Research on Point Cloud Registration Algorithm for Concealed Gap Profiles

Abstract

Abstract. To achieve precise extraction of fitting area contours of automotive cover parts in automated vehicle quality inspection, particularly when multiple angled shots with a line structure light camera are needed to obtain a complete contour, a point cloud preprocessing scheme suitable for the coherent point drift (CPD) registration algorithm is proposed for the precise registration of concealed gaps. First, an “improved DBSCAN clustering algorithm” is proposed to extract contour curves. Next, a “Bezier equation-based tangent point positioning algorithm” is designed to locate tangent points accurately. Finally, the traditional CPD algorithm is combined to achieve precise point cloud registration. The experimental results show that the proposed clustering algorithm has good application effects; the tangent point positioning algorithm achieves an accuracy of up to 0.03 mm; the root-mean-square error (RMSE) of the registration algorithm can be controlled within 1.24 × 10−3, reducing it by 99.6% compared with the traditional CPD algorithm; and the registration success rate reaches 99%, with significantly improved stability.