A Study On Algorithm Of Edge Detection Based On The Eight Directions Sobel Operator

With the wide application of digital image, the requirements for image accuracy havegradually increased. Edge is the boundary of the target image and the background image, aswell as one of the most basic features of the image. Edge contains a wealth of intrinsic imageinformation, such as the direction, nature of step and shape, etc. Therefore, the edge detectionis a critical subject in the field of image analysis and identification. As a pre-treatment process,edge detection is widely used in the field of image segmentation, pattern recognition andmotion analysis, etc. Traditional edge detection method mainly uses spatial convolutionalgorithm, to extract the edge by convoluting the template and the image. Due to itsadvantages, such as universality and rapidness in calculation, it has been widely used.However, with the continuous improvement of people’s living standards, the demand forcolor image process is growing. Experts from different circles have paid more attention to theresearch of color image. Till now, most of edge detection algorithm of the color image isderived and developed from the edge detection algorithm of gray-scale image. However, thecolor image contains not only information of brightness, but also the information ofchrominance that the gray-scale image does not possess. Thus, the grayscale image edgedetection theory is not suitable for color images. Research indicates that the color imagesprovide more information of edge than that of the grayscale image. Therefore, edge detectionof color image has greater practical significance.In this thesis, the background information of image edge detection and research status athome and abroad have been described in the beginning. By understanding the profile of theimage edge detection research, we expand the research of this article. We introduce somebasic concepts of edge detection and discuss the research methods and the characteristics ofthe traditional gray image edge detection. On this basis, in order to tackle the problem ofineffectiveness exists in the traditional edge detection method; we propose an algorithm ofedge detection based on the eight directions Sobel operator. Inspired by the above method, weimprove the traditional Prewitt edge detection algorithm and give a comparative analysis of the experimental results. Secondly, we have further studied the color image edge detection.The expression form of color image is different in different color space. Based on color spaceconversion technologies, a color image edge detection algorithm based on CIELAB space isproposed. And the true edge can be detected under the premise of not losing the informationof color image. Thus, it makes meaningful attempts on a wider application of image edgedetection. This research mainly focuses on the following aspects:(1) For the questions in the traditional edge detection method, this paper proposes aalgorithm of edge detection based on the eight directions sobel operator. Combined withmulti-directional edge, the directions of the traditional edge detection algorithm is extended toeight directions. It can better detect the edge in different directions by using0o、22.5o、45o、67.5o、90o、112.5o、135o、157.5oeight directions template.(2) We take into account that the distance of neighborhood pixel to the center pixel isdifferent, the contribution of the neighborhood pixel to the center pixel is different. aalgorithm of edge detection based on the eight directions sobel operator weights theneighborhood pixel according to the Euclidean of it to the center pixel. We let the pixel who isnearer to the center pixel has greater weight. And derive a weighted formula digital images.(3) The traditional Prewitt edge detection algorithm was improved accordingly, and animproved Prewitt operator edge detection algorithm is introduced. Then we gives acomparative analysis of the experimental results.(4) We propose a color image edge detection algorithm based on CIELAB space. Usingthe color space conversion technology, the algorithm transformed from an RGB color spaceinto the CIELAB color space uniformly. We used the difference between two color pixels withCIELAB color space to define eight direction detection template. The above grayscale imageedge detection method is extended to the field of color images. Simulation results show thatthe algorithm can detect more accurate edge.