The medical images have pathological change region and background. The pathological change region that is so-called the region of interest (ROI) is the emphasis of image segmentation. Aiming at the problems of sensitivity to noise and over-segmentation existing in traditional watershed algorithm，an image segmentation strategy on the combination of multi-scale morphology gradient and fuzzy enhancement and marker controlled watershed is proposed. Firstly，combined with respective advantage of the big structural elements and small structure elements，over-segmentation is weakened by multi-scale morphology edge detection; Secondly，the edge of original medical image is enhanced by using the fuzzy enhancement algorithm; Finally，the image is segmented with a marker-based watershed algorithm. Simulation experiments show that the algorithm can effectively overcome the serious watershed transformation over-segmentation problem and get significant regional segmentation，and it has strong regional outline positioning ability and don't need to carry on the following merger processing，and segmentation algorithm is simple，and has the characteristics of multiple scales，and it can adapt to the demand of the medical image classification and information extraction.