The Rearch Of Color Image Retrieval Based On Color Feature Of Object

With the advent of multimedia technology age, we can get a large number of image information. How to get the image information effectively and efficiently has become a hot research field, and content-based image retrieval (CBIR) technology emerges as the raised. For the present, content-based image retrieval has focused mostly on the low level features on global. The global features are relative simple, fast to calculate. But the retrieval effect will be affected by background information in the image. Retrieval based on object features can reduce the disadvantage in global retrieval. The method extracts the object interest by user with the image segment.This paper proposes image retrieval base on the object. Firstly, we extract the target region by segment the image, then retrieval the image in the object. In the segment part, we find the dominant color from the image by find the peak in quantized histogram at first, pixels are clustered to get the dominant color image, then using the OTSU to find the best threshold, segment the image by this threshold into two parts, object and background.In the feature extraction part, In order to get this descriptor, the dominant color and its percentage value are computed, Every image was divided into image blocks, and give every block a proper weight. The weight is correlation by the interesting of the user. At last using the Euclid distance as the similarity measure.Experimental results indicate the method can extract the object, and the retrieval is prior to using global features, improve the effectiveness.