为了解决传统分水岭算法的过分割问题，提出一种使用形态学梯度重构和标记提取技术进行图像预处理的分水岭图像分割方法。该方法基于多尺度概念，进行梯度重构时采用了不同尺寸的结构元素，在对重构后的各梯度图像的区域极小值进行标记后，将各标记点的并集作为最终标记图像，用其修改梯度图像，然后进行分水岭变换，实现图像的区域分割。实验结果表明，该方法既能有效解决分水岭算法的过分割问题，又保留了各尺度下的重要目标，并且可以根据图像特点和具体的分割要求，调整分割过程中所选参数，得到满意的图像分割效果。

A watershed segmentation method combining multiscale morphological gradient reconstruction with marker extraction is proposed. Considering the idea of multiscale, this method employs different sizes of structure elements to reconstruct morphological gradient image and extracts markers of regional minima from each gradient image by using thresholds. The union set of a series of marker images is regarded as the final marker image. Then the markers are used to modify morphological gradient image. Finally, the watershed transformation of the marker modified gradient image is performed to achieve the regional segmentation of the image. Experimental results show that this method can effectively avoid over segmentation of watershed algorithm and maintain important objects at different scales. Furthermore, better segmentation results can be achieved by adjusting the chosen parameters during segmentation process according to the features of the image and specific requirements.