Square Pixels Research Articles

μCT ANALYSIS AND 3D MODELLING OF COMPARATIVE FLUID PERMEABILITY OF HUMAN TALI FOLLOWING THREE DIFFERENT CLINICAL PROCEDURES: NANOFRACTURE, MICROFRACTURE, AND FINE WIRE DRILLING

AbstractObjectivesThe aim of this work was to compare the different techniques and the different fluid permeability of the tissue following each technique through assessing the flow of radiopaque contrast agent using μCT image analysis and 3D modelling.MethodsDonated human tali specimens (n=12) were prepared through creating a 10mm diameter chondral defect in three different regions of each talus. Each region then underwent one of three surgical techniques: 1) Fine wire drilling, 2) Nanofracture or 3) Microfracture, equidistant sites in each defect to ensure even distribution. Each region then had an addition of 0.1 ml radiopaque contrast agent (Omnipaque™ 300), imaged using a clinical μCT scanner (SCANCO Medical AG, 73.6 μm resolution). Each μCT scan was segmented using Slicer 3D software (The Slicer Community, 2023 3D Slicer (5.2.2)). The segmentation package was used to segment the bone and contrast agent regions in each different surgical site of each sample. Each defect site was created into a cylinder and the ratio of segmented pixels of contrast agent against bone.ResultsThe μCT analysis indicated that across the 12 samples, eight nanofracture regions demonstrated flow of the contrast agent either to the depth of the fracture site or deeper. Some lateral flow was also observed in these sites. eight microfracture regions demonstrated that the flow of the contrast agent was localised to the fracture site and a preferential flow laterally. In only one sample, did a fine wire drilling region demonstrate any fluid flow. In this sample, contrast agent had permeated through the drilling site to the bottom and some sub-site permeation was observed. However, in all samples that showed no permeation of contrast agent through the fracture site, a layer of contrast agent on the chondral surface or minor permeation through to the sub-chondral surface. Segmentation of each sample site showed a significant increase (n=12, p<0.05) in fluid flow of the contrast agent in the nanofracture sites (11%) compared to microfracture (5%) and fine wire drilling (2%).ConclusionsNanofracture showed significantly improved fluid permeability throughout the surrounding trabecular structure, when compared to microfracture and fine wire drilling. Microfracture appears to allow some fluid flow, but only confined to the immediate area around the fracture site, while fine wire drilling appears to allow a comparably small amount, if not no fluid flow through the surrounding trabecular tissue. This conclusion is reinforced by previous literature that concluded the damage to the structure of the trabecular tissue is reduced when using nanofracture, compared to the other two techniques.Declaration of Interest(a) fully declare any financial or other potential conflict of interest

Shot classification for human behavioural analysis in video surveillance applications

Human behavior analysis plays a vital role in ensuring security and safety of people in crowded public places against diverse contexts like theft detection, violence prevention, explosion anticipation etc. Analysing human behaviour by classifying of videos in to different shot types helps in extracting appropriate behavioural cues. Shots indicates the subject size within the frame and the basic camera shots include: the close-up, medium shot, and the long shot. If the video is categorised as Close-up shot type, investigating emotional displays helps in identifying criminal suspects by analysing the signs of aggressiveness and nervousness to prevent illegal acts. Mid shot can be used for analysing nonverbal communication like clothing, facial expressions, gestures and personal space. For long shot type, behavioural analysis is by extracting the cues from gait and atomic action displayed by the person. Here, the framework for shot scale analysis for video surveillance applications is by using Face pixel percentage and deep learning based method. Face Pixel ratio corresponds to the percentage of region occupied by the face region in a frame. The Face pixel Ratio is thresholded with predefined threshold values and grouped into Close-up shot, mid shot and long shot categories. Shot scale analysis based on transfer learning utilizes effective pre-trained models that includes AlexNet, VGG Net, GoogLeNet and ResNet. From experimentation, it is observed that, among the pre-trained models used for experimentation GoogLeNet tops with the accuracy of 94.61%.

Computer Aided Classification of X-ray Images from Pediatric Pneumonia Subjects Collected in Developing Countries

Pneumonia is a lower tract respiratory infection due to bacteria or viruses. It is a severe disease in the pediatric population. Pneumonia is the leading cause of mortality in children under five years worldwide. One of the problems with pneumonia is the diagnosis, as the symptoms of pneumonia may overlap with other diseases, such as asthma and bronchiolitis. In this work, we propose to develop a method for classifying pneumonia and non-pneumonia using X-ray images. We collected 60 X-ray images from Dr. Sardjito Hospital, Yogyakarta, Indonesia, and the dataset from Kaggle. We processed these images through pre-processing algorithms to enhance the image quality, segmentation, white pixel computation, and classification. The novelty of our method is using the ratio of the white pixels from edge detection using the Canny algorithm with the white pixels from segmentation for classifying pneumonia/non-pneumonia. In the Kaggle dataset, our proposed method achieved an accuracy of 86.7%, a sensitivity of 100%, and a specificity of 85%. The classification using the dataset from Dr. Sardjito Hospital yields sensitivity, specificity, and accuracy of 80%, 60%, and 66.7%, respectively. Despite the low performance in the results, we proved our novel feature, ratio of white pixels, can be used to classify pneumonia/non-pneumonia. We also identified that the local dataset is essential in the algorithm development as it has a different quality from the dataset from modern countries. Further, our simple method can be developed further to support pneumonia diagnosis in resource-limited settings where the advanced computing devices or cloud connection are not available.

Towards robust registration of heterogeneous multispectral UAV imagery: A two-stage approach for cotton leaf lesion grading

Multiple source images acquired from diverse sensors mounted on unmanned aerial vehicles (UAVs) offer valuable complementary information for ground vegetation analysis. However, accurately aligning heterogeneous UAV images poses challenges due to differences in geometry, intensity, and noise resulting from varying imaging principles. This paper presents a two-stage registration method aimed at fusing visible RGB and multispectral images for cotton leaf lesion grading. The coarse alignment stage utilizes Scale Invariant Feature Transform (SIFT), while the refined alignment stage employs a novel correlation coefficient-based template matching. The proposed method first employs the EfficientDet network to detect infected cotton leaves with lesions in RGB images. Subsequently, lesion leaves in multiple spectral imagery (red, green, red edge, and near-infrared bands) are located using the perspective transformation matrix derived from SIFT and the coordinates of lesion leaves in RGB images. Refined registration between RGB and multispectral imagery is achieved through template matching with the new correlation coefficient. The registered reflectance data from the different spectral bands and RGB components are utilized to classify pixels in each infected leaf into lesion, healthy, and soil parts. The lesion grade is determined based on the ratio of lesion pixels to the total corresponding leaf area. Experimental results, compared with manual assessment, demonstrate a lesion leaves detection model with a mAP@0.5 of 91.01% and a leaf lesion grading accuracy of 92.01%. These results validate the suitability of the proposed method for UAV RGB and multispectral image registration, enabling automated cotton leaf lesion grading.

DCMF-AFNet: An anchor-free photovoltaic hot-spot fault detection network based on deformable context transformer and bi-branch multi-level feature fusion

In the process of photovoltaic hot-spot detection by thermal infrared sensors, the fault features cannot be effectively represented due to low pixel ratios and complex environmental interference, which makes it difficult for the detection network to accurately detect hot-spot faults. Therefore, an anchor-free photovoltaic hot-spot fault detection algorithm based on deformable context Transformer and bi-branch multi-level feature fusion is proposed. First, to improve the feature extraction ability of the backbone network for small-scale hot-spot faults, a deformable context Transformer module is constructed. By building an offset network in a multi-headed self-attention mechanism, dynamic and static context information with small-scale fault features can be explored from shallow feature maps. Second, to solve the problem of low target saliency due to the complex background interference, a bi-branch multi-level feature fusion module is designed to aggregate global and local multi-level features in a parallel fusion manner, enabling the detection network to rapidly focus on the fault target region in complex environments. Then, an anchor-free mechanism is introduced, and a dynamic task alignment prediction module is proposed to avoid feature misalignment in the classification and localization tasks and further improve the algorithm detection accuracy. Finally, to verify the superiority of the proposed network, seven detection algorithms are selected for comparison experiments. The experimental results show that the DCMF-AFNet network can accurately detect multi-scale hot-spot targets under various harsh conditions, and the detection accuracy can reach 87.3%.

A novel image compression method using wavelet coefficients and Huffman coding

Compressing medical images to reduce their size while maintaining their clinical and diagnostic information is crucial. Because medical images can be large and demand a lot of storage and transmission capacity, effective compression methods aid medical institutions in better storing and transmitting medical images, reducing costs, speeding up data transfer, and simplifying managing image databases. However, it is essential to note that image compression in medical imaging can also introduce drawbacks, such as loss of information and poor output image quality. Therefore, a suitable compression algorithm and parameter must be chosen to balance file size and visual fidelity. This paper suggests an effective image compression method employing the Discrete Wavelet Transform (DWT), followed by a reduction operation and Huffman coding to produce a mere lossless encoding to transmit the images over a channel. The extracted DWT coefficients are mapped to the nearest integral value. All four sub-bands of DWT are joined, and then a window of 3 × 3 is selected for reduction operation by choosing the origin as the pivot element. The Huffman coding algorithm is used to compress the processed image. The pivot origin element is used in the reversible reduction while uncompressing the image. When sending compressed data across an unreliable route, the window size and pivot element selection keep the compressed data secure. Standard measures such as bits per pixel (BPP) and compression ratio (CR) are used to assess the suggested approach. The efficiency of the suggested course of action is supported by the research's findings, which use a peak signal-to-noise ratio (PSNR) of 54.66 dB.

Road Traffic Sign Detection Method Based on RTS R-CNN Instance Segmentation Network.

With the rapid development of the autonomous driving industry, there is increasing research on related perception tasks. However, research on road surface traffic sign detection tasks is still limited. There are two main challenges to this task. First, when the target object's pixel ratio is small, the detection accuracy often decreases. Second, the existing publicly available road surface traffic sign datasets have limited image data. To address these issues, this paper proposes a new instance segmentation network, RTS R-CNN, for road surface traffic sign detection tasks based on Mask R-CNN. The network can accurately perceive road surface traffic signs and provide important information for the autonomous driving decision-making system. Specifically, CSPDarkNet53_ECA is proposed in the feature extraction stage to enhance the performance of deep convolutional networks by increasing inter-channel interactions. Second, to improve the network's detection accuracy for small target objects, GR-PAFPN is proposed in the feature fusion part, which uses a residual feature enhancement module (RFA) and atrous spatial pyramid pooling (ASPP) to optimize PAFPN and introduces a balanced feature pyramid module (BFP) to handle the imbalanced feature information at different resolutions. Finally, data augmentation is used to generate more data and prevent overfitting in specific scenarios. The proposed method has been tested on the open-source dataset Ceymo, achieving a Macro F1-score of 87.56%, which is 2.3% higher than the baseline method, while the inference speed reaches 23.5 FPS.

Left Ventricle Segmentation in Echocardiography with Transformer.

Left ventricular ejection fraction (LVEF) plays as an essential role in the assessment of cardiac function, providing quantitative data support for the medical diagnosis of heart disease. Robust evaluation of the ejection fraction relies on accurate left ventricular (LV) segmentation of echocardiograms. Because human bias and expensive labor cost exist in manual echocardiographic analysis, computer algorithms of deep-learning have been developed to help human experts in segmentation tasks. Most of the previous work is based on the convolutional neural networks (CNN) structure and has achieved good results. However, the region occupied by the left ventricle is large for echocardiography. Therefore, the limited receptive field of CNN leaves much room for improvement in the effectiveness of LV segmentation. In recent years, Vision Transformer models have demonstrated their effectiveness and universality in traditional semantic segmentation tasks. Inspired by this, we propose two models that use two different pure Transformers as the basic framework for LV segmentation in echocardiography: one combines Swin Transformer and K-Net, and the other uses Segformer. We evaluate these two models on the EchoNet-Dynamic dataset of LV segmentation and compare the quantitative metrics with other models for LV segmentation. The experimental results show that the mean Dice similarity of the two models scores are 92.92% and 92.79%, respectively, which outperform most of the previous mainstream CNN models. In addition, we found that for some samples that were not easily segmented, whereas both our models successfully recognized the valve region and separated left ventricle and left atrium, the CNN model segmented them together as a single part. Therefore, it becomes possible for us to obtain accurate segmentation results through simple post-processing, by filtering out the parts with the largest circumference or pixel square. These promising results prove the effectiveness of the two models and reveal the potential of Transformer structure in echocardiographic segmentation.

Evaluation of the effect of leaf spatial aggregation on chlorophyll content retrieval in open-canopy apple orchards

The real-world crowns of broadleaf tree species feature green leaves surrounding branches, resulting in leaf spatial aggregation effect in the crown. However, the impact of such leaf spatial aggregation on chlorophyll content retrieval has not yet been determined. This study investigated the effect of leaf spatial aggregation on chlorophyll content retrieval in two distinct apple orchards with open canopies. The “PROSPECT + LESS” model was used for canopy reflectance simulation, and 25 hyperspectral vegetation indices (VIs) were analyzed to identify universal VIs for various leaf aggregations. Sensitivity analysis was conducted to evaluate the impact of leaf aggregation on the relationships between VIs and chlorophyll content. An artificial neural network regression algorithm was used to retrieve chlorophyll content by reversing the radiative transfer model (RTMs). The results show that leaf aggregation significantly affects the relationships between VIs and chlorophyll content as a result of the variability in the ratio of photosynthetic vegetation pixels to background pixels captured by the sensor at the top of the canopy. TCARI/OSAVI was found to be resistant to confounding factors (e.g., leaf area index and dry matter content) and maintained stable relationships with chlorophyll content. Leaf spatial aggregation had a significant impact on chlorophyll content retrieval, especially when leaves were highly aggregated. In such cases, the spectral variation driven by the photosynthetic vegetation was masked by the background, leading to a large divergence between simulated and observed spectra. Low to moderate levels of leaf aggregation, on the other hand, provided accurate chlorophyll content retrieval in both apple orchards (R2 = 0.49 to 0.67). In conclusion, when using 3D RTMs to retrieve chlorophyll content, it is recommended to configure low to moderate levels of leaf aggregation to ensure high accuracy and efficiency.

Potential Disease Severity Assessment Approach for Potato Late Blight with Smartphone-Based Image Capturing

Accurate and timely evaluation of the severity of Potato Late Blight (PLB) is vital in disease control and being confident that fungicides are utilized efficiently. PLB is caused by Phytophthora infestans and is one of the most destructive diseases affecting potatoes. Knowledge gap in determining the severity of the PLB using the Late Blight Key (LBK) results in erroneous disease evaluation, unregulated chemical use, and exorbitant management expenses for PLB. The purpose of this research was to build a novel technique based on smartphones to measure the PLB severity in relation to the LBK. Using a smartphone camera, digital images of sick PLB leaves were taken at a consistent height, and simultaneously the related visual disease severity rating score was recorded using the LBK. The MATLAB (R2019b) image processing toolbox was used to process the photos, which included cropping, manually defining the areas of the leaf and lesion, and determining the pixel ratio (the ratio of PLB lesion to leaf pixels). When data from all the images were merged at the conclusion, regression analysis revealed a virtually perfect linear association (R2=86.6%) between the image-derived pixel ratio and the visual ranking scores of the PLB sickness severity. According to these results, image processing analysis that has been built may precisely identify the leaf areas that are infected with PLB and assess the disease's severity level.

Starvation combined with constant anode potential triggers intracellular electron storage in electro-active biofilms

The accumulation of electrons in the form of Extracellular Polymeric Substances (EPS) and poly-hydroxyalkanoates (PHA) has been studied in anaerobic processes by adjusting the access of microorganisms to the electron donor and final electron acceptor. In Bio-electrochemical systems (BESs), intermittent anode potential regimes have also recently been used to study electron storage in anodic electro-active biofilms (EABfs), but the effect of electron donor feeding mode on electron storage has not been explored. Therefore, in this study, the accumulation of electrons in the form of EPS and PHA was studied as a function of the operating conditions. EABfs were grown under both constant and intermittent anode potential regimes and fed with acetate (electron donor) continuously or in batch. Confocal Laser Scanning Microscopy (CLSM) and Fourier-Transform Infrared Spectroscopy (FTIR) were used to assess electron storage. The range of Coulombic efficiencies, from 25 to 82%, and the biomass yields, between 10 and 20%, indicate that storage could have been an alternative electron consuming process. From image processing, a 0.92 pixel ratio of poly-hydroxybutyrate (PHB) and amount of cells was found in the batch fed EABf grown under a constant anode potential. This storage was linked to the presence of living Geobacter and shows that energy gain and carbon source starvation were the triggers for intracellular electron storage. The highest EPS content (extracellular storage) was observed in the continuously fed EABf under an intermittent anode potential, showing that constant access to electron donor and intermittent access to the electron acceptor leads to the formation of EPS from the excess energy gained. Tailoring operating conditions can thus steer the microbial community and result in a trained EABf to perform a desired biological conversion, which can be beneficial for a more efficient and optimized BES.

A Review of Critical Issues in High-Speed Vat Photopolymerization.

Vat photopolymerization (VPP) is an effective additive manufacturing (AM) process known for its high dimensional accuracy and excellent surface finish. It employs vector scanning and mask projection techniques to cure photopolymer resin at a specific wavelength. Among the mask projection methods, digital light processing (DLP) and liquid crystal display (LCD) VPP have gained significant popularity in various industries. To upgrade DLP and LCC VPP into a high-speed process, increasing both the printing speed and projection area in terms of the volumetric print rate is crucial. However, challenges arise, such as the high separation force between the cured part and the interface and a longer resin refilling time. Additionally, the divergence of the light-emitting diode (LED) makes controlling the irradiance homogeneity of large-sized LCD panels difficult, while low transmission rates of near ultraviolet (NUV) impact the processing time of LCD VPP. Furthermore, limitations in light intensity and fixed pixel ratios of digital micromirror devices (DMDs) constrain the increase in the projection area of DLP VPP. This paper identifies these critical issues and provides detailed reviews of available solutions, aiming to guide future research towards developing a more productive and cost-effective high-speed VPP in terms of the high volumetric print rate.

LONG-TERM PROGNOSIS OF CHOROIDAL NEOVASCULARIZATION COMPLICATING ANGIOID STREAKS.

To report the very long-term visual prognosis of choroidal neovascularization complicating angioid streaks in the antivascular endothelial growth factor era. Retrospective monocentric study aimed at analyzing patients' demographics, choroidal neovascularization features, angioid streak-associated conditions, and previous and current therapies for choroidal neovascularization. The main outcome measures were the quantitative measurement of central retinal pigment epithelial atrophy enlargement by comparing the ratio of pixels involved on automated infrared images acquired by spectral-domain optical coherence tomography and the changes in best-corrected visual acuity. The secondary outcome measures were the number of intravitreal injections and the changes in central choroidal thickness and central retinal thickness. Subgroup analyzes were performed to compare macular atrophy extent between eyes of patients with or without proven pseudoxanthoma elasticum ("PXE" or "no PXE") and between eyes previously treated or not with photodynamic therapy ("PDT" or "no PDT"). Thirty-three eyes of 23 patients were included. The mean best-corrected visual acuity decreased significantly from 66 ± 19 Early Treatment Diabetic Retinopathy Study letters at the time of the first antivascular endothelial growth factor injection to 52 ± 23 Early Treatment Diabetic Retinopathy Study letters at the end of the follow-up (mean follow-up duration: 109 ± 42 months, range: 47-175 months). The ratio of central retinal pigment epithelial atrophy enlargement was 201%, 110%, 240%, and 111% in the PXE, no PXE, PDT, and no PDT groups, respectively. Despite the use of antivascular endothelial growth factor agents, the very long-term prognosis appeared relatively poor, especially in patients with PXE. This study also suggests that PDT should be used with caution in the management of choroidal neovascularization in eyes with angioid streaks.

Simulation of video image detection in leisure sports tourism industry based on convolutional neural network

Due to the continuous development of information technology, emerging technologies such as 5G networks, Turing devices and deep learning are widely used in people's lives. At present, the sports and tourism industry has become a pillar industry supporting the national economy. Sports events have attracted many people's attention, and people are willing to devote themselves to sports activities. Participants and visitors are also increasing day by day. The article also conducts research on the rapid detection of human video behavior, obtains more representative features for this demand. This research is of great help to the development of the industry. This allows the network to manipulate the input video behavior at different scales and extract deeper video information. Therefore, this paper proposes another 3D convolutional neural network model that combines a multi-level pyramid network with an attention mechanism. The visual attributes in the video behavior data set are explicitly learned to refine the classification of similar features. This paper also proposes an integrated neural network based on visual attribute enhancement. It also shows good adaptability, and can complete video data processing based on time series. The extraction of time and space functions is realized by the two-stream neural network method, the two networks learn the corresponding pixel ratios, and it is not easy to lose the clues of the action recognition. This paper also proposes a spatiotemporal convolutional neural network video for human body recognition algorithm, which is applied to the leisure tourism industry according to the specific analysis of the above method, and promotes the development of the industry.

JAED-Net: joint attention encoder–decoder network for road extraction from remote sensing images

In road extraction from remote sensing images, the road environment is complex and blocked by trees, buildings, and other objects, making it impossible to extract practical (continuous and complete) road information. We propose a joint attention encoder–decoder network (JAED-Net) for road extraction from remote sensing images to solve these problems. First, JAED-Net encodes a modified residual network as the backbone for road feature extraction. A joint attention module is added to the encoder to enhance the network’s ability to learn and express road features. Then, strip convolution is added to the decoder, so the network retains more spatial features, such as the width and connectivity of roads during upsampling. Finally, a hybrid weighted loss function is introduced to train the network and ensure stability because of the unbalanced ratio of road and background pixels in remote sensing images. Experimental validation of the proposed network is performed on three publicly available datasets.

Measuring liquid-liquid volume composition in transparent tubing using a linear CMOS sensor

Recent advances in milli- and microscale fluidics have led to the miniaturization of liquid-liquid extraction systems. Carryover is a primary mode of failure in these systems, where unwanted liquid appears at the outlet of the opposite phase. Active monitoring of carryover is key to realizing automation and deployment of such fluidic technologies. In this paper, we present an inexpensive sensor and simple methods of analysis to quantify carryover. The sensor measures the volume composition of immiscible liquids using off-the-shelf tubing, a LED light source, a linear CMOS sensor, and shadowgraphy. Beam4 optical modeling is included to provide insight into this sensor's optical behavior. We report measurements of single and two-phase flows and explore average pixel ratio (signal:background) and menisci counting analyses. As an added benefit, we use average pixel ratio to identify bulk liquids and their constituents, specifically toluene, water, and 0–50 wt% sucrose in water.

Maize (Zea mays L.) Stem Target Region Extraction and Stem Diameter Measurement Based on an Internal Gradient Algorithm in Field Conditions

The target region and diameter of maize stems are important phenotyping parameters for evaluating crop vitality and estimating crop biomass. To address the issue that the target region and diameter of maize stems obtained after transplantation may not accurately reflect the true growth conditions of maize, a phenotyping monitoring technology based on an internal gradient algorithm is proposed for acquiring the target region and diameter of maize stems. Observations were conducted during the small bell stage of maize. First, color images of maize plants were captured by an Intel RealSense D435i camera. The color information in the color image was extracted by the hue saturation value (HSV) color space model. The maximum between-class variance (Otsu) algorithm was applied for image threshold segmentation to obtain the main stem of maize. Median filtering, image binarization, and morphological opening operations were then utilized to remove noise from the images. Subsequently, the morphological gradient algorithm was applied to acquire the target region of maize stems. The similarity between the three types of gradient images and the manually segmented image was evaluated by pixel ratio extraction and image quality assessment indicators. Evaluation results indicated that the internal gradient algorithm could more accurately obtain the target region of maize stems. Finally, a checkerboard was employed as a reference for measurement assistance, and the stem diameter of maize was calculated by the pinhole imaging principle. The mean absolute error of stem diameter was 1.92 mm, the mean absolute percentage error (MAPE) was 5.16%, and the root mean square error (RMSE) was 2.25 mm. The R² value was 0.79. With an R² greater than 0.7 and a MAPE within 6%, the phenotyping monitoring technology based on the internal gradient algorithm was proven to accurately measure the diameter of maize stems. The application of phenotyping monitoring technology based on the internal gradient algorithm in field conditions provides technological support for smart agriculture.

Roadway Snow Detection Using Dual-Spectrum Camera Images and Computer Vision

Unfavorable roadway conditions, such as slippery roads, can negatively affect the safety of highway transportation. We aimed to develop a convenient tool capable of evaluating multi-lane road slippery conditions in winter seasons. In this work, field data collection using a dual-spectrum camera was first performed at a field site in the state of Utah, U.S. We analyzed optical and infrared images covering a field of view over three lanes through two snowstorms. Image processing techniques, including image registration, morphological operation, and segmentation, were implemented on both types of images collected under different illumination and temperature conditions. Moreover, the ratio of snow-covered pixels was computed to quantify the snow coverage rate of individual lanes. Finally, we verified the system performance by comparing our estimation with the ground truth via a confusion matrix. The high accuracy, precision, true positive rate, and true negative rate suggest the developed approach can support satisfactory performance for roadway snow detection. Besides, the performance of the unsupervised k-means clustering algorithm and supervised support vector machine (SVM) were evaluated on a dataset of 22 optical images and 19 infrared images. Both the k-means clustering and SVM can support a reasonable image segmentation for roadway snow coverage estimation. Thus, the developed technique offers the potential to facilitate local agencies’ decision-making on snow-plowing resource planning and performance evaluation and support winter safety for connected vehicles.

Measurement Method for Wall Crack Size Based on Pixel Ratio

The identification and measurement of building cracks can reduce the harm caused by accidents to people’s lives. Reducing manual work can improve the efficiency of crack inspection and reduce inspection errors. In this paper, (1) Machine vision with high measurement accuracy and fast response speed is used, the crack contour is extracted by OpenCV splicing, segmentation, morphological processing, and wall cracks can be accurately detected. (2) The crack size is calculated using the pixel ratio algorithm, the image pixel ratio is calculated by the reference object method, and the size of the wall crack is measured. (3) The measurement error is small. In comparing the experimental results and manual measurement, the width measurement error rate is the highest, 0.314%, and the height error rate is the highest, 2.9%. (4) Size information can be displayed to prevent the hazard caused by cracks in advance. So, the system has a good wall crack identification and measurement effect, which reduces labor costs.

Unsupervised Change Detection in Wide-Field Video Images Under Low Illumination

In low-illumination environments such as at night, due to factors such as the large monitoring field of an eagle eye, short sensor exposure time, and high-density random noise, the video images collected by image sensors generally have poor visual quality and low signal-to-noise ratio, which makes it difficult for surveillance systems to detect weak changes. To solve this problem, we propose a method for image change detection (CD) in surveillance video based on optimized k-medoids clustering and adaptive fusion of difference images (DIs). First, for the input multitemporal video surveillance images, two DIs are obtained by log-ratio and extremum pixel ratio operators. Then, the two DIs are adaptively fused by combining the local energy of DIs and the Laplacian pyramid. Simultaneously, the fused DI is compressed by the normalization function, and the final DI is obtained via the improved adaptive median filter. Finally, the changed image is obtained by using the optimized k-medoids clustering algorithm. The experimental results show that the proposed method can accurately and effectively detect weak changes in the eagle eye surveillance picture in a low-illumination environment. Compared with those of other methods, the accuracy and robustness of the proposed method are higher, and the running time of the algorithm is shorter. Moreover, it will not generate a false alarm due to the influence of noise in unchanged scenes.