Edge Detection Filter Research Articles

Implementation of Sobel filter using CUDA

The Sobel filter is an edge detection filter that results in highlighting and focusing the edges. It is used for extracting the image gradient and contours in image processing applications. The data sets we are using in the paper are PGM and TIFF images with different rows and column pixel dimensions. The sobel filter operations used are implemented in C and CUDA programming language and the time taken for the operations in the serial and parallel platforms respectively are analyzed.

Digital filters applications on aeromagnetic data for identification of hidden objects

The hidden objects are located, as sixteen objects (anomalies) on the reduction to magnetic pole data (RTP). The follow-up and identification of these objects by applying several digital filters, as upwarding continuation at different level, then the horizontal derivative (HD) and vertical derivative (VD). Accordingly, the edge detection filter techniques are dependent on the tilt derivative (horizontal, vertical derivatives) and the tilt angle. This study shows closing to the basic dykes (shallow anomalies) that imprinted in the RTP map and the upward continuation maps at shallow levels. The residual features are mostly starting to disappear gradually in the upward continuation maps at intermediate and deeper levels. The last upwarding maps illustrate the penetrated roots of these bodies to deep levels (deep anomalies). The advanced filter technique of edge detection is the enhanced total horizontal derivative of the tilt angle (ETHDR). This study delineates and compares the hidden objects by applying the different filter techniques, as the analytical signal (AS), first vertical derivative (FVD), total horizontal derivative (THDR), tilt angle (TA), horizontal tilt angle (TDX), and ETHDR. The applied derivative-based filters, as ETHDR, produce more detailed results for the shallower and deeper magnetized structures. The compared filter results such as TA, TDX, and ETHDR give sharp edges of the magnetized sources (hidden objects). Quantitatively, the determined depths have been evaluated, using the 2D modeling algorithm. The depths and relief of the basement surface across eight profiles are defined, and the depths and shapes of the hidden objects are outlined. Meanwhile, magnetic susceptibility contrasts of the hidden objects are varied with different expanding shapes and subsurface depths.

Retinal blood vessels segmentation using classical edge detection filters and the neural network

Abstract Retinal blood vessels analysis is of interest for medical screening, especially in the diagnosis of diabetic retinopathy. In this paper, we propose a new method for the segmentation of blood vessels in retinal photographs. This method is based on classical edge detection filters and artificial neural networks. Firstly, edge detection filters are applied to extract the features vector. The resulting features are used to train an artificial neural network in order to recognize each pixel as belonging to blood vessels or not. The obtained algorithm is evaluated with the publicly available DRIVE, CHASE and STARE datasets, containing retinal images frequently used for this goal. The performance of the proposed system is calculated in terms of detection accuracy, sensitivity, specificity, and the area under the ROC curve. Our model is compared to other vessel segmentation models with encouraging results obtained. The proposed algorithm is a suitable tool for automated retinal image analysis.

PlantifyAI: A Novel Convolutional Neural Network Based Mobile Application for Efficient Crop Disease Detection and Treatment

Crop diseases are a major threat to human food security. Around the world, more than 80% of agricultural production is generated by farmers, and over 50% of their yield is lost due to pests and pathogens, leading to mass disruption in food supply and a large number of hungry people. Identifying a disease correctly is a crucial first step for efficient treatment but remains difficult in many parts of the world due to limited access to agricultural experts and professional infrastructure. The purpose of this research was to create a free, easy-to-use, and widely accessible mobile application that efficiently and accurately, diagnoses 26 diseases of 14 crop species. Furthermore, this application provides treatment steps, common symptoms, and access to recommended curing products for each disease. The real-time crop disease diagnosis is based on a convolutional neural network (CNN) that was trained, validated, and tested on a dataset of 87,860 leaf images split into 38 classes. To design an optimal CNN, 16 different CNNs were designed and tested. MobileNetV2 using the Canny Edge Detection filter was chosen as it had the highest classification accuracy of 95.7% and an F1 score of 96.1. Multi-level testing and data analysis was conducted for this application, it has been verified to be functional in the real world through field testing at local garden centers. This application is a novel and accessible tool for crop disease management and can be deployed as a free service to farmers for ecologically sustainable production, overall increasing food security.

Исследование методов выделения границ в задаче распознавания объектов на изображениях с низкой контрастностью

The article discusses edge detection methods separately and combinations of edge detection filters with antialiasing filters in the task of pattern recognition on images with low contrast. Sobel, Canny, Otsu and thresholding filters are considered as edge detection methods. Median and Gaussian filters are considered as smoothing filters. The performance of the filters is assessed using the peak signal-to-noise ratio (PSNR) and the structural similarity index (SSIM).

Transforming UTE-mDixon MR Abdomen-Pelvis Images Into CT by Jointly Leveraging Prior Knowledge and Partial Supervision.

Computed tomography (CT) provides information for diagnosis, PET attenuation correction (AC), and radiation treatment planning (RTP). Disadvantages of CT include poor soft tissue contrast and exposure to ionizing radiation. While MRI can overcome these disadvantages, it lacks the photon absorption information needed for PET AC and RTP. Thus, an intelligent transformation from MR to CT, i.e., the MR-based synthetic CT generation, is of great interest as it would support PET/MR AC and MR-only RTP. Using an MR pulse sequence that combines ultra-short echo time (UTE) and modified Dixon (mDixon), we propose a novel method for synthetic CT generation jointly leveraging prior knowledge as well as partial supervision (SCT-PK-PS for short) on large-field-of-view images that span abdomen and pelvis. Two key machine learning techniques, i.e., the knowledge-leveraged transfer fuzzy c-means (KL-TFCM) and the Laplacian support vector machine (LapSVM), are used in SCT-PK-PS. The significance of our effort is threefold: 1) Using the prior knowledge-referenced KL-TFCM clustering, SCT-PK-PS is able to group the feature data of MR images into five initial clusters of fat, soft tissue, air, bone, and bone marrow. Via these initial partitions, clusters needing to be refined are observed and for each of them a few additionally labeled examples are given as the partial supervision for the subsequent semi-supervised classification using LapSVM; 2) Partial supervision is usually insufficient for conventional algorithms to learn the insightful classifier. Instead, exploiting not only the given supervision but also the manifold structure embedded primarily in numerous unlabeled data, LapSVM is capable of training multiple desired tissue-recognizers; 3) Benefiting from the joint use of KL-TFCM and LapSVM, and assisted by the edge detector filter based feature extraction, the proposed SCT-PK-PS method features good recognition accuracy of tissue types, which ultimately facilitates the good transformation from MR images to CT images of the abdomen-pelvis. Applying the method on twenty subjects' feature data of UTE-mDixon MR images, the average score of the mean absolute prediction deviation (MAPD) of all subjects is 140.72 ± 30.60 HU which is statistically significantly better than the 241.36 ± 21.79 HU obtained using the all-water method, the 262.77 ± 42.22 HU obtained using the four-cluster-partitioning (FCP, i.e., external-air, internal-air, fat, and soft tissue) method, and the 197.05 ± 76.53 HU obtained via the conventional SVM method. These results demonstrate the effectiveness of our method for the intelligent transformation from MR to CT on the body section of abdomen-pelvis.

A new high resolution filter for source edge detection of potential field data

Determining the source edges is a frequently requested task in the analysis of potential fields. However, the edge detection methods have some drawbacks or shortcomings, for example, blurred responses to weak amplitude anomalies, the estimated peaks over the edges are low in gradient or bring out overestimated edges in the resulting maps. In this paper, a new method based on an arcsine function using the ratio of the vertical gradient to the total gradient of the amplitude of the horizontal gradient field is introduced for outlining the source edges. The validity of the method is tested on both synthetic and real data and the edge detection results are compared with those detected from some other existing methods. The results from the proposed filter provide more accurate and higher resolution edges, and can avoid producing the spurious edges in the results.

Determination of the Grain Size in Single-Phase Materials by Edge Detection and Concatenation

This paper presents a novel approach for edge detection and concatenation. It applies the proposed method on a set of optical microscopy images of aluminium alloy Al 99.5% (ENAW1050A) samples with different grain size values. The performance of the proposed approach is evaluated based on the intercept method and compared with the manual grain size determination method. Edge detection filters have proven inefficient in grain boundaries’ detection of the presented microscopy images. To some extent only the Canny edge-detection filter was able to compute grain boundaries of lower-resolution images adequately, while the presented method proved to be superior, especially in high-resolution images. The proposed method has proven its applicability, and it implies higher automatisation and lower processing times compared to manual optical microscopy image processing.

Graphic map load comparison of two Czech school atlases using edge detection

Abstract. Map complexity has been studied by cartographers for decades. Unfortunately, it is not being systematically measured as an essential map property and used to optimize cartographic products during the map-making process. Therefore, the appropriate load of map content in a map usually remains dependent only on the subjective assessment of the cartographer designing the map.The paper aims to describe and examine newly developed easy-to-use metric for graphic map load evaluation. A plugin for an open-source software Gimp was developed to provide an accessible tool to measure graphic map load of various geospatial visualizations in an image file format. This tool was then used to compare two Czech school atlases produced by Kartografie PRAHA, a.s. publisher in terms of their graphic map load. The examined atlases include: the latest World School Atlas (in Czech Školní atlas světa) and the new equivalent called Pupil’s Atlas (originally Školní atlas světa) intended to fit especially to younger pupils. Both the atlases are captured in Figure 1.The equivalent maps (the same or very similar map extents, scales, captured regions and map types) in both atlases were compared to investigate differences in graphic map load. For this, a developed metric implemented in the form of an extension to well-known image processing software Gimp was used. The extension advances an edge detection filter to identify both sharp and smooth edges in order to estimate the graphic map load of map samples. Also, a distribution of map load across the map layout is shown. Subsequently, measured values were compared and used to estimate the differences in the potential readability of those products.

A Method for Extracting Some Key Terrain Features from Shaded Relief of Digital Terrain Models

Detection of terrain features (ridges, spurs, cliffs, and peaks) is a basic research topic in digital elevation model (DEM) analysis and is essential for learning about factors that influence terrain surfaces, such as geologic structures and geomorphologic processes. Detection of terrain features based on general geomorphometry is challenging and has a high degree of uncertainty, mostly due to a variety of controlling factors on surface evolution in different regions. Currently, there are different computational techniques for obtaining detailed information about terrain features using DEM analysis. One of the most common techniques is numerically identifying or classifying terrain elements where regional topologies of the land surface are constructed by using DEMs or by combining derivatives of DEM. The main drawbacks of these techniques are that they cannot differentiate between ridges, spurs, and cliffs, or result in a high degree of false positives when detecting spur lines. In this paper, we propose a new method for automatically detecting terrain features such as ridges, spurs, cliffs, and peaks, using shaded relief by controlling altitude and azimuth of illumination sources on both smooth and rough surfaces. In our proposed method, we use edge detection filters based on azimuth angle on shaded relief to identify specific terrain features. Results show that the proposed method performs similar to or in some cases better (when detecting spurs than current terrain features detection methods, such as geomorphon, curvature, and probabilistic methods.

Edge detection for weed recognition in lawns

The rapid propagation of weeds is a major issue for turfgrass management (both ornamental and sports turf). While pesticides can ensure weed eradication, they pose a risk to human health and the environment. In this context, the early detection of weeds can allow a dramatic reduction in the amount of pesticide required. Here we present the use of edge detection techniques to identify the presence of these invasive plants in ornamental lawns and sports turf. Regarding the former, images from small experimental plots in the facilities of IMIDRA were used while images for the latter were taken on a golf course. Up to 12 different filters for edge detection were tested on the images collected. Aggregation techniques, with a range of cell values, were applied to the results of the three most effective filters (sharpening (I), sharpening (II), and Laplacian) to minimise the number of false positives. After the tests with different cell sizes, two filters were selected for more in-depth analysis. Box plots were selected to define the best cell size and identify the filter with the best performance. The sharpening (I) filter and the aggregation technique with the minimum value and a cell size of 10 offered the best results. Finally, we determined the most appropriate threshold value on the basis of the number of false positives, false negatives, and derived indexes (Precision, Recall, and F1-Score). A threshold of 78 gave the best performance. The results achieved with this methodology differed slightly between ornamental and sports turf.

Determining minimum number and placement of fault detectors in transmission network for fault location observability

This paper presents a new minimal placement strategy of fault detectors for fault-location on the transmission network. The proposed methodology includes two offline and online algorithms. First, the topology of a transmission network is analyzed, and the proposed offline algorithm is employed to determine the minimum number and location of detectors. Based on the arrival time of initial traveling wave-front acquired by the installed detectors, the online algorithm is introduced to identify the fault occurrence time as well as fault location. In order to extract the traveling wave from post-fault signals and construct the online algorithm, morphological edge detection (MED) filter is adopted. Extensive simulation studies based on the electromagnetic transients program and MATLAB Software have been carried out to evaluate the accuracy and reliability of the proposed method.

Mapping shoreline change using machine learning: a case study from the eastern Indian coast

The continuous shift of shoreline boundaries due to natural or anthropogenic events has created the necessity to monitor the shoreline boundaries regularly. This study investigates the perspective of implementing artificial intelligence techniques to model and predict the realignment in shoreline along the eastern Indian coast of Orissa (now called Odisha). The modeling consists of analyzing the satellite images and corresponding reanalysis data of the coastline. The satellite images (Landsat imagery) of the Orissa coastline were analyzed using edge detection filters, mainly Sobel and Canny. Sobel and canny filters use edge detection techniques to extract essential information from satellite images. Edge detection reduces the volume of data and filters out worthless information while securing significant structural features of satellite images. The image differencing technique is used to determine the shoreline shift from GIS images (Landsat imagery). The shoreline shift dataset obtained from the GIS image is used together with the metrological dataset extracted from Modern-Era Retrospective analysis for Research and Applications, Version 2, and tide and wave parameter obtained from the European Centre for Medium-Range Weather Forecast for the period 1985–2015, as input parameter in machine learning (ML) algorithms to predict the shoreline shift. Artificial neural network (ANN), k-nearest neighbors (KNN), and support vector machine (SVM) algorithm are used as a ML model in the present study. The ML model contains weights that are multiplied with relevant inputs/features to obtain a better prediction. The analysis shows wind speed and wave height are the most prominent features in shoreline shift prediction. The model’s performance was compared, and the observed result suggests that the ANN model outperforms the KNN and SVM model with an accuracy of 86.2%.

Automatic Lineament Extraction Method in Mineral Exploration Using CANNY Algorithm and Hough Transform

Copper mineralization in Kahang porphyry mining district as a tectonically active region in the center of Iran, at the middle of Urmia–Dokhtar magmatic assemblage (UDMA), has been drastically controlled by structural lineaments. Determination of these footprint features is a good guide to identify the location of ore occurrences. One way to recognize lineaments is to process satellite imagery along with geophysical data. Although the visual extraction of lineaments from such data set is the most common method, automatic methods for detecting lineaments can highly reduce user errors and runtime. The most efficient automated methods in this regard are those that simultaneously take edge detection filters with lineament extraction algorithms into consideration. In this work, the CANNY algorithm was employed as an edge-detector filter at first and later Hough transform was used to extract linear features from satellite imagery and geophysical magnetometry data. The proposed methods were implemented on the high-resolution imagery data collected by QuickBird satellite along with ground-based magnetometry data to extract the shallow and deep-seated lineaments. After investigating and plotting this structural controller in a map, the dominant orientation was in the NE–SW direction perpendicular to the UDMA. Generated lineament density map also indicated that the eastern, southeast, and western portions of the area had high potential for porphyry copper mineralization.

Edge detection, depth estimation and 3D-inverse modelling of the Red Sea and Zagros gravity anomalies using the gravity data extracted from EGM2008 Geo-potential Model

Using geophysical methods and measuring physical properties of subsurface rocks are good solutions for investigating the subsurface structures and exploring underground buried resources (such as oil, gas, water, minerals, etc.). This research investigates the anomaly sources of Zagros and the Red Sea by using the derivative filters, regularized filters, analytic signal, local-phase filter, 3D-inverse modelling with the Li-Oldenburg method. For this purpose, these filters are first applied to artificial models to determine the capability of each of these filters, a comparisons is also will be made between edge detection filters and finally applied to the real gravity of Zagros and Red Sea regions (taken from the EGM2008 Global Model). The overall result is that the effective depth of the sources of gravity anomalies of the Red Sea is approximately 200 km, and incoherently, up to a depth of 300 km. The effective depth of the Zagros anomalies sources is also about 180 km and since then it has continued inconsistently up to 400 km.

GLoG: Laplacian of Gaussian for Spatial Pattern Detection in Spatio-Temporal Data.

Boundary detection has long been a fundamental tool for image processing and computer vision, supporting the analysis of static and time-varying data. In this work, we built upon the theory of Graph Signal Processing to propose a novel boundary detection filter in the context of graphs, having as main application scenario the visual analysis of spatio-temporal data. More specifically, we propose the equivalent for graphs of the so-called Laplacian of Gaussian edge detection filter, which is widely used in image processing. The proposed filter is able to reveal interesting spatial patterns while still enabling the definition of entropy of time slices. The entropy reveals the degree of randomness of a time slice, helping users to identify expected and unexpected phenomena over time. The effectiveness of our approach appears in applications involving synthetic and real data sets, which show that the proposed methodology is able to uncover interesting spatial and temporal phenomena. The provided examples and case studies make clear the usefulness of our approach as a mechanism to support visual analytic tasks involving spatio-temporal data.

An improved image steganography technique using edge based data hiding in DCT domain

Steganography is a type of secret communication which is used to hide the confidential information from the intruder. The security during information communication is a key concern which makes image steganography a popular technique. Image steganography techniques may be broadly classified into two domains i.e., spatial domain and frequency domain. This paper proposes an improved image steganography technique in which data is embedded in the edge pixels of the carrier image. An intruder has less suspicion about the existence of data bits in edges, because edge pixels appear to be different than their neighbours and thus ensures better security. The proposed technique uses different types of edge detection filters like Prewitt, Sobel, Laplacian and Canny in the existing image steganography using edge based data hiding in DCT domain algorithm. Furthermore, the proposed technique results in excellent performance with PSNR of 99% and SNR of 96%. However, the proposed steganography technique surprisingly results in reduction of image size mainly because of image compression.

Type-2 fuzzy control for line following using line detection images

This work presents a comparative analysis of Type-1 and Type-2 fuzzy controllers to drive an omnidirectional mobile robot in line-following tasks using line detection images. Image processing uses a Prewitt filter for edge detection and determines the error from the line location. The control systems are tested using four different paths from the Robotino® SIM simulator. Also, two different strategies in the design and implementation of the controllers are presented. In the first one, a PD controller scheme is extended by using a fuzzy system to have adaptive parameters P and D, additionally, Type-2 Fuzzy sets are used to give robustness to the controller. In the second case, a Fuzzy controller is designed to compute in a direct way the control variables and it is extended to Type-2 Fuzzy controller. Finally, experimental results and comparative analysis are presented for the five control schemes by comparing the running time and the standard deviation to measure the robustness of the control systems.

Image colour edge detection using hypercomplex convolution

Quaternions are considered for colour image edge detection. Most work on quaternions is based on a linear quaternion system (LQS) which applies multi-directional kernels (horizontal, vertical, and diagonal) using hypercomplex convolution, each kernel producing an edge map for a specific direction, and the final result is a combination of these maps. This paper introduces a new colour image edge detection filter based on LQS convolution. The process starts by applying quaternion convolution with the proposed filter, and then generating the final edge map by computing the magnitude of the result. The proposed filter is able to highlight both colour and greyscale edges in multiple directions using a single LQS convolution pass. The validity of the proposed filter is demonstrated, and its performance is supported experimentally through a set of comparisons with state-of-the-art methods.

Optimized Edge Detection Technique for Brain Tumor Detection in MR Images

Genetic algorithms (GAs) are intended to look for the optimum solution by eliminating the gene strings with the worst fitness. Hence, this paper proposes an optimized edge detection technique based on a genetic algorithm. A training dataset that consists of simple images and their corresponding optimal edge features is employed to obtain the optimum filter coefficients along with the optimum thresholding algorithm. Qualitative and quantitative performance analyses are investigated based on several well-known metrics. The performance of the proposed genetic algorithm-based cost minimization technique is compared to the classical edge detection techniques, fractional-order edge detection filters, and threshold-optimized fractional-order filters. As an application for the proposed algorithm, a strategy to detect the edges of the brain tumour from a patient’s MR scan image of the brain is proposed. First, Balance Contrast Enhancement Technique (BCET) is applied to improve the image features to provide better characteristics of medical images. Then the proposed GA edge detection method is employed, with the appropriate training dataset, to detect the fine edges. A comparative analysis is performed on the number of MR scan images as well. The study indicates that the proposed GA edge detection method performs well compared to both classical and fractional-order edge detection methods.