Performance Evaluation In Edge Detection Algorithms Of Gray Images

Edge detection is one of the most fundamentals in image processing and analyzing, which is still unsolved. Image's edges include image's features such as position and outline, which belong to the fundamental features. Edge detection is widely used in image analysis and processing such as feature description, image segmentation, image enhancement, pattern recognition and image compression etc., so Edge Detection is the research hotspot in the technology of image processing and analysis all the while, on which the new theory and methods are put forward constantly.Some edge detection algorithms are researched, including Roberts, Sobel, Prewitt, LoG, Canny, Kirsch etc., and algorithms which gained more attention, such as Fuzzy Enhancement, Gray Morphology, Statistic, Surface fitting. These algorithms are introduced in detail, and the steps of realized are introduce too. Experiments are taken.Main aim of the paper is studying the Performance Evaluation in edge detection algorithms, and the detection results are compared. Edge detection performances are evaluated by the existing methods of qualitative and quantitative analysis. In the existing methods, Canny criteria, Pratt's figure of merit measurement, and Strickland Edge Equality Score(EQS) are mainly used to judge the features of edges, including continuity, Smoothness, Thinness, Localization, Detection, Noisiness and Time Complexity. On the basis of Pratt's figure of merit measurement, a new evaluation method is brought forward in order to determine Detection and Location. Then images are divided into 3 kinds in the paper. The algorithms which have good performances or fit some special kind of images are chosen by qualitative method.The experiments results show that the algorithms can be a useful reference of latter parts of image processing. It can be a reference for selecting algorithms in Image Segmentation and Image Edge detection.