Analysis of Neural Networks for Edge Detection

2014

Neural network edge detection is associate rising field that consists of complementary parts of fuzzy logic, neural computing and biological process computation. Neural network edge detection techniques have found wide applications. One among the foremost necessary applications is edge detection for image segmentation. The method of partitioning a digital image into multiple regions or sets of pixels is termed image segmentation. Edge could be a boundary between 2 consistent regions. Edge detection refers to the method of characteristic and locating sharp discontinuities in an image. During this paper, the main aim is to survey the idea of edge detection for image segmentation exploitation neural network supported the fuzzy logic, Genetic algorithmic rule and Neural Network. Similar to an individual's observer, an automatic image vision system is ready to recognize most components of associate object if the system may accurately trace and reflect its true form. This has prompted...

2013 International Conference on Electrical Engineering and Software Applications, 2013

In this work a new method for image edge detection based on multilayer perceptron (MLP) is proposed. The method is based on updating a MLP to learn a set of contours drawn on a 3x3 grid and then take advantage of the network generalization capacity to detect different edge details even for very noisy images. The method is applied first to Gray scale images and can be easily extended to color ones. Simulations on synthetic and real image show much promised results in term of precision and localization. Moreover the method works well even for very low contrast images for which other edge operators fail.

This study employed the use of Cellular Neural Networks (CNN) for Edge Detection in images due to its high operational speed. The process of edge detection is unavoidable in many image processing tasks such as obstacle detection and satellite picture processing .The conventional edge detector models such as Sobel Operator, Robert Cross, among which Canny is the best, have high computational time. The CNN Model is a class of Differential Equation that has been known to have many application areas and high operational speed. The work investigated four parameters: resolutions, processing time, false alarm rate, and usability for performance evaluation. The CNN Model was modified by using hyperbolic tangent (tanh x) and Von Neumann Neighborhood. The modified CNN Model and enhanced Canny Model were implemented using MATLAB 7.0 running on Pentium III and 128 MB RAM Personal Computer. A series of images served as input for both Canny and modified CNN Model. With several images tested, the overall results indicated that the two models have similar resolutions with average computational time of 1.1078 seconds and 2.293 seconds for CNN-based and Enhanced Canny Model respectively. The hyperbolic entry a 22 of the cloning template A made our work fully controllable since max(tanh x) = +1 and min(tanh x) = -1. A consideration of the set of digital images showed that edge maps which result from Canny Model have adjacent boundaries that tend to merge. The CNN model obviously produced the optimum edge map with edges that are one-pixel wide and unbroken. The false alarm rate of noise variance probability     2  F P for which an edge detector can easily declare an edge pixel given that there is no edge, showed the value 0.8525 for CNN Model and 0.4595 for Canny Model. The CNN parameters can be adjusted and modeled to solve Partial Differential Equations and Maximum likelihood problems for edge detection while Canny Model cannot be easily adjusted for any other functionality. The CNN-based edge detector performed better than the popular canny operator in terms of the computational time required, usability, and false alarm rate.

Frontiers in Computational Neuroscience, 2019

Insects can detect the presence of discrete objects in their visual fields based on a range of differences in spatiotemporal characteristics between the images of object and background. This includes but is not limited to relative motion. Evidence suggests that edge detection is an integral part of this capability, and this study examines the ability of a bio-inspired processing model to detect the presence of boundaries between two regions of a one-dimensional visual field, based on general differences in image dynamics. The model consists of two parts. The first is an early vision module inspired by insect visual processing, which implements adaptive photoreception, ON and OFF channels with transient and sustained characteristics, and delayed and undelayed signal paths. This is replicated for a number of photoreceptors in a small linear array. It is followed by an artificial neural network trained to discriminate the presence vs. absence of an edge based on the array output signals. Input data are derived from natural imagery and feature both static and moving edges between regions with moving texture, flickering texture, and static patterns in all possible combinations. The model can discriminate the presence of edges, stationary or moving, at rates far higher than chance. The resources required (numbers of neurons and visual signals) are realistic relative to those available in the insect second optic ganglion, where the bulk of such processing would be likely to take place.

This paper introduces a proposed method based on a backpropagation artificial neural network using Scaled Conjugate Gradient (SCG) training algorithm so as to gain the edges of any image. A new training image model is suggested to train this artificial neural network, then using this network to find the edges of any image. Computer experiments are carried out for extracting edge information from real images; the results presented are compared with those from classical edge detection methods like Canny. Using this new method does not need to tune any parameter to find the edge of any image, as well as using this method the false edges is reduced.

2000

In this paper, we investigate the feasibility of characterizing signi"cant image features using model-based neural network with modular architecture. Instead of employing traditional mathematical models for characterization, we ask human users to select what they regard as signi"cant features on an image, and then incorporate these selected features directly as training examples for the network. As a "rst step, we consider the problem of the characterization of edges, which are usually regarded as signi"cant image features by humans. Unlike conventional edge detection schemes where decision thresholds have to be speci"ed, the current NN-based edge characterization scheme implicitly represents these decision parameters in the form of network weights which are updated during the training process, and which thus allow automatic generation of the "nal binary edge map without further parameter adjustments. Experiments have con"rmed that the resulting network is capable of generalizing this previously acquired knowledge to identify important edges in images not included in the training set. In particular, one of the important attributes characterizing the current approach is its robustness against noise contaminations: the network can be directly applied to noisy images without any re-training and alteration of architecture, as opposed to conventional edge detection algorithms where re-adjustment of the threshold parameters are usually required.

International Journal of Computer Applications, 2013

The edges provide important visual information since they correspond to major physical and geometrical variations in scene object. Edge detection is a terminology in image processing that refers to algorithms which aim at identifying edges in an image. In this paper a Feedforward Neural Network (FNN) based algorithm is proposed to detect edges in gray scale images. The backpropagation learning algorithm is used to minimize the error. Standard deviation and gradient values are used as training patterns. In the end the network is tested for a number of different kinds of grayscale images. The proposed scheme is compared with Prewitt, Roberts, Sobel, LoG and other neural network based method in which binary training patterns are used. Our method has performed significantly better as compared to other methods.

This paper depicts three methods for edge detection. The first method is one of the promising method for edge detection based on canny edge detection. In the second method neural network has been used for edge detection. Third method is inspired by the behavior of biological receptive fields and the human visual system. A network model based on spiking neuron is proposed to detect edges in visual image. SNN is able to perform edge detection within a processing time which is consistent with human visual system. SNNs improve the representation capacity and the processing abilities of neural networks. Successful computer simulation results are obtained.

Journal of Image and Graphics, 2013

In image processing based applications there are very important requirements of noise removal and edge detection. In this paper universal approximation characteristic of feedforward neural is taken to achieve both of these requirements in a simple but very efficient way. Concept of local pattern generated by small region pixels which define the possibility of relation among pixels is presented. This approach facilitates the solution as universal solution for different types of noise removal compare to conventional solutions which are based on noise characteristics. Same model of neural network with little extension can also be utilized as edge detector has also presented. Another benefit of proposed model is contrast enhancement without any extra computation cost. In effect this solution can be considered as universal solution for noise reduction, edge detection and contrast enhancement. Comparison has made with well established solution like median filter and adaptive Wiener filter for noise reduction where as Canny and Prewitt detectors have taken for edge detection comparison.

Edges of an image are considered a type of crucial information that can be extracted by applying detectors with different methodology. This research paper presents a brief study of the fundamental concepts of the edge detection operation, theories behind different edge detectors, and some simple self-written Matlab edge detection functions with the simulation results. Previous works on edge detection models are reviewed and simulated. The Matlab results coincide with the first and second order derivative edge detection models.