Research on Image Retrieval based on multi-features

Content-based Image Retrieval(CBIR) is the hotspot of image retrieval. Extracting features and computing similarity are two key points of CBIR. Color feature is the most popular feature used in CBIR, which is high efficient but always lacks the shape information. In this paper, we propose a method combining the color feature with the edge feature, and use the combing feature to retrieve images. Experiments show, compared to the original CBIR method based on color feature or edge feature, our method significantly resaises the average precision and recall ratio, while keeps the high efficency.