Silhouette shape and detail texture based garment style recognition

摘要

Garment style is closely related to silhouette shape and detail shape. This paper proposes image processing technology to separate the silhouette curve of the garment and obtain the image representing the drape pleating shadow. Firstly, 13 descriptors representing the shape characteristics and detail texture are acquired. After that, a three layer BP neural network is employed to recognition the garment style. Finally, the proposed method is verified in the skirt style classification and the experimental result shows the effective of the method.