A New WEB Image Retrieval Method Combined with Color and Spatial Information
Lihui Lin · Journal of Chongqing University of Science and Technology · 2012
In order to improve the retrieval efficiency,this paper uses the Non-uniform quantization of the HSV color model to extract the main colors of the images,reduces the feature vectors of the WEB images,then uses the histogram intersection as a similarity measure for the first image retrieval.After obtaining the top parts of search results that are similar on the global color distribution,considering the spatial distribution of color,this paper uses the method of colors to improve the retrieval accuracy.Proved by the experiments the method used in this paper can effectively improve the performance of web image retrieval from the efficiency and accuracy.