基于数学形态学和区域合并的医学CT图像分割*

针对传统分水岭算法分割腹部CT图像存在的过分割情况，提出了一种基于形态学优化和区域合并的分水岭分割算法。该方法先利用多尺度数学形态学方法检测出梯度图像，并用形态学重构去除细密纹理和噪声引起的局部极值，然后进行分水岭变换。为了产生有意义的分割，采用简单的区域灰度均值对变换后的图像进行有效的合并。实验结果表明，该方法能有效解决分水岭算法的过分割问题，得到较好的分割效果。

Medical CT image segmentation based on mathematical morphology and region merging

In order to reduce the over segmentation of the traditional watershed algorithm, the algorithm was modified based on mathematical morphology and region merging. First, obtained gradient image by the multi-scale morphological method. Employed opening operation and closing operation of reconstruction to the gradient image to remove close texture and the extremes caused by noise. Then applied watershed transformation to get the initial segmentation. Furthermore, adopted region merging region to get a significative result, whose criteria was mean value of sub-area. Experimental results show that this method can effectively avoid over-segmentation of watershed algorithm, and better results can be achieved.