New watershed segmentation algorithm based on hybrid gradient and self-adaptive marker extraction

In order to overcome the problem of over-segmentation in the traditional watershed algorithm, this paper proposes a new watershed segmentation algorithm, based on hybrid gradient and self-adaptive marker extraction. The algorithm uses color space to calculate image gradient, and improves the gradient function with the information entropy theory. Then, this paper suggests a self-adaptive calculation method of h-minima threshold to mark the gradient image, and uses the marker to modify the gradient image. Finally, the watershed segmentation is applied by using the modified gradient. A series of experiments proved that the new algorithm can effectively enhance weak edges, and overcome the problem of traditional marker extraction which needs to set threshold parameters by users. Also the new approach can obtain a good watershed segmentation result, and effectively reduce the over segmentation phenomenon.