaker and microphone during the recording process. We assume that a deep learning model trained on an audio that simulates the various acoustic configurations of the original and replayed audios can classify the acoustic configurations of the original and replay audios well. To validate this, we performed pre-training to classify the audio generated by the multi-order acoustic simulation into three classes: clean signal, audio simulating the acoustic configuration of the original audio, and audio simulating the acoustic configuration of the replay audio. We also set the weights of the pre-training model to the initial weights of the replay voice spoofing detection model using the existing replay voice spoofing dataset and then performed fine-tuning. To validate the effectiveness of the proposed method, we evaluated the performance of the conventional method without pre-training and proposed method using an objective metric, i.e., the accuracy and F1-score. As a result, the conventional method achieved an accuracy of 92.94%, F1-score of 86.92% and the proposed method achieved an accuracy of 98.16%, F1-score of 95.08%. Full article

We demonstrated a new optical fiber modal interferometer (MI) for airflow sensing; the novelty of the proposed structure is that an MI is fabricated based on a piece of HAF, which makes the sensitive MI itself also a hotwire. The interferometer is made by applying arc-discharge tapering and then flame tapering on a 10 mm length high attenuation fiber (HAF, 2 dB/cm) with both ends spliced to a normal single mode fiber. When the diameter of the fiber in the processing region is reduced to about 2 μm, the near-infrared dispersion turning point (DTP) can be observed in the interferometer’s transmission spectrum. Due to the absorption of the HAF, the interferometer will have a large temperature increase under the action of a pump laser. At the same time, the spectrum of the interferometer with a DTP is very sensitive to the change in ambient temperature. Since airflow will significantly affect the temperature around the fiber, this thermosensitive interferometer with an integrated heat source is suitable for airflow sensing. Such an airflow sensor sample with a 31.2 mm length was made and pumped by a 980 nm laser with power up to 200 mW. In the comparative experiment with an electrical anemometer, this sensor exhibits a very high air-flow sensitivity of −2.69 nm/(m/s) at a flowrate of about 1.0 m/s. The sensitivity can be further improved by enlarging the waist length, increasing the pump power, etc. The optical anemometer with an extremely high sensitivity and a compact size has the potential to measure a low flowrate in constrained microfluidic channels. Full article