【摘要】 本文针对三维物体识别中的距离图象分割问题，提出了一个新的边与区域混合的鲁棒算法．该算法包括两个步骤：首先使用一种自适应加权拟合技术对原始图象进行平滑以获取边与面的初始估计；然后是一个基于边与区域混合的细分割过程以获得边与面的精确描述．场景实验以及与已有算法进行的比较实验充分证明，该技术明显优于文献中已有的方法，具有良好的适用性．更多还原

【Abstract】 This paper presents a new robust algorithm for range image segmentation based on the integration of region and edge information. An adaptive weighted fitting technique called AWF is developed to smooth the image and obtain the rough estimation Of edges and surface patches. AWF can preserve the depth and normal discontinuities precisely as well as remove local noises. Then with integration of region and edge information the rough segmentation is refined to achieve accurate final result. Comparisons demonstrate that performance of the proposed algorithm is superior over classical ones in its simplicity,accuracy and robustness.更多还原