An Improved Image Retrieval Algorithm

Abstract

This paper introduces the whole process of retrieval based on color characteristics.First of all,this article presents the HSV color model in brief,then HSV color model is selected to verify other links to the choice of methods.Next,the color space is quantized and the quantization standard can be divided into two kinds:equidistant quantification and non-equidistant quantification.Then it is choosing proper feature extraction method.General color histogram can only express the global statistical information of the image without the color space,so this article adopts the method of local cumulative histogram,and in the choice of similarity measure,this paper compares Euclidean distance and the weighted distance,and then concludes the more effective retrieval results.Finally,through the contrast,this paper chooses the optimal image retrieval algorithm:HSV color model-Local accumulate histogram-Euclidean distance-Equidistant quantification.