An Efficient System For Cloth Segmentation And Color Extraction Based On An Improved Graphcut Approach

Auto object segmentation in unconstrained images has been widely applied in several areas of image processing. Among those areas, cloth matching is now adding its appeal for common mobile Internet users. This paper presents an efficient system for the two basic steps of cloth matching, cloth segmentation and color extraction. We use a light and convenient improved Graphcut approach to achieve the balance of efficiency and effectiveness. The experimental result demonstrates that we arrive at an eminently suitable point for the cloth segmentation. This paper describes how the system is constructed and depicts how the algorithm-based workflow is designed in the domain of clothes.