Three-dimensional point cloud registration by matching surface features with relaxation labeling method

摘要: Automated approaches for the conversion of multiple overlapped three-dimensional (3D) point clouds into an integrated surface shape measurement in form a complete polygon are important general field reverse engineering. Traditionally, process is achieved semi-automated manner that requires extensive user interaction. In this work, automated methods set registration developed and experimentally validated using reconstruction to represent raw, 3D obtained from non-contacting systems. Using local differential properties extracted representation data set, robust sculpture feature-matching method described automatically obtaining initial orientation mismatch estimates each set. both simulated measured experimental quantify performance method, it shown features appropriate metrics identifying common initializing relative positions individual clouds, thereby providing basis automating integration processes while improving speed distance minimization process.