Vehicle monitoring video image segmentation based on improving Watershed Algorithm

摘要

This article proposes an improved watershed algorithm based on morphological, using morphological open and close reconstruction to reduce the noise of image. Besides, with the Weber Perception Principle, it divides the morphological gradient image in the form of gray scale and non-linear and marked the gradient image with Watershed Algorithm. So it solved the over-segmentation problem of traditional watershed algorithm and the suppression problem of noise and fine-grained before watershed mark. By the above method, the target image can be extracted from Vehicle surveillance video, which simplifies the steps of traditional segmentation of the video image, and improve the accuracy of image segmentation.