An Image Retrieval System Based On Color Emotion

Nowadays images in the Internet, as one of the carrier of the massive information, have been developed by leaps and bounds. One of the disadvantages of the traditional content-based image retrieval technology is the "semantic gap", which could not get the semantic meaning from the basic image features. In order to explore the new direction of new image feature extraction, which is intuitive and closely to the human sensory, and shorten the semantic gap between image low-level features and upper semantic meaning, we proposed a scheme of image retrieval system based on the color emotion.Affective computing model of monochrome color is based on color psychology research, mapping colors to three-dimensional color-emotion model color activity, color weight, color heat. Then we proposed mixed emotions model. Combined with the determination of the color center of the image, you can extract the emotional value of the global image.Through analysis on image in the same theme, we use color emotion histogram intersection as the similarity of the image retrieval calculation model. Subsequently combined the relevant feedback with the fast particle swarm algorithm to accelerate the retrieval algorithms.By the experiment of the10categories of image data, the contrast ratio of the color histogram index method using the average precision of the search algorithm of the color emotions yield in a better result. The image retrieval system of color emotions based on the image retrieval system can be used as the image source management, and would be widely used in the field of image processing.