In order to realize the system design objectives of quad-rotor UAVs, such as autonomous flying, hovering and ground-target tracking, a camera is added to the system to capture the ground information. The result of image processing is used together with sudoku segmentation method of vision to obtain the location information of the searched target. The system consists of an ultrasonic height-setting module. The motion control of UAVs is simplified from the traditional six degrees-of-freedom to four degrees-of-freedom. The location information is introduced to the PID control system, thus to obtain the control commands of the four motors, so that the motors can move in all directions. In addition, dual controllers are used. One is used to control the motion of the aircraft, and the other to control each module of the system, so as to reduce the burden of data processing and speed up the processing. Finally, the expected autonomous tracking effect is achieved. Compared with the existing methods, the motion control idea is more simplified, and the tracking of the aerial vehicle is more sensitive.