摘要： 在细胞图象、纳米材料图象等颗粒图象中,经常会有颗粒重叠粘连的现象,这给图象后续的分析处理带来了许多困难。针对这一情况,提出一种有效的分离算法,应用于二值化后的数字图象,先对图象作距离变换,使颗粒群中的每个颗粒在距离图象中都产生一个局部最大值,再利用分水岭算法,从一个初始的高阈值逐渐降低,就能找到边界点,从而将颗粒群分开。

关键词: 颗粒图象, 距离变换, 分水岭算法, 倒角

Abstract: The overlapping phenomenon often appears in such granule images as cell image,labuan material image etc.The phenomenon makes it hard for image consequent processing.In the paper,an effective segementing method is proposed in order to solve the problem.First,distance transform is applied to the thresholded digtal image ,and so the maxvalue according to each granule of the clustering granules is produced as a result,secondly,by use of the watershed algorithm,we mark the edge points and then break apart clusters of granules. ;;;

中图分类号: 

• TP391.4