An Image Edge Detection Method Based on Marked Watershed Algorithm

Aiming at the problems of traditional watershed edge detection algorithm in the study of abnormal behavior images of rail transit crowd, such as susceptibility to noise and over-segmentation phenomenon, an image edge detection method based on marked watershed algorithm is proposed. The method first uses a mathematical morphology method for smoothing and edge enhancement, and then uses an improved watershed algorithm based on marker for edge detection to obtain the closed edge of the target image, so as to mark the edge of the target image. The experiments show that compared with the traditional watershed edge detection algorithm, this method can reduce the influence of noise, effectively suppress the over-segmentation phenomenon, improve the edge detection effect, and detect clear and uniform edges, which provides a new image edge detection technique and method for the study of abnormal behavior images of rail transit crowds.