Research And Implementation Of Parallel Medium Edge Detection Algorithm Based On Multi-core Environment

The edge is the most basic feature of images. Image Edge Detection is a basic research problem in the field of image processing and computer vision. The Image Edge Detection gets more and more extensive applications in the field of modern information processing, especially in Biomedical Science and Industrial Manufacturing. So the researches of Image Edge Detection Algorithm have the very practical meaning. The classic Image Edge Detection Algorithms have high efficiency, but the result of these algorithms are not so satisfactory. The new algorithms such as Medium Edge Detection has good effection, but the execution speed of this algorithm can not matchs people’s expectation. On basis of maintaining the excellent results,how to improve the detecting speed is the focus of the image Edge detection research. With the popularity of multi-core computers and the rapid development of parallel computing technologies, designing parallel algorithms of Image Edge Detection on multi-core platforms to improve detecting speed has important practical value.In this thesis, we review the development of Image Edge Detection Algorithm especially the Medium Edge Detection Algorithm and parallel computing technologies, and introduce the methods of parallel computing based on Matlab. Next we design a parallel Medium Edge Detection Algorithm,mainly by effort of divide the task of computing the DR Function Matrix to design a parallel Medium Edge Detection Algorithm. Then apply the parallel algorithm to Grayscale image and achieve parallel algorithm by using Matlab language, finally analyse the Speedup ratio. To design a parallel Edge Line Detection based on Hough Transform, use the task divide mode. Achieve the parallel algorithm by using Matlab language and analyse the Speedup ratio. At last, based on the last two chapter ’s improved algorithm, this thesis design a Medium Edge Detection System based on multi-core platforms. It can do pretreatment to grayscale image, Parallel Medium Edge Detection and Parallel Edge Line Detection. By analysis,the system’s expandability is satisfying.