Abstract: In order to realize the content-based men’s suit image emotional semantic recognition, the primitive features of men’s suit image need to be mapped to emotion semantic space. Based on the formed two-dimension image emotional factor space and the primitive features of men’s suit image (ten-dimension of brightness, warm-cold fuzzy histogram; seven-dimension of saturation, warm-cold fuzzy histogram and color contrast value) in our previous work, the machine learning, BP neural network could help to realize the mapping that from the primitive features of men’s suit image to the emotion semantic factor space. The primitive features of image color could finish the image emotional factor values and description values calculation automatically, and the new image data could be added after identification to the image emotion database automatically. Experiments have proved that BP neural network method can be a better method to realize the content-based men’s suit image emotional semantic recognition.

Key words: men's suit image, emotional semantic, color feature, BP neural network, image recognition