Garment image segmentation combined with a salient region and JSEG

摘要

Many model pictures will be adopted in the network marketing of garment products because this method is the most intuitive. Human eyes can easily identify the garment styles, while computer programs struggle to automatically do so. This paper proposes a method for garment image segmentation in which the JSEG algorithm is applied to the image salient region segmentation, and the background region and dressing region are identified with the help of face detection. This method was proven effective based on actual image samples tested at an online shopping mall, which lays a foundation for extracting the garment structure and characteristics as well as visualized retrieval.