Digital Image Retrieval Technology Based on Multi-feature

At present the most commonly used retrieval method is based on the text. The disadvantage to this approach is the big workload, and it need to artificially add tags, in which the subjectivity is strong. Therefore the retrieval effect is not ideal. Therefore, content-based image retrieval technology is becoming a hot spot of research, which use of image color, texture and shape features mainly for image retrieval, and so far there is not a universal retrieval method for any image. In this paper, the specific gallery - cheongsam library is selected to study. Based on texture feature of cheongsam library, this paper selects two kinds of texture retrieval method -- wavelet transform and LBP, and an integrated feature retrieval algorithm is put forward.