Similarity measures of sectional contour based surface feature extraction from point clouds

Surface feature extraction from point clouds is an important technology of 3D digital geometric signal understanding. The existing surface feature extraction methods have limit of precision while segmenting point clouds from complex surfaces with open contours, branching and blending features. A new practical method is presented for surface feature extraction based on similarity measures of sectional contour. The shape description of arc length and rotation angle is discussed. The similarity of feature is determined according to the normalized cross correlation coefficient of sectional contour feature point clouds. Branch, blend, dissimilarity and precise feature point are distinguished based on the similarity measures of sectional contour feature point. Surface feature is automatically extraction by feature points of point clouds. Experiments show that the proposed surface extraction method can accurately segment the complex surface with open contour, branching and blending feature. Complex surface is segmented into individual surfaces according to similarity measure rules, reflecting the original design intent.