Research And Implementation On Algorithm Of Image Object Contour Extraction

Image segmentation is a basic operation of the image processing, is animportant content of the image engineering. All fields related to the machinevision use image segmentation. Summary of the existing image segmentationmethods, most algorithms are based on the underlying characteristics of theimage, segments the image into some scattered regions and can not get acomplete and meaningful object. With the ways to acquire images more andmore, the number of images need to deal with has increased dramatically. Howto access the images and extract the useful information in them, is a big anddifficult problem that the image processing need to solve. However, thedisadvantages exiting in the image segmentation, makes it can’t meet theneeds of solving the problem.With this background, the concept of the image object segmentation hasbeen invented and got a high attention and in-depth resrarch from researchers.This paper studies the existing image object segmentation methods, analysistheir performance characteristics, and uses the experience of the idea thatsegmenting object by contour segments matching.This paper examines the most classical edge detection operators, andmakes comparative analysis of their performance by smulating images. Basedon the analysis of the simulation results, the performances of Canny algorithmare superior to other operators, so this article chooses Canny algorithm inedge detection. This article also examines several classic edge thinningalgorithms, studys a frame edge thinning algorithm, and proposes an improvededge thinning algorithm on the basis of the original original. After the aboveedge extraction processing, the edges we got may be unconnected. So thispaper studys an edge growing algorithm and proposes a improved edgegrowing algorithm by solving the existing problems of the original algorithm,and seal the unconnected edges with it to get a complete object contour.Through the above operators, this article uses the adaptive threshold segmentation and area screening, to get the edges of object from complexbackground. At last, this paper builds a interface and realizes the function ofeach steps by program.