Affiliation of Author(s):计算机科学与技术学院/人工智能学院/软件学院
Journal:Trans. Nanjing Univ. Aero. Astro.
Abstract:The car surface scratch detection adopts the traditional manual detection with poor efficiency and high missing rate. Because of the gray mark and the background of car surface scratches, the traditional edge detection algorithm cannot meet the needs of car surface scratch detection. Therefore, the directional SUSAN algorithm based on CIELab color space is adopted in this paper. The direction template and the circle template to calculate the color difference of the color image are used in the algorithm which has been converted to the CIELab space. The edges and scratches are eliminated by matching and contrasting the detected image with the edge template. Experimental results show that the algorithm can effectively detect scratches on the surface of cars, improve the detection speed and reduce the undetected rate. © 2019, Editorial Department of Transactions of NUAA. All right reserved.
ISSN No.:1005-1120
Translation or Not:no
Date of Publication:2019-06-01
Co-author:王振,XT20872
Correspondence Author:xq