Sobel Edge Detection Research Articles

Modified restoring array-based power efficient approximate square root circuit and its application

In this paper, the authors present an investigation into the benefits of approximate computing in energy-efficient error-resilient applications. It proposes an approximate square root (SQR) circuit based on the modified restoring array square root (MRAS) architecture. The MRAS design incorporates strategic elimination, optimization, and simplification of restoring subtractor cells (SC) within a conventional restoring array SQR circuit. As a result, the proposed MRAS circuit efficiently computes the square root of a 2n-bit unsigned integer, offering a 27 percent reduction in area, 13 percent faster computation, and 50 percent lower power consumption compared to the exact SQR design. To introduce approximation, configurable SCs are integrated into the MRAS circuit, enabling dual operation modes for both accuracy and approximation. This feature allows the proposed approximate modified restoring array SQR (AMRAS) design to cater to both error-resilient and error-sensitive applications. The evaluation involves a thorough analysis of accuracy and design metrics for 16-bit unsigned exact, state-of-the-art, and proposed square rooters. The designs are implemented on Artix7 FPGA using Verilog-HDL and simulated in the Xilinx Vivado simulator. The results demonstrate that the proposed AMRAS achieves a good trade-off between accuracy and hardware, presenting an impressive 80 percent reduction in power consumption and a 36 percent faster computation. Additionally, the paper showcases the practical application of the proposed AMRAS square rooters in the Sobel edge detection for image processing.

Direct feature extraction and image co-registration of morphological structure from Bayer pattern raw planetary images

Feature extraction from planetary remote sensing images is a primary image processing task for object recognition, crater counter, morphological structure dimension and unpaired image co-registration. The paper presents a novel methodology to extract features directly from Bayer Pattern raw planetary images. Gradient information is extracted from Bayer Pattern image using standard edge operator that follows Color Difference Constancy (CDC) assumption. The proposed method's advantage is that it can skip the computationally intensive image processing pipeline and can have the Bayer Pattern raw planetary image flow directly for useful information extraction. Sobel Edge Detector is applied on Indian Mars Color Camera (MCC) Bayer Pattern raw images, and Gradient Magnitude Map (GMM) is generated at different Martian terrains. In addition, we have developed a direct image co-registration approach for MCC Bayer intensity raw image with respect to Mars Digital Image Model (MDIM) 2.1 reference using Mode-Mean Combo Patch Filler and Gradient Intensity induced Scale Invariant Feature Transform (GI-SIFT) based feature matching. The outlier matched points are removed by Feature Similarity Score guided Random Sample Consensus (FSS-RANSAC) estimation technique. The visual evaluation and quantitative metrics indicate that GMM from MCC Bayer Pattern raw image has negligible degradation with respect to GMM extracted from MCC demosaic image. The Root Mean Square Error (RMSE) is computed at different Mars regions, and it is found that the average image co-registration accuracy is less than 0.5 pixel.

Fast Feature Extraction Method for Brillouin Scattering Spectrum of OPGW Optical Cable Based on BOTDR.

Brillouin optical time domain reflectometry (BOTDR) detects fiber temperature and strain data and represents one of the most critical ways of identifying abnormal conditions such as ice coverage and lightning strikes on optical fiber composite overhead ground wire (OPGW) cable. Existing BOTDR extracts brillouin frequency shift (BFS) features with cumulative averaging and curve fitting. BFS feature extraction is slow for long-distance measurements, making realizing real-time measurements on fiber optic cables challenging. We propose a fast feature extraction method for block matching and 3D filtering (BM3D) + Sobel brillouin scattering spectroscopy (BGS). BM3D takes the advantage of non-local means (NLM) and wavelet denoising (WD) and utilizes the spatial-domain non-local principle to enhance the denoising in the transform domain. The global filtering capability of BM3D is utilized to filter out the low cumulative average BGS noise and the BFS feature extraction is completed using Sobel edge detection. Simulation verifies the feasibility of the algorithm, and the proposed method is embedded in BOTDR to measure 30 km of actual OPGW line. The experimental results show that under the same conditions, the processing time of this method is reduced by 37 times compared to that with the 50,000-time cumulative averaging + levenberg marquardt (LM) algorithm without severe distortion of the reference resolution. The method improves the sensor demodulation speed by using image processing technology without changing the existing hardware equipment, which is expected to be widely used in the new generation of BOTDR.

ADCGNet: Attention-based dual channel Gabor network towards efficient detection and classification of electrocardiogram images

Heart disease is a major health issue, and accurate diagnosis of irregular heartbeats and heart failure is crucial. Current diagnostic processes can be time-consuming, requiring significant effort from clinicians. An effective classifier, ADCGNet: Attention-based Dual Channel Gabor Network is proposed to address this challenge by accurately classifying anomalies. ADCGNet involves pre-processing every ECG beat into two-dimensional images using Analytical Morlet transform and then applying thirty-two Gabor filters and Sobel edge detection to enhance features. ADCGNet comprises three blocks, with the first block using dual channels to extract essential features in the images efficiently. The second block includes a multi-head attention mechanism to focus on relevant features, and the third block uses a SoftMax activation function to perform classification tasks. Extensive experiments with public datasets from PhysioNet, and comparison with several state-of-the-art classifiers indicate ADCGNet is superior. Specifically, ADCGNet achieved an accuracy of 99.17%, 98.98% in precision, a recall of 98.87%, an F1-score of 98.82% and AUC, 98.75% with optimal hyperparameters. Further, a GRAD-CAM visualization of activated areas on the test samples gives graphical insight into the performance of ADCGNet. The proposed ADCGNet classifier has promising potential for enhancing the diagnosis of heart disease, and we believe it will be of much interest to the medical community.

Sobel edge detection for quantifying the effectiveness of focused ultrasound thalamotomy for tremor relief

Abstract Focused ultrasound on the thalamus is employed to relieve tremor in movement disorders. The treatment efficacy is currently assessed by subjective observation of patient tremor during hand movements. The purpose of this study was to create a simple quantitative method to assess tremor based on traditional pen-and-paper spiral drawing that could be employed during treatment and follow-up visits. Image processing was employed to automatically and rapidly clean scanned spiral and line drawings and to isolate the minimal part of the drawing that could provide sufficient data for tremor detection. Sobel edge detection and easily-interpretable statistics of the distribution of edge-angle orientations in the spirals relative to the spiral centres were used as tremor measures. The method was applied to scanned drawings of 122 tremor patients before treatment and in follow-up visits. Discrimination analysis of the tremor measure between drawings in the session before treatment and the follow-up sessions was carried out. The processing yielded correctly cropped and clean spirals. The tremor measure demonstrated consistent tremor reduction in the treated hand drawings in all five follow-up visits during the first year following treatment date in 76 patients from the study cohort. In this study we developed a computational method that rapidly computed tremor severity estimate in an explainable and clinically useful manner from a traditional pen-and-paper spiral drawing. This rapid quantitative and objective tremor assessment can easily be implemented during focused ultrasound treatment and replace the subjective assessment currently employed.

High-Accuracy Measurement of Three-Dimensional Inclinations of Power Transmission Towers Using Reconstruction Model of Stereo Images

The traditional inclination measurement method usually measures one-dimensional or two-dimensional inclinations of power transmission towers. This research proposed a method, which is feasible to measure three-dimensional (3D) inclinations using the stereo images of power transmission towers. The main work of this method is the 3D reconstruction of power transmission tower model using stereo images and the calculation of 3D inclinations. As for the 3D reconstruction, a method of calculating the matching cost using a combination of multichannel red–green–blue stereo pairs and Sobel edge detection stereo pairs is proposed to obtain accurate matching cost, and a method of invalid disparity judgment is proposed to detect the invalid disparities in small closed areas; as for the calculation of 3D inclinations, a 3D inclination calculation method is put forward to measure 3D inclinations by rotation and translation of different 3D reconstruction models. The experiment shows that the 3D reconstruction quality is good and the results of the 3D inclinations are accurate, which is of great value in the static or dynamic measurement of 3D inclinations of power transmission towers.

A Novel Approach to Evaluate the Wear Behavior of CuO Doped Sintered Alumina Ceramics Using Image Processing Models

The present research describes the post wear SEM images of nano-CuO doped alumina ceramics in the light of three different image processing algorithms, viz. entropy analysis, Sobel edge detection techniques and entropy filtered image histogram analysis. The 2 wt% CuO doped sintered alumina has shown the best performance towards wearing due to maximum bulk density, hardness and fracture toughness. These materialistic properties are fitted to correlate with the lowest value of entropy and edge density index at the level of 2 wt% doping. The entropy filtered image histogram of 2 wt% doping is the closest to the Gaussian Bell shape distribution with the lowest skewness factor. These three-image computed methods validate their suitability in developing a real-life implementation for a possible wear resistance estimator.

A geometric calibration of the tip of the red giant branch in the Milky Way using Gaia DR3

ABSTRACT We use the latest parallaxes measurements from Gaia DR3 to obtain a geometric calibration of the tip of the red giant branch (TRGB) in Cousins I magnitudes as a standard candle for cosmology. We utilize the following surveys: SkyMapper DR3, APASS DR9, ATLAS Refcat2, and Gaia DR3 synthetic photometry to obtain multiple zero-point calibrations of the TRGB magnitude, $M_{I}^{TRGB}$. Our sample contains Milky Way halo stars at high galactic latitudes (|b| > 36) where the impact of metallicity, dust, and crowding are minimized. The magnitude of the TRGB is identified using Sobel edge detection, but this approach introduced a systematic offset. To address this issue, we utilized simulations with parsec isochrones and showed how to calibrate and remove this bias. Applying our method within the colour range where the slope of the TRGB is relatively flat for metal-poor halo stars (1.55 < (BP − RP) < 2.25), we find a weighted average $M_{I}^{TRGB} = -4.042 \pm 0.041$ (stat) ±0.031 (sys) mag. A geometric calibration of the Milky Way TRGB has the benefit of being independent of other distance indicators and will help probe systematics in the local distance ladder, leading to improved measurements of the Hubble constant.

A Novel Algorithm of Distance Calculation Based-on Grid-Edge-Depth-Map and Gyroscope for Visually-Impaired

This paper presented a new algorithm for determining the distance of an object in front of a stereo camera placed on a helmet. By using a stereo camera with a Sum of Absolute Difference with a Sobel edge detector, our previous Grid-Edge-Depth map algorithm could calculate the objects’ distance up to 500 cm. The problem started when a vision disability person used the device with an unfixed stereo camera angle. The unspecified angle caused by the helmet’s movement influenced the distance calculation result. This novel process started with calculating the distance from a Grid-Edge-Depth map considering unfixed angle data of the x-axis from a gyroscope sensor placed on the stereo camera using the trigonometry formula. The angle data used was the x-axis data. The distance measurement results by the system were then computed based on the unfixed angle compared to the actual distance. The test was carried out with three scenarios which required the user to stand at a distance of 100 cm, 125 cm, and 150 cm from a table, chair, or wall, with 30 tests for each scenario. The test results showed an average accuracy of 96.05% with three experimental scenarios, which meant that this machine was feasible to implement.

Automatic vehicle number plate recognition system

In simple terms, Automatic Vehicle Number Plate Recognition System (AVNPRS) can be described as a computer vision image processing technology. This technology captures images of vehicles and recognizes their number plates. As the world is becoming more sophisticated on each new day, security as well as management of transportation system have become a vital aspect in controlled places. Such places include modern campuses, shopping malls and other institutions. With today’s rate at which motor vehicles are increasing in numbers, there is need for an effective, affordable and efficient AVNPRS. Thus, this project proposes, outlines and discusses an efficient AVNPRS. The system under consideration in this project can be installed at main entrances of modern institutions. This is because the AVNPRS ensures that only authorized vehicles can automatically have access to such institutions. The AVNPRS ensures that it captures the image of the number plate once the input sensor detects the vehicle. The Sobel edge detection and Laplacian edge detection techniques are used at this critical stage. Thereafter, the Bounding box technique is used to find the number plate leading to character segmentation. After capturing, an image undergoes extraction and character investigation via the Optical Character Recognition (OCR). In addition, in achieving character recognition, matching between the computer template and segmented image is done via the OCR method. It is vital to mention that this system is sustainable as it successfully detects, recognizes and processes vehicle number plates on real images. The AVNPRS can be used for both traffic control and security. Therefore, the main aim of this piece of writing is to develop from a theoretical perspective an AVNPR system that can detect and capture vehicle number plate images.

Identification of land and water Centella asiatica leaf herbal plants using digital imagery with the Sobel Edge Detection algorithm

Centella asiatica leaves or gotu kola leaves are wild plants that grow in Asian countries such as China, Indonesia, Japan and India. Since thousands of years ago, this gotu kola leaf has been known to treat various diseases. This plant is even used as a traditional herbal medicine in China and India. Centella asiatica is an annual herbaceous plant that grows and flowers throughout the year. Plants will thrive if the soil and environment are suitable to be used as a ground cover. Types of gotu kola that are often found are red gotu kola and green gotu kola. Centella asiatica is also known as antanan taman or antanan batu because it is found in rocky, dry and open areas. Centella asiatica grows with stolons and has no stems, but has rhizomes (short rhizomes). Meanwhile, green gotu kola is often found in rice fields and on the sidelines of the grass. Based on this problem, a study is needed to develop a system to determine the shape of leaf fiber density with a comparison of ground gotu kola and water gotu kola using image processing techniques to find the diameter. This measurement process uses the Matlab application and tests with the Sobel edge detection method and image processing to see edges that are more clearly visible. The results showed that the developed system was capable of obtaining images and identifying the fiber density of Centella asiatica leaves. The system was designed with Jupyter Notebook Python-based programming language analysis with image data taken via internet sources as research material.

Smart Edge Detection Technique in X-ray Images for Improving PSNR using Sobel Edge Detection Algorithm with Gaussian Filter in Comparison with Laplacian Algorithm

Aim: The aim of this study is to propose smart edge detection techniques in x-ray images for improving PSNR using the sobel edge detection algorithm and comparing it with the laplacian algorithm. Materials and methods: For the design of edge detection technique to improve PSNR Sobel edge detection algorithm is used along with the gaussian filter and it is compared with the laplacian algorithm. Sobel edge detection algorithm and laplacian algorithm are the two groups considered in this study. For each group, the sample size is 20 and the total sample size is 40. Sample size calculation was done using clinicalc. com by keeping g-power at 80%, confidence interval at 95%, and the threshold at 0.05%. Result: When comparing the two algorithms, it is clear that the Sobel edge detection algorithm has a higher mean PSNR value of 39.15db than the laplacian algorithm 36.79db. It is observed that the Sobel edge detection algorithm performed better than the laplacian algorithm by performing an independent sample t-test. The p value is 0.09 which is greater than the normal value(p>0.05). Conclusion: Sobel edge detection has insignificantly greater PSNR when compared to the Laplacian algorithm.

Penerapan Convolutional Neural Networks Menggunakan Edge Detection Untuk Identifikasi Motif Jenis Batik

Batik is the work of the Indonesian nation which is a combination of art and technology by the ancestors of the Indonesian people. UNESCO designated batik as a Humanitarian Heritage for Masterpieces of the Oral and Intangible Heritage of Humanity, followed by a presidential decree on 2 October 2009 which was designated as Indonesia's National Batik Day. In maintaining the existence of batik in the era of technological advances, they have utilized Artificial Intelligence and Machine Learning technologies. Even though research on batik motifs has become a common topic, there are still many mistakes found during the process of identifying motifs. As is well known, Indonesia has various types of traditional batik which have very diverse colors, patterns and motifs. This is the main problem in the identification of batik motifs, where it is found that many batik motifs adopt or have similarities in either pattern or color to other batik motifs, giving rise to thin predictions when identified with other motifs. This study uses the Convolutional Neural Networks (CNN) method using Edge Detection to identify batik motifs. Using the main dataset of 1106 images divided into 4 classes, namely Kawung, Megamendung, Merak Ngibing, and Parang motifs. The research is focused on comparing the results of prediction effectiveness on the CNN model with Canny edge detection (CNN-Canny) and the CNN model with Sobel edge detection (CNN-Sobel). The training process for each model is applied the same configuration to the dataset with a ratio of 8:2. Then in the model augmentation the functions of random flip, random zoom, and random invert are applied. The results obtained from learning the machine learning model at the testing and validation stage on the CNN-Sobel model obtained an accuracy of 91.2% in the training process and 91.8% in the validation process. Meanwhile, CNN-Canny got 90.5% accuracy in the training process and 86.2% in the validation process. The results of the comparison of the performance of the two models which are mapped to the confusion matrix table on 16 data testing in the form of images of batik motifs that have never been studied by each model show that the CNN-Sobel model can work more optimally than the CNN-Canny model with an accuracy comparison percentage of 94%. compared to 76% in the process of identifying batik motifs.

A Novel Defect Segmentation for Pulse Infrared Images Based on Improved Fuzzy C-Means Algorithm With Weighted Distance

Pulsed infrared nondestructive testing (NDT) technology can effectively identify defects in engineering materials or structures. However, the original infrared images have the problems of low signal-to-noise ratio (SNR) and blurred edges. A novel defect segmentation method for pulsed infrared images based on improved fuzzy C-means algorithm with weighted distance is proposed. First, the original infrared images are denoised by wavelet transform. Second, an improved fuzzy C-means algorithm based on weighted distance (WDFCM) is proposed to cluster the infrared images. Afterward, the maximum between-cluster variance method (OTSU) is used to automatically select the thresholds and the binarization is performed to extract defect features. The Sobel edge detection operator is used to completely extract the structural properties of the defect regions. Finally, an experimental platform for pulsed infrared thermal imaging was built with two kinds of materials: flexible pipes and steel structures. The method proposed in this article was used for image segmentation subsequently. The results verify that the performance of the proposed method is optimal. The accuracy of defect detection can reach 98.95%, which is 16.82% higher than the original method.

Image processing and machine learning‐based bone fracture detection and classification using X‐ray images

AbstractOne of the most important problems in orthopedics is undiagnosed or misdiagnosed bone fractures. This can lead to patients receiving an incorrect diagnosis or treatment, which can result in a longer treatment period. In this study, fracture detection and classification are performed using various machine learning techniques using of a dataset containing various bones (normal and fractured). Firstly, the X‐ray images obtained are subjected to image preprocessing stages and are prepared for the feature extraction stage. Then, in addition to the Canny and Sobel edge detection methods used in the image processing stage, feature extraction of X‐ray images is performed with the help of Houhg line detection and Harris corner detector. The data obtained by performing feature extraction is given to 12 different machine learning classifiers and the results are presented. Setting hyperparameters for classifiers is done by the grid search method, and the study is tested for 10‐fold cross‐validation. Classifier results are presented comparatively as accuracy, training time, and testing time, and linear discriminant analysis (LDA) reaches the highest accuracy rate with 88.67% and 0.89 AUC. The proposed computer‐aided diagnosis system (CAD) will reduce the burden on physicians by identifying fractures with high accuracy.

CLASSIFICATION OF RICE USING CONVOLUTIONAL NEURAL NETWORK (CNN)

This paper describes the technique for automatic recognition and classification of different rice grain samples using neural network classifier. The Red Green Blue (RGB), Hue Saturation Intensity (HSI) and Hue Saturation Value (HSV) color models of the image were considered for extracting 18 color features. The classification was carried out using color and texture features separately. The color image was converted to Gray scale image and the Gray Level Co-occurrence Matrixes (GLCM) for four different directions was calculated. A total of eight texture features were calculated from the Co-occurrence matrices. Convolutional Neural Network (CNN) is used for the classification process. The classification accuracy with color features and texture features were compared. Result shows that the classification base on texture features outperform the color feature-based classification even with lesser number of features. It is found that Convolutional Neural network was able to classify two varieties of rice with 100% accuracy using texture features and the edge detection with Sobel and Canny edge detection of the fiber features in the food grain.

Sobel Edge Detection Algorithm with Adaptive Threshold based on Improved Genetic Algorithm for Image Processing

In this paper, a novel adaptive threshold Sobel edge detection algorithm based on the improved genetic algorithm is proposed to detect edges. Because of the influence of external factors in actual detection process, the result of detection is often not accurate enough when the configured threshold of the target image is far away from the real threshold. Different thresholds of images are calculated by improved genetic algorithm for different images. The calculated threshold is used in edge detection. The experimental results show that the image processed by the improved algorithm has stronger edge continuity. It is shown that proposed algorithm has a better detection effect and applicability than the traditional Sobel algorithm.

Automatic Optic Disc Detection in Retinal Images Using FKMT‒MOPDF

In recent days, detecting Optic Disc (OD) in retinal images has been challenging and very important to the early diagnosis of eye diseases. The process of detecting the OD is challenging due to the diversity of color, intensity, brightness and shape of the OD. Moreover, the color similarities of the neighboring organs of the OD create difficulties during OD detection. In the proposed Fuzzy K‒Means Threshold (FKMT) and Morphological Operation with Pixel Density Feature (MOPDF), the input retinal images are coarsely segmented by fuzzy K‒means clustering and thresholding, in which the OD is classified from its neighboring organs with intensity similarities. Then, the segmented images are given as the input to morphological operation with pixel density feature calculations, which reduce the false detection in the small pixel of the OD. Finally, the OD area is detected by applying the Sobel edge detection method, which accurately detects the OD from the retinal images. After detection optimization, the proposed method achieved Sensitivity (SEN), Specificity (SPEC) and Accuracy (ACC), with 96.74%, 96.78% and 96.92% in DiaretDB0 (Standard Diabetic Retinopathy Database Calibration level 0), 97.12%, 97.10% and 97.75% in DiaretDB1 (Standard Diabetic Retinopathy Database Calibration level 1) and 97.19%, 97.47% and 97.43% in STARE (Structured Analysis of the Retina) dataset respectively. The experimental results demonstrated the proposed method’s efficiency for segmenting and detecting OD areas.

Image Processing and Feature Extraction for Hull Structure GMAW Based on Weld Pool Visual Sensing

Image processing and feature information extraction based on the visual perception of the weld pool are considered essential components of intelligent welding quality monitoring of hull structure gas metal arc welding (GMAW). The unstable characteristics, such as large spatter, much smoke, and strong arc light during hull structure GMAW, lead to the blurring of image acquisition and the difficulty of contour extraction of the weld pool. The present study is aimed at addressing the practical issues from two perspectives, i.e., a spectrum‐visual‐sensing acquisition system and an image‐processing and feature extraction algorithm. First of all, by analyzing the light energy distribution law and acquiring the optical parameters relevant to the cut‐off composite dimming and near‐infrared narrowband filtering, spectral sensing is employed in establishing models of arc light radiation to detect the strength of continuously distinctive spectral lines. Besides, an appropriate high‐speed charge‐coupled device (CCD) camera is selected to build a visual acquisition system, which can reduce the external interference of the arc light on the image acquisition of the weld pool. Afterwards, the implementation of an image‐processing fusion model based on the spatial information fuzzy C‐means (FCM) clustering analysis and Sobel edge detection operator accompanies the investigation of the geometric aspects of the weld pool image. In terms of clear segmentation of the interest region, the edge detection and accurate extraction of the target contour are successfully obtained. In the subsequent section, the Hough transform analysis is adopted to establish the geometric feature extraction model of the weld pool, with corner detection, conversion, and camera calibration as the core technology. Additionally, the left and right views of the image contour are calibrated to achieve the lossless conversion of corner pixels and physical coordinates. Finally, three other distinct image‐processing methods are designed to compare the segmentation effect of the edge contour with the fusion model, and then, the extraction accuracy of the geometric features of the weld pool is verified. The interference of the arc light and smoke has been demonstrated to be substantially diminished, which is attributable to the visual‐sensing system during image acquisition of the weld pool. The results of edge fusion of the weld pool image show that based on the GMAW using the FCM‐Sobel fusion method, a superior extraction accuracy of geometric features characterized by smoothness, continuousness, no breakpoints, and less noise has fulfilled the engineering requirements.

CFLHCF: Simultaneous Detection of the Optic Disc and Exudates Using Color Features, Local Homogeneity and Contextual Features

Detecting Optic Disc (OD) and Exudates (EXs) in the fundus images has been challenging and demanding for the computer-aided diagnosis system. Existing algorithms for detecting OD and EXs are mainly based on traditional learning methods that heavily rely on enhanced OD and EXs features. Unlike traditional learning methods, a novel simultaneous detection of OD and EXs is presented. In this proposed novel Color Features, Local Homogeneity and Contextual Features (CFLHCF), the input original fundus image is preprocessed by color normalization, contrast enhancement, noise removal, and OD localization which the EXs is differentiated from its background information. Then, the preprocessed images are given as the input to Mathematical Morphology Binary Segmentation (MMBS) with Sobel Edge Detection (SED) technique, which detects the EXs from the given fundus images. An MMBS with a SED technique is implemented to boost a highly accurate segmented model for small EXs regions of EXs. The DiaretDB0, DiaretDB1, and STARE datasets are used to validate the proposed method. On the DiaretDB0 dataset, this technique archived an average sensitivity of 98.44% for EXs and a specificity of 98.72% for non‒EXs, which can potentially classify EXs even when the regions are trivial. With respect to sensitivity and specificity values, this method outperformed the previous state‒of‒the‒art methods by roughly 1.24% and 1.09% in the detection of EXs. Additionally, we show the EXs‒diagnosis in ~7 seconds per image.