Color image segmentation using a morphological gradient-based active contour model

Segmenting objects of dynamic shapes and various colors is still challenging to the computer vision of natural images, because of slow computation, inaccuracy, and loss of information. In this paper, vie propose a novel segmentation algorithm based on active contour models, to overcome these weaknesses. First, vie apply a morphological gradient-based edge detector to an image, to extract its edge map. Because this step is performed directly on color images, it helps us avoid losing color characteristics, compared with grayscale conversion. Second, this edge map will be used as a clue to provide both good edge information and good region information for an active contour, without a re-initialization model. As a result, our proposed algorithm allows the contour to be initialized more flexibly, evolves the contour faster, and segments the boundary of objects more precisely in color images. Results attained on diverse natural images show its promising performance, compared with other models, for both accuracy and computational time.