An image retrieval system with color emotion query

The traditional content-based image retrieval technology can´t map underlying image features to semantics, leading to a “semantic gap” between the needs of users and the retrieval results. In this paper, we implement a color image retrieval system based on color emotions. By analyzing the distribution of color emotion histograms of the same theme image, we build a similarity calculation model, and then we use regional growth algorithm and fast particle swarm algorithm to accelerate the computation, and improve the accuracy on relevant feedback. Ultimately we achieve an image classification model based on emotional semantics. The test image data is from 10 classes of Corel standard image database. Experiments show that the color emotion-based image retrieval system in this paper can effectively extract the image emotional features, achieve a preferable retrieval results.