In order to overcome the problem of over-segmentation, a novel algorithm of watershed segmentationbased on morphological gradient reconstructing is proposed in this paper. In the algorithm, morphological gradient reconstruction is introduced in watershed segmentation, and opening and closing by reconstruction operators are employed to reconstruct gradient image. And then a lot of gradient pixels with low value are removed and some gradient pixels with high value are preserved. Therefore, watershed segmentation arithmetic is developed.
The improved algorithm is applied to reconstruct gradient image. The whole segmentation process needs no post-segmentation which reduced the complexity of the segmentation in some degree. The experimental results show that the improved algorithm can efficiently eliminate over- segmentation, and hold the position of region contours without evident bias.