WEIGHTED-TYPE IMAGE SEGMENTATION MODEL VIA COUPLING HEAT KERNEL CONVOLUTION WITH HIGH-ORDER TOTAL VARIATION

Image segmentation is an essential step for many applications in the field of the image analysis. One of the main challenges for this task is how to accurately locate complicated boundary and properly segment a region of interest efficiently. To this end, this paper provides a new scheme by combining the adaptive weight function and the high-order total variation term to improve the robustness of the classical active contour model. In order to reduce the computational complexity, our model uses the heat kernel convolution with adaptive weight to approximate the perimeter of the segmentation area. Due to the nonsmoothness of the proposed model, we adopt the alternating direction method of multipliers to solve it. Numerical implementations on several different types of images illustrate that our proposed scheme demonstrates better segmentation performance and robustness than several existing state-of-theart segmentation models.