一种基于对象的连通区域标记方法

A New Algorithm Connected Component Labeling Based on Objects

传统连通区域标记方法主要为二次扫描法和区域增长法,针对目标主要是单个像素。这里将给出一种基于对象探索的连通区域标记方法,该方法无需通过多次扫描即可区分不同连通区。我们的研究对象依然是四连通区域,依靠新的探索方法单独提取出每个对象的边界,为栅格图像矢量化的后续工作打下基础。

The traditional method of connected component labeling is the twice scanning method and the region growingmethod, which are both directed against a single pixel. Present a method of connected region labeling based on object exploration, which can distinguish different components without multiple scans. Our research object is also four--neighborhood connected region, relying on a new exploration method to extract the boundary of each object separately. By this, we can make it easier to vector raster images in our following work.