区域生长法，手动选种子点吗，它是一个迭代的过程，这里每个种子像素点都迭代生长
直到处理过每个像素，因此形成了不同的区域，这些区域它们的边界通过闭合的多边形定义。
在区域生长中的主要问题如下:
(1)表示区域的初始化种子的选择:在区域生长过程中，这些不同区域点合适属性的选择。
(2)基于图像具体属性的像素生长不一定是好的分割。在区域生长过程中，不应该使用连通性或邻接信息。
(3)相似性:相似性表示在灰度级中观察在两个空间邻接像素之间或像素集合的平均灰度级间的最小差分，它们将产生不同的区域。如果这个差分比相似度阈值小，则像素属于相同的区域。(Region growth method, manually select the seed point, it is an iterative process, where each seed pixel points are iterative growth
Until each pixel is processed, different regions are formed, and their boundaries are defined by closed polygons.
The main problems in regional growth are as follows:
(1) represents the selection of the initialization seed of the region: in the region growing process, the selection of these different region points is appropriate.
(2) pixel growth based on image specific attributes is not necessarily good segmentation. Connectivity or adjacency information should not be used during regional growth.
(3) similarity: the similarity representation, in gray level, is to observe the minimum difference between the two spatial adjacent pixels or the average gray level of the set of pixels, which produce different regions. If the difference is smaller than the similarity threshold, the pixels belong to the same region.)