关键词: 点云, 优化, 曲率分布, 误差分布, 逼近曲线

Abstract: To improve accuracy of shape information so as to optimize the approximate curve of point-clouds, an optimization algorithm was proposed based on curvature distribution and error distribution. Firstly, a non-optimal B-spline curve within the error limit was rapidly established to obtain more accurate curvature distribution. And then, the corner-points which were greater than the curvature threshold were chosen as the seed points. The cubic spline approximate curve with sequence number parameters was established by this seed points. If the error of approximate curve was greater than the error limit, the added position of a new dominant point was determined based on the error distribution so as to decrease the maximum error after adding every time. Experiments showed that the control points number of approximate curves was remarkably decreased.

Key words: point-clouds, optimization, curvature distribution, error distribution, approximate curves, control points