Author(s)

Zhang Xiaomeng, Zhang Haibo and Zhang Zekun

Corresponding Author

Zhang Haibo

ABSTRACT

Emotional semantics identifying and retrieving of images is one of the hotspots in current research. It can provide certain basis for clothing design, clothing wearing and clothing matching. As a kind of male T-shirt, the recognition and retrieval of emotional semantics of images has also attracted much attention. On the basis of establishing the relationship between emotional semantic space and low-level features of T-shirt images based on the previous knowledge of clothing field, 40 T-shirt images were trained and a machine learning model was established by using support vector machine (SVM). The quantitative mapping relationship between emotional semantic space and low-level features of T-shirt images was established, which could be calculated automatically. The emotional semantics of the image is calculated to realize the emotional semantic recognition of the image. Similarity measure algorithm is also used to measure the emotional semantics of images, and image emotional semantic retrieval is realized. Through the experiment of software programming and development, a better recognition and retrieval effect have been achieved, thus verifying the feasibility and effectiveness of emotional semantic analysis of images combined with domain knowledge.

KEYWORDS

Male T-shirt, emotional semantic, support vector machine, image recognition, image retrieval