模式识别国家重点实验室知识库

Three-dimensional (3D) inspection based on machine vision is high-precision and efficient. In this paper, a laser scanning system using the triangulation measurement is designed to obtain dense point clouds of the workpiece surface, and the point clouds are projected to two-dimensional (2D) range images. Fast Directional Chamfer Matching (FDCM) is a reliable algorithm for object detection and localization, and it is improved to accelerate the time-consuming 3D registration. Although the author of FDCM greatly improved the directional chamfer matching, it remains very slow in practice. We mainly improved the line fitting process and distance transform in this algorithm, and they greatly accelerate the scanning process. Experimental results show that the scanning system with the improved algorithm can highlight the deformation of the workpiece in real time.