Abstract: Aiming at the shortcomings of the traditional watershed segmentation algorithm, such as noise sensitivity and high rate of false segmentation, we proposed a novel image segmentation method based on improved watershed algorithm in order to obtain better image segmentation result. Firstly, the original image was filtered to eliminate the interference of noise to image segmentation, and the edge information of target was effectively protected. Secondly, image was transformed to enhance contrast of the image, and then image was segmented by watershed algorithm. Finally, several types of images were tested on the MATLAB2014 platform. Simulation results show that the proposed method can avoid effect of noise on image segmentation results, and can obtain better  image segmentation results, moreover, the segmentation accuracy and efficiency are better than other image segmentation methods.

Key words: noise interference, filtering processing, marker watershed algorithm, image segmentation, edge information