Content-based Image Retrieval Using Color Difference Histogram in Image Textures

The aim of content-based image retrieval system is finding similar images to the query image from a database based on its visual content. In this paper, a novel retrieval system based on human vision is proposed. A factor that has a high impact on the search process is a set of features which are used in. The recent studies emerged that the human eye system considers the image content, texture, and color properties more than other features. Therefore, to retrieve more precisely the images, features should be used that are close to the human eye system. In the current paper, at first, the texture is extracted from the images using the local binary patterns algorithm. After that, the color differences of two adjacent pixels with the same texture are calculated in the HSV color space. Afterward, the histogram is taken from the color difference values. The obtained features from the histogram can describe the visual content of the images in more detail. Finally, the effective features are selected based on their entropy value. The prominent advantage of the proposed method is the lack of implementation of segmentation, clustering, training, and any other method of machine learning, which requires a lot of processing and time. The method is evaluated on two standard Corel 10K and Corel 5K databases, and its retrieval rate is significantly improved compared to some recent methods.