Strip Image Edge Detection Based on Ant Colony Algorithm

摘要：

A novel method based on Ant colony algorithm (ACA) is proposed for detecting edges in strip image according to the fuzzy clustering ability of ACA. In this paper, we employ two feature-extraction elements, namely wavelet entropy and multiscale wavelet transform module gradient value to compose the characteristic vector of each pixel. To avoid the early convergence or stability in basic ACA, the evaporating coefficient is set to change dynamically with the number of valid paths the ants passed. Finally, we use ACA to cluster edges. Experimental results demonstrate the proposed algorithm possesses high convergence speed, low computing complexity and good noise restraining performance. In addition, the proposed algorithm also overcomes the randomicity of the searching process in basic ACA.

关键词： ant colony algorithm strip image characteristic vector edges

作者: LIU Yue HU Shaoxing

作者单位: Laboratory of 3D Laser scan & Industry Computer Tomography Beihang University School of Mechanical Engineering & Automation, Beihang University, Beijing, 100083

会议类型: 国际会议

会议名称: 第七届国际测试技术研讨会

会议地点: 北京

会议语种:英文

在线出版日期: 2007-08-05（万方平台首次上网日期，不代表论文的发表时间）