摘要
分水岭算法由于其分割速度快、精确而受到很大的关注．但它存在过分割的问题。经过研究给出了改进算法，从两个方面来改进：（1）在算法执行前对输入图像进行滤波降噪处理；（2）在算法执行中结合动态合并准则直接对算法本身的形成的过分割区域进行抑制。实验结果表明，该方法能有效地处理过分割现象，是一种行之有效的方法。
Watershed segmentation algorithm becomes the focus because of its fast, accurate and attention, but there is over-segmentation problem. After the study shows the improved algorithm, to improve in two ways： （1）makes filtering noise reduction on the input image before the realizing; （2）in the algorithm with dynamic consolidation process of implementation guidelines for the formation of directly on the algorithm itself, to suppress the over-segmentation regions. The experimental result shows that the method can effectively deal with the phenomenon of over-seg- mentation, is an effective method.
出处
《现代计算机》
2011年第14期28-31,共4页
Modern Computer
关键词
医学图像分割
分水岭算法
数学形态学
动态合并
Medical Image Segmentation
Watershed Algorithm
Mathematical Morphology
Dynamic Merge
