Canny Edge Research Articles

Two-Stage Method for Clothing Feature Detection

The rapid expansion of e-commerce, particularly in the clothing sector, has led to a significant demand for an effective clothing industry. This study presents a novel two-stage image recognition method. Our approach distinctively combines human keypoint detection, object detection, and classification methods into a two-stage structure. Initially, we utilize open-source libraries, namely OpenPose and Dlib, for accurate human keypoint detection, followed by a custom cropping logic for extracting body part boxes. In the second stage, we employ a blend of Harris Corner, Canny Edge, and skin pixel detection integrated with VGG16 and support vector machine (SVM) models. This configuration allows the bounding boxes to identify ten unique attributes, encompassing facial features and detailed aspects of clothing. Conclusively, the experiment yielded an overall recognition accuracy of 81.4% for tops and 85.72% for bottoms, highlighting the efficacy of the applied methodologies in garment categorization.

Detecting Wear and Tear in Pedestrian Crossings Using Computer Vision Techniques: Approaches, Challenges, and Opportunities

Pedestrian crossings are an essential part of the urban landscape, providing safe passage for pedestrians to cross busy streets. While some are regulated by timed signals and are marked with signs and lights, others are simply marked on the road and do not have additional infrastructure. Nevertheless, the markings undergo wear and tear due to traffic, weather, and road maintenance activities. If pedestrian crossing markings are excessively worn, drivers may not be able to see them, which creates road safety issues. This paper presents a study of computer vision techniques that can be used to identify and classify pedestrian crossings. It first introduces the related concepts. Then, it surveys related work and categorizes existing solutions, highlighting their key features, strengths, and limitations. The most promising techniques are identified and described: Convolutional Neural Networks, Histogram of Oriented Gradients, Maximally Stable Extremal Regions, Canny Edge, and thresholding methods. Their performance is evaluated and compared on a custom dataset developed for this work. Insights on open issues and research opportunities in the field are also provided. It is shown that managers responsible for road safety, in the context of a smart city, can benefit from computer vision approaches to automate the process of determining the wear and tear of pedestrian crossings.

Semantic segmentation of diseases in mushrooms using Enhanced Random Forest

Mushrooms are a rich source of antioxidants and nutritional values. Edible mushrooms, however, are susceptible to various diseases such as dry bubble, wet bubble, cobweb, bacterial blotches, and mites. Farmers face significant production losses due to these diseases affecting mushrooms. The manual detection of these diseases relies on expertise, knowledge of diseases, and human effort. Therefore, there is a need for computer-aided methods, which serve as optimal substitutes for detecting and segmenting diseases. In this paper, we propose a semantic segmentation approach based on the Random Forest machine learning technique for the detection and segmentation of mushroom diseases. Our focus lies in extracting a combination of different features, including Gabor, Bouda, Kayyali, Gaussian, Canny edge, Roberts, Sobel, Scharr, Prewitt, Median, and Variance. We employ constant mean-variance thresholding and the Pearson correlation coefficient to extract significant features, aiming to enhance computational speed and reduce complexity in training the Random Forest classifier. Our results indicate that semantic segmentation based on Random Forest outperforms other methods such as Support Vector Machine (SVM), Naïve Bayes, K-means, and Region of Interest in terms of accuracy. Additionally, it exhibits superior precision, recall, and F1 score compared to SVM. It is worth noting that deep learning-based semantic segmentation methods were not considered due to the limited availability of diseased mushroom images.

Sketch Assisted Face Image Coding for Human and Machine Vision: A Joint Training Approach

Image coding is one of the most fundamental techniques and is widely used in image/video processing and multimedia communications. Current image coding methods are mainly human-oriented, and the visual quality is always unsatisfactory, especially at low bitrates. Moreover, the recent emergence of machine vision goes beyond the scope of current coding. With these considerations, we proposed a sketch assisted face image coding for human and machine vision by a joint training approach. In the proposed approach, we design a new feature representation: a color sketch, which aims to satisfy both low-frequency features of human vision and high-frequency features of machine analysis. Then, we present a novel end-to-end image codec framework with joint training that consists of three models: an image-to-image translation module, a coding module, and a two-stage reconstruction module. Specifically, the input image is first translated into the edge map with the Canny edge as the auxiliary label to merely preserve the structure information. Afterward, the backpropagation from reconstruction module guides the edge map to increase or decrease the information through joint training, which results in the generation of color sketch. Then, the generated sketch is compressed into the bitstream and decompressed back to a sketch in the coding module. Finally, the decompressed sketch is reconstructed to support the machine and human tasks, respectively. In this way, the color sketch is designed to bridge the gap between human and machine vision, and the joint training strategy helps to adjust the low-frequency information in the sketch. The experimental results on challenge datasets demonstrate that our proposed algorithm offers 40.9%-86.6% bitrate savings on machine vision and is comparable to state-of-the-art image coding methods on human vision.

IMPLEMENTASI METODE DETEKSI TEPI CANNY UNTUK MENGHITUNG JUMLAH UANG KOIN DALAM GAMBAR MENGGUNAKAN OPENCV

Edge detection is often used in image processing to find the edge boundaries of image objects and this is important for selecting the edges of image parts. The image edge detection used in this study counts coins in an image, so that it can easily pass through the human eye but not by a computer. In a computer system, it is necessary to provide programs and test data so that the computer can read and count coins like a human. The purpose of this study is to count coins in an image using Canny edge detection and Gaussian filtering methods. The images used in this study are in the form of *jpg and tested using the Visual Studio Code application using OpenCV and Python. The results of this study indicate that the results of the number of coins read from the images inputted to the system can be read properly, with a data accuracy of 90%. Canny's edge detection method can be used to count the number of coins in an image.Keywords: Pengolahan Citra; Deteksi tepi Canny; Gaussian filtering; OpenCV;

Research on Edge Detection and Image Segmentation of Cabinet Region Based on Edge Computing Joint Image Detection Algorithm

Image segmentation (IE) in several disciplines of image processing and computer vision is an essential topic. Segmentation splits a picture into the areas or items that it constitutes. Image segmentation may be achieved with many approaches, some easier than others because of sophisticated programming requirements. The most common technique for segmenting pictures is edge detection (ED) based on sudden (locomotive) intensity fluctuations. This paper aims to study edge detection approaches for the division of images and acquired experimental findings, Sobel, Prewitt, Robert, CannyLoG (Laplacian of Gaussian). It is vital to ensure that picture segmentation algorithms deliver correct results quickly and efficiently for computer vision to reach its full potential. Computer vision approaches require more investigation in hierarchical architectural IoT networks created for seeing the world. In this work, the new way to provide joint image detection (JID) algorithm is to provide multi-scaling approaches for edge detection and segmentation using IoT edge computing (EC). This JID-EC method avoids the requirement to choose and track the edge explicitly. This study provides an overview of fundamental ideas, techniques, and algorithms common to segment images and edge detection, focusing on the segmentation and visualization of joint-articular cartilage images. The reason for this failure is that it is an image noise-sensitive high pass filter. The need for improved algorithms to meet a suitable value of low and high thresholds should thus be stressed for picture noise such as a canny edge, and the performance is achieved with an efficiency of 95.2%.

Computed tomography reconstruction based on canny edge detection algorithm for acute expansion of epidural hematoma

Backgroundand purpose: Acute enlargement of epidural hematoma often indicates a poor prognosis. This paper aims to explore the predictive value of computed tomography (CT) reconstruction based on the Canny segmentation algorithm for acute expansion of epidural hematoma. MethodsA retrospective analysis was made of 73 patients with epidural hematoma in our hospital. They were divided into two groups according to whether there was an acute expansion of hematoma during and after operation: the expansion group and the non-expansion group. Canny edge detection algorithm was used to segment the focus, and three-dimensional (3D) reconstruction was performed to predict the acute expansion of epidural hematoma. Evaluate the effect of 3D reconstruction of epidural hematoma and the application value of 3D CT reconstruction in acute hematoma enlargement. ResultsThe Dice value of the hematoma segmentation result based on Canny's edge detection algorithm was 0.919. The accuracy, sensitivity, and specificity of 3D CT reconstruction were 88.46%, 87.03, and 91.67%, respectively. ConclusionThe Canny edge detection algorithm has good accuracy in segmenting epidural hematoma, and CT reconstruction has an excellent predictive effect on the acute expansion of epidural hematoma.

An algorithm to detect overlapping red blood cells for sickle cell disease diagnosis

In Africa, Uganda is among the countries with a high number of babies (20,000 babies) born with sickle cell, contributing between 6.8% of the children born with sickle cell every year worldwide and approximately 4.5% of the children born with hemoglobinopathies worldwide. It is estimated that by 2050, sickle cell cases will increase by 30% if no intervention is put in place. To facilitate early detection of sickle cell anaemia, medical experts employ machine learning algorithms to detect sickle cell abnormality. Previous research revealed that algorithms for recognizing shape of a sickle cell from blood smear by fractional dimension, cannot detect sickle cells if applied on blood samples containing overlapping red blood cells. In this research, the authors developed an algorithm to detect overlapping red blood cells for sickle cell disease diagnosis. The algorithm uses canny edge and double threshold machine learning techniques and takes overlapping red blood cells images as inputs to detect if these cells are sickle cell anaemic. These images have a scale magnification of (200×, 400×, 650×) pixel taken using a microscope. The algorithm was tested on a total of 1000 digital images and the overall accuracy, sensitivity and specificity were 98.18%, 98.29% and 97.98% respectively.

Robust Interactive Method for Hand Gestures Recognition Using Machine Learning

The Hand Gestures Recognition (HGR) System can be employed to facilitate communication between humans and computers instead of using special input and output devices. These devices may complicate communication with computers especially for people with disabilities. Hand gestures can be defined as a natural human-to-human communication method, which also can be used in human-computer interaction. Many researchers developed various techniques and methods that aimed to understand and recognize specific hand gestures by employing one or two machine learning algorithms with a reasonable accuracy. This work aims to develop a powerful hand gesture recognition model with a 100% recognition rate. We proposed an ensemble classification model that combines the most powerful machine learning classifiers to obtain diversity and improve accuracy. The majority voting method was used to aggregate accuracies produced by each classifier and get the final classification result. Our model was trained using a self-constructed dataset containing 1600 images of ten different hand gestures. The employing of canny's edge detector and histogram of oriented gradient method was a great combination with the ensemble classifier and the recognition rate. The experimental results had shown the robustness of our proposed model. Logistic Regression and Support Vector Machine have achieved 100% accuracy. The developed model was validated using two public datasets, and the findings have proved that our model outperformed other compared studies.

Modified cuckoo search algorithm: Feature subset selection &Shape,Color and Texture Features Descriptors for Content-Based Image Retrieval

With the recent advances in knowledge, the complication of multimedia has increased expressively and new areas of research have opened up in search of new multimedia content. Content-based image retrieval (CBIR) are used to extract images associated with image queries (IQs) from huge databases. The CBIR schemes accessible at present have limited functionality because they only have a partial number of functions. This document presents an improved cookie detection algorithm with coarse sentences for processing large amounts of data using selected examples. The improved cuckoo detection algorithm mimics the behavior of brood attachment parasites in some cuckoo species, including some birds. Modified cuckoo recognition uses approximate set theory to create a fitness function that takes into account the sum of features and the quality of classification as a small amount. For an image entered as IQ from a database, distance metrics are used to find the appropriate image. This is the central idea of CBIR. The projected CBIR method is labelled and can extract shape features based on the RGB color using the and canny Edge (CED) and neutrosophic clustering algorithm scheme. After YCbCrcolor cut, and the CED to get the features to extract the vascular matrix. The combination of these techniques improves the efficiency of the CBR image recovery infrastructure. In this thesis recursive neural network techniques are used to measure the similarity. In addition, the accuracy of the results is: The recall score is measured to evaluate system performance. The proposed CBIR system provides more precise and accurate values than the complex CBIR system.

YOLO 를 이용한 CNN 기반의 물체 인식과 파지 중심점 위치오차 최소화 알고리즘 연구 및 파지 성능 평가

In the last few years, the use of robot manipulators has attracted increasing attention in various industries. Accordingly. Researchers have proposed unique ideas for co-robot control using vision sensors. In this study, you only look once (YOLO) based on convolutional neural network (CNN), and grasping center point position error minimization algorithms were proposed to reduce object misrecognition and increase the performance for grasping an object. In addition, a gripping algorithm was designed for six degree of freedom (DOF) robot manipulators. In addition, machine vision algorithms, including a Grayscale, Gaussian filter, Canny edge, and Contouring, were implemented to detect objects features, such as centroids and orientation. Furthermore, the coordinate system of the vision sensor was converted into a coordinate system of the robot manipulator using a transformation matrix to accurately move the end effector of the robot arm to the center point of the object. The logic implemented in this study not only detected the trained object on the workstation, but also minimized the positional error of the transformation matrix. Additionally, experiments were performed on the 6-DOF robot manipulators. The results revealed that the end effector of the 6-DOF manipulators successfully moved to the center of the detected object, and each of the eight objects was normally gripped.

A Deep Learning Framework for Intrusion Detection and Multimodal Biometric Image Authentication

Nowadays, a demand is increased all over the world in the field of information security and security regulations. Intrusion detection (ID) plays a significant role in providing security to the information, and it is an important technology to identify various threats in network during transmission of information. The proposed system is to develop a two-layer security model: (1) Intrusion Detection, (2) Biometric Multimodal Authentication. In this research, an Improved Recurrent Neural Network with Bi directional Long Short-Term Memory (I-RNN-BiLSTM) is proposed, where the performance of the network is improved by introducing hybrid sigmoid-tanh activation function. The intrusion detection is performed using I-RNN-BiLSTM to classify the NSL-KDD dataset. To develop the biometric multimodal authentication system, three biometric images of face, iris, and fingerprint are considered and combined using Shuffling algorithm. The features are extracted by Gabor, Canny Edge, and Minutiae for face, iris, and fingerprint, respectively. The biometric multimodal authentication is performed by the proposed I-RNN-BiLSTM. The performance of the proposed I-RNN-BiLSTM has been analysed through different metrics like accuracy, f-score, and confusion matrix. The simulation results showed that the proposed system gives better results for intrusion detection. Proposed model attains an accuracy of 98% for the authentication process and accuracy of 98.94% for the intrusion detection process.

Towards Grading Chest X-rays of COVID-19 Patients Using a Dynamic Radial Basis Function Network Classifier

The high volume of COVID-19 Chest X-rays and less number of radiologists to interpret those is a challenge for the highly populous developing nations. Moreover, correct grading of the COVID-19 stage by interpreting the Chest X-rays manually is time-taking and could be biased. It often delays the treatment. Given the scenario, the purpose of this study is to develop a deep learning classifier for multiple classifications (e.g., mild, moderate, and severe grade of involvement) of COVID-19 Chest X-rays for faster and accurate diagnosis. To accomplish the goal, the raw images are denoised with a Gaussian filter during pre-processing followed by the Regions of Interest, and Edge Features are identified using Canny's edge detector algorithm. Standardized Edge Features become the training inputs to a Dynamic Radial Basis Function Network classifier, developed from scratch. Results show that the developed classifier is 88% precise and 86% accurate in classifying the grade of illness with a much faster processing speed. The contribution lies in the dynamic allocation of the (i) number of Input and Hidden nodes as per the shape and size of the image, (ii) Learning rate, (iii) Centroid, (iv) Spread, and (v) Weight values during squared error minimization; (vi) image size reduction (37% on average) by standardization, instead of dimensionality reduction to prevent data loss; and (vii) reducing the time complexity of the classifier by 26% on average. Such a classifier could be a reliable assistive tool to human doctors in screening and grading COVID-19 patients and in turn, would help faster management of the patients as per the stages of COVID-19.

Crop Phenology-Based, Object-Oriented Classification Approach Using SENTINEL-2A and NDVI Time Series: Sunflower Crops in Kırklareli TURKEY

The aim of this study is to develop a methodology for determining sunflower cultivated areas with the help of high resolution SENTINEL-2A satellite images time series representing the phenological stages of the crop growth cycle, and its application in Kırklareli province. Spectral information representing phenological periods was obtained with the help of satellite images and normalized difference vegetation index (NDVI) time series, and an object-oriented classification approach was developed based on this spectral information database. Segmentation and classification decision tree algorithms were produced by using this spectral information database, object shape criteria and other auxiliary thematic maps. The best performance in segmentation was achieved by increasing the weight coefficient of the "Canny edge” layer, which is the edge determination layer defined in the multiresolution method of "Canny edge” algorithm to define the agricultural parcels. Object-oriented classification was carried out based on the this segmented parcels. First, summer, winter, fallow and continuous green areas were determined through the classification decision tree algorithms. The summer and winter crops were classified using the parcel spectral information of the crop-based learning samples that allocated in field work. The crops for which class definition could not be made were passed through a second elimination in the "unclassified" group and later assigned to their classes. In the last stage, parcels whose class definition could not be made were named as "other" class. According to the confusion matrix and accuracy analysis results, sunflower, which was determined in two classes as early and late sowing, was classified at 98% and 92% accuracy, respectively.

A Novel Approach to Detect Abnormal Chest X-rays of COVID-19 Patients Using Image Processing and Deep Learning

The study proposes a novel approach to automate classifying Chest X-ray (CXR) images of COVID-19 positive patients. All acquired images have been pre-processed with Simple Median Filter (SMF) and Gaussian Filter (GF) with kernel size (5, 5). The better filter is then identified by comparing Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR) of denoised images. Canny's edge detection has been applied to find the Region of Interest (ROI) on denoised images. Eigenvalues [-2, 2] of the Hessian matrix (5 × 5) of the ROIs are then extracted, which constitutes the 'input' dataset to the Feed Forward Neural Network (FFNN) classifier, developed in this study. Eighty percent of the data is used for training the said network after 10-fold cross-validation and the performance of the network is tested with the remaining 20% of the data. Finally, validation has been made on another set of 'raw' normal and abnormal CXRs. Precision, Recall, Accuracy, and Computational time complexity (Big(O)) of the classifier are then estimated to examine its performance.

An improved K-means algorithm for underwater image background segmentation

Conventional algorithms fail to obtain satisfactory background segmentation results for underwater images. In this study, an improved K-means algorithm was developed for underwater image background segmentation to address the issue of improper K value determination and minimize the impact of initial centroid position of grayscale image during the gray level quantization of the conventional K-means algorithm. A total of 100 underwater images taken by an underwater robot were sampled to test the aforementioned algorithm in respect of background segmentation validity and time cost. The K value and initial centroid position of grayscale image were optimized. The results were compared to the other three existing algorithms, including the conventional K-means algorithm, the improved Otsu algorithm, and the Canny operator edge extraction method. The experimental results showed that the improved K-means underwater background segmentation algorithm could effectively segment the background of underwater images with a low color cast, low contrast, and blurred edges. Although its cost in time was higher than that of the other three algorithms, it none the less proved more efficient than the time-consuming manual segmentation method. The algorithm proposed in this paper could potentially be used in underwater environments for underwater background segmentation.

Palmprint Recognition Based on Edge Detection Features and Convolutional Neural Network

Research on biometric technology get much attention from researchers who interest in the recognition system. One of the biometric objects that will continue to be developed is the palmprint. The hand palm line has a unique characteristic in each person or may not be the same. The palmprint image is easy to capture because clearly visible, so it does not require a specific sensor. This paper presents the automatic extraction feature with Convolutional Neural Network (CNN) technique to get a unique characteristic of palmprint image and identify a person. CNN will get easier to classify the image database if it has many data. CNN belongs to Supervised Learning, which requires training data to create a knowledge base. In the dataset with little training data, the system must increase the training data using augmentation methods like zoom, shear, and rotate. Still, in the palmprint, that augmentation method can change the original character of the palmprint. Our proposed method is adding training data with an edge detection image from the original image. Edge detection used in our method is Canny and Sobel. The addition of Canny and Sobel edge detection for training data is the best combination scenario for palmprint recognition. The experiment results showed that palmprint recognition using Convolution Neural Network with Canny and Sobel edge detection for training data resulted in an accuracy rate of 96.5% for 200 classes, and the Equal Error Rate (ERR) value is 3.5%. This method has been able to recognize 193 palms of 200 people.

Nonlinear RANSAC with crossline correction: An algorithm for vision-based curved cable detection system

Accurate curved line shape object detection is important for assistant Autonomous Underwater Vehicles)AUVs( to complete underwater cable inspection tasks. In this article an algorithm based on adaptive canny's edge detection and an improved random sample consensus (RANSAC) algorithm, called Crossline Correction Non-linear RANSAC (CCNL-RANSAC), is proposed to detect curved line object. The algorithm is featured on adaptive threshold in the canny's operator to improve correctness of edge detection, preliminary check and cross line correction introduced in a nonlinear RANSAC method to locate cables with better success rate and less execution time. Comprehensive underwater cable detection experiments were carried out by a range-gated imaging system in a big water tank. With the proposed CCNL-RANSAC algorithm, we increased success rate to 96% ~99% in line detection for the images of curved cable taken as far as 21 m long in clear water. The average execution time is reduced from 202 ms (using standard RANSAC) to 88 ms on average using CCNL-RANSAC, with 80% time reduction at most. With the help of CCNL-RSNSAC algorithm proposed in the paper, vision based AUVs is able to make auto inspection of curved cables and pipeline underwater, which will make the periodical checking of those marine structures more efficiently.

Clicker Bot for Gacha Games Using Image Recognition

Gacha games has tedious activities need to be done by the players to progress the game besides using real money. The usage of bots and macros to automate the game are not uncommon nowadays with the high demand of grinding within the game. This research purposes is to create a bot that can replace the user task to click buttons on game stages, thus helping users in their grinding. The automatic clicker applied image recognition to find the area to be clicked. Firstly, the template image is manually taken by the user to be used as the matching. The screen image, which captured on the go, and the template image are pre-processed using three variations techniques to find the best precision of the coordinate area. Afterwards, the cursor will be moved to the coordinate position returned from the matching between template image and screen image. The coordinate returned from the matching is randomized using Gaussian Distributions but still within the matched positions between template image and screen image. The experiment shows that Canny Edges gives the good coordinate position. On the other hand, the grayscale gives a poor result on giving the coordinate position. Canny Edge and combination of Canny Edge and grayscale always return the target’s position. With this bot, user does not need to supervise the game every time the game stage has finished and restarting the stage.

Erratum to: Identification of reservoir fractures on FMI image logs using Canny and Sobel edge detection algorithms

Erratum to: Identification of reservoir fractures on FMI image logs using Canny and Sobel edge detection algorithms