Abstract: Binary connected component detection and labeling is an essential step in many image analysis techniques, highly efficient connected component labeling algorithm can greatly speed up image processing. Aiming at this point, a new algorithm based on run-length encoding is presented. It scans the image, records all the run-length codes and appends the equivalent pairs to the equivalent linked list. It analyzes the equivalent linked list by recursive method, obtains the mapping between old labels and the new ones and corrects the old labels obtained. The comparative experiments with several traditional algorithms and two improved methods show that it is more effective.

Key words: binary image, run-length encoding, connected component labeling, recursion

摘要： 二值图像连通区域的检测和标记在图像分析中是十分重要的步骤，高效的连通区域标记算法能大大提高图像处理速度。针对此，提出一种新的基于游程编码的连通体标记算法。扫描图像，记录所有的游程编码并将等价对添加到等价对链表中。通过递归方法对等价对链表进行分析，得到旧标记和新标记之间的映射关系，并修正得到的游程编码标记。与几种传统方法和两组改进方法的对照实验表明该算法是更高效的。

关键词: 二值图像, 游程编码, 连通区域标记, 递归