Image Edge Detection Based On Kirsch Operator

Edge detection is the most critical and most basic problem, and also is one of technicalproblems in the field of digital image processing. The results of edge detection have a majorimpact on the process of feature extraction, feature description, target recognition and imageunderstanding. Therefore, edge detection has an important position which is based on thetechnology of image edge, such as computer vision, pattern recognition, imagesegmentation, and many other applications.This paper researches the process of image edge detection technology, especially theclassical edge detection methods. The simulation results of the methods with noise or notare compared, then the evaluation standards of the edge detection are briefly reviewed inthis paper. The multi-scale algorithm is studies due to the existing algorithm for adaptiveproblem of different detection scales, and applies the wavelet multi-scale algorithm insmoothing filter of image edge protection. The article focuses on Kirsch operator in edgedetection, and makes use of loop displacement in eight direction template and matrixmanipulation in order to reduce the amount of calculation. Finally, the local entropy edgedetection method is applied to gray image binarization method.Compared with traditional median filter, the experiments verify that the wavelet edgesmoothing filter is better to protect image edge, and also verify the validity of the method oflocal entropy threshold value. The improved algorithm not only can protect the smooth filteredge of image, but also can make binarization have an obvious improvement in image edgedetection.