文献类型：期刊文献

英文题名：Multi-resolution-based contour corner extraction algorithm for computer vision-based measurement

作者：Li, Jianhua[1]; Liao, Lin[1]

机构：[1] Department of Computer Science and Engineering, East China University of Science and Technology, Shanghai, 200237, China

年份：2018

卷号：1A-2018

外文期刊名：Proceedings of the ASME Design Engineering Technical Conference

收录：EI(收录号：20184706127861)

语种：英文

外文关键词：Design - Extraction

摘要：Corner-based registration of the industry standard contour and the actual product contour is one of the key steps in industrial computer vision-based measurement. However, existing corner extraction algorithms do not achieve satisfactory results in the extraction of the standard contour and the deformed contour of the actual product. This paper proposes a multi-resolution-based contour corner extraction algorithm for computer vision-based measurement. The algorithm first obtains different corners in multiple resolutions, then sums up the weighted corner values, and finally chooses the corner points with the appropriate corner values as the final contour corners. The experimental results show that the proposed algorithm, based on multi-resolution, outperforms the original algorithm in the aspect of the corner matching situation and helps in subsequent product measurements. Copyright ? 2018 ASME.