出了一种适合医学图像分割的改进分水岭算法，针对分水岭算法存在的过分割问题，采用数学

形态学重构滤波器和Ostu阈值分割方法对图像中感兴趣的目标和背景进行标记。根据标记的二值图像，

运用形态学极小值标定技术对原有梯度图像进行修正。最后，使用分水岭算法对修正的梯度图像进行分割。该方法能有效抑制过分割现象，而且计算复杂程度较低、分割效果较好。-Out of a medical image segmentation for improved watershed algorithm for segmentation of watershed algorithm over the problem, using the mathematical morphology reconstruction filters and Ostu thresholding segmentation interested in the image and the object and the background to tag. Binary image according to the marks, the use of minimum calibration technique morphological gradient image of the original amendment. Finally, the gradient of the modified watershed algorithm to segment the image. This method can effectively suppress the phenomenon of over-segmentation, and low computational complexity, segmentation is better.