The desire to explore the resources of the seas and oceans is the main reason for the development of underwater robots. Searching through underwater environment for detection and tracking targets is a useful action in this erea. The solution of this problem will be useful in other branches of science such as Mechanical engineering, Control, Biology and Artificial Intelligence. From the perspective of artificial intelligence and robotics, Computer Vision, the underwater environment is considered for intelligently acting. While active sensors such as sonars which have high energy and price were an ideal choice to use in underwater environment, there is a lost comings guide us to use other approaches. This study aims to assess the needs of building an autonomous underwater robot using vision sensors. Among the most important achievements of this research can be pointed to Reliance on very low cost components and sensors on the Internal market. The idea of ??combining the two methods of key point (SURF) matching and color pursuit algorithm (CAMShift) to find objects in the underwater environment id another new work. According to the results, we can say it is not enough merely to use color in underwater environments, beacause it would be misguided robot, due to the perturbation is caused by the different colors in the environment. Also, in many cases, failure of the robot will be followed by the target. Due to the use of local features (SURF), the work has been done in this study, less will failure. Keywords: Autonomous Underwater Vehicle, Computer Vision, SURF, CAMShift, Detecting and Tracking Target
