摘要
本文针对医学图象的特点 ,改进了医学图像的分水岭算法 ,并用于医学图像的分割处理。这种分割通常应用分水岭算法 ,但是它有过分割的严重问题。本文闸述了在分水岭算法的基础上做的一些改进 ,其内容是 :在分水岭算法之前 ,引入浮点 ,活动图像作为分水岭算法的输入 ,在分水岭算法之后 ,在面积控制的基础上 ,同时根据面积控制和对比度控制的准则 ,将某些被分割的小区并入邻近较大的区域。这种改进的方法使过分割现象得到了很好的抑制 ,而且医学图像中的病变小区被分割出来了 ,效果很好。
Considered the uniqueness of medical image the watershed algorithm is improved and applied to the segmentation of the medical image. Watershed algorithm is commonly applied to the segmentation, but there is a sever problem of over-segmentation. This paper is to expound some improvements based on the current base of watershed algorithm. These improvements are: before watershed algorithm, to introduce a floating-point active-image as the input of watershed algorithm and after watershed algorithm, to incorporate some small segmented regions into the near bigger region based on the area control, and at the same time according to the incorporated criterions of area and contrast control. By means of the improved method, the phenomena of over-segmentation are well restrained, and small regions with pathological change are well segmented. The effect of this method is very well.
出处
《微电子技术》
2003年第4期39-42,共4页
Microelectronic Technology
关键词
分水岭算法
浮点活动图象
形态梯度
Watershed algorithm
Floating-point active-image
Morphological grade
