分水岭变换是一种非常适用于图象分割的形态算子，然而，基于分水岭变换的图象分割方法，其性能在很大程度上依赖于用来计算待分割图象梯度的算法。为了高效地进行分水岭变换，提出了一种计算图象形态梯度的多尺度算法，从而对阶跃边缘和“模糊”边缘进行了有效的处理，此外，还提出了一种去除因噪声或量化误差造成的局部“谷底”的算法，实验结果表明，图象采用本文算法处理后，再进行分水岭变换，即使不进行区域合并，也能产生有意义的分割，因而极大地减轻了计算负担。

Watershed transformation is a powerful morphological operator for image segmentation.It is performed on the gradient of the image to be segmented.Each minimum of the gradient leads to a region in the resulting segmentation.However,conventional gradient operators generally produce many local minima. Produced by noise and quantization error. Experimental result, which are caused by noise or quantization error.Hence,watershed transformation with a conventional gradient operator usually results in over segmentation.To alleviate this problem,this paper presents a multiscale algorithm for computing morphological gradient images,with effective handling of both step and blurred edges.We also present an algorithm to eliminate the local minima produced by noise and quantization error. Experimental results demonstrate that the proposed algorithm can effectively enhance step and blurred edges and reduce the number of local minima.Watershed transformation with the proposed algorithm produces meaningful segmentations,even without a region merging step.The proposed algorithm can significantly reduce the computational load of watershed based image segmentation methods.