The Research Of FSVM And Its Application In Image Edge Detection

The image edge detection is an important part of image processing. The task of finding the edgesin an image that correspond to true physical boundaries remains a hot topic of image processing.However, classical edge detection algorithms always have localization, because of its relativelyfixed template. And they are only applied to the limited ideal situation. With the development ofmachine learning and data mining, Support Vector Machine (SVM) has gradually been used to theimage edge detection. However, the traditional SVM is very sensitive to noise and outliers, it isdifficult to suppress image noise while extracting the edge.For practical engineering applications, a new fuzzy support vector machine (FSVM) is proposedin this paper. And it is applied to the image edge detection. FSVM is an improved SVM. It gives thedifferent membership function to different samples in order to reduce the influence of noise. Theperformance of FSVM can be improved by creating the new membership function. And the FSVMclassification algorithm is proposed to extract the edge of image in this paper.The following innovations in this paper are as follows:(1) A new Fuzzy Support Vector Machine which is applied to the image edge detection isproposed;(2) Constructing a new fuzzy classification membership so that improving the accuracy ofclassification;(3) Using the new FSVM classification algorithm to construct the new model of edgedetection, and applying the model to edge detection of the ideal image and the image of havingnoise. Simulation results show the proposed method gives good performance on reducing the effectsof noise and has good performance of image edge detection.