基于最优邻域的区域填充算法

A region filling algorithm based on optimal neighborhood relativity

为了提高对感兴趣区域识别和填充的效率,提出一种基于最优邻域的区域填充算法。该算法结合优化的邻域判断次序来确定当前扫描位置的临时标记值,提高算法的扫描效率;并利用一维动态数组保存临时标记与最终标记的等价关系,通过集合来记录临时标记之间的等价关系,最后采用集合运算为表示同一连通域的所有临时标记分配最终标记。实验结果表明,该算法可快速填充图像上复杂形状的连通区域,具有较好的鲁棒性。

In order to improve the efficiency of recognition and filling for the region of interest, a region filling algorithm based on optimal neighborhood is presented. The temporary label of the object pixels is firstly determined by the neighborhood in the prior judgment order, which significantly accelerates the scan speed. Then a one - dimension array is used to store the equivalent relationship of the temporary la- bel and final label, and sets are utilized to record the equivalent relationship among the temporary labels. Finally, set operations merge all temporary labels representing the same connected component with a final label. Experimental results show this optimized algorithm can effectively and stably fill connected domains with complex shapes.