Laplacian Neural Network based Image Edge Detection Method using Supervised Learning System

Currently, edge detection techniques make extensively attention because of its high applicability and diversity in many human life applications. In order to extract important information, an efficient edge detection method is required. Edges are represented as the boundaries where particular intensity changes or discontinuities happen. Experimentally, discovered that it is hard to design an edge detector that's able of finding all the true edges in an image. Lots of conventional algorithms have been recommended to detect the edges such as Sobel, Prewitt, Roberts, and Laplacian. This paper presents the design of Laplacian Neural Network and convolutional neural network based edge detection methods. The proposed methods first determines the features used as the input to the supervised learning system, features are selected carefully for each input image. Next, Laplacian edge detector was utilized as the training for the proposed Back propagation Artificial Neural Network. The structure of the system is simple and defined by different features for the input images, number of neuron in the hidden layer, and the output layer. The suggested system applied on the tested images and produces subjectively accurate results in term of true\ false edges compared with Sobel, Prewitt, and classical Laplacian detectors.