Abstarct: In this thesis, a fast and accurate method is proposed for image retrieval. In the recent years, the number of images increased drastically in social networks in internet and public databaxses. According to the subject, retrieval systems became important. Image retrieval has many usages in the fields of medical, military, industrial and multimedia environment. A Content baxsed image retrieval system retrieves the image baxsed on its visual contents which is different from text baxsed image retrieval. So the speed of computer systems is becoming a significant feature. Beside acceptable accuracy, the speed is also a priority. In the proposed method, both color and edge features are used. The combination of color histogram features HSV and edge histogram features (SRF) baxsed on local binary pattern (LBP) form the proposed system. Beside the two mentioned features of Canberra match criteria, it is used for increasing efficiency. How to retrieve image is that the system assess the sample with all images after receiving the query image and provide the most similar image. Wang databaxse has been used to evaluate the proposed method including 1000 images in 10 concept classes. Eventually, according to evaluating proposed system with the up to date and current systems, it is seen that our system have better accuracy and speed than current methods.