With the advancement of technology and the increasing use of the Internet, image search and retrieval has been become one of the most important issues during last decades. Content-based image retrieval is one of the techniques for image retrieval .This technique uses image content instead of metadata like keywords and annotation in order to retrieve images. Because all the pixel values of the image are not important, the image content should be described using some features.The most common low-level features for describing the image are color and texture. Many descriptors have been proposed in order to extract each of these features. After extracting features from images and their description, the similarity between query image and each database image is computed. Subsequently the images will be sorted according to these criteria and the result will be shown to the user. Semantic gap is one of the problems of these systems which mean the retrieved images may be different from the willing of the user. To overcome this difficulty, relevance feedback for image retrieval has been introduced. In these systems, if the user is not satisfied with the result, the user will be asked to label retrieved images as relevant or irrelevant. The use of these images is a way to improve performance. In this thesis, image retrieval systems without feedback and based on feedback have been proposed .In the proposed systems the Bag of Visual Word technique has been used for constructing image’s descriptor.This technique has some drawback, for address such problems soft-assignment and spatial bag of visual word approach has been used. In order to assign appropriate weight to each feature and select the appropriate combination of these features, genetic algorithm and forward selection has been used, respectively. For contributing user feedback, in image retrieval system based on the feedback, a method of weighting for each component is presented. Implementation results indicate good accuracy of the proposed methods. Keywords: Content-based image retrieval, Bag of Visual Word, SIFT, Hue descriptor
