Comparison of min-cut models for image segmentation

Image segmentation is a core task in computer vision and plays an important role in object recognition and image analysis. Image segmentation based on s/t graph cut combines region information and boundary information. It achieves segmentation by transforming the image segmentation problem into an energy function minimization problem and applying the min-cut algorithm to find a globally optimal solution. s/t graph cut methods are globally recognized for their high efficiency and accuracy in several image segmentation scenarios and their applicability to the N-D problem. In this paper, we detail the combinatorial optimization framework of the s/t graph, construct different min-cut models, use the optimization solver Gurobi to solve these models, compare and analyze the effectiveness of these models in image segmentation, and evaluate the performance of these models in different scenarios and conditions.