Research of Image Retrieval Based on Uniting Features

Research the current content-based image retrieval technology. In order to improve the limitations of single feature image retrieval and the computational complexity of integrated features image retrieval is so complexity, so it brings up a new method based on uniting features. The mainly theory is that combines Tamura texture feature with wavelet transform texture analysis of resist geometric distortions to extract 13-componet texture features; after the Daubechies3 wavelet transform, and extracts the color histogram and 7 shaped invariant moments at the low-frequency region; then implement a system of uniting features image retrieval. The Experiments show that it has better search results with no apparent increasing amount of calculation than the feature of a single or other integrated feature to retrieval.