In order to solve the problem of external interference on human face detection and accurate tracking and avoid relying too heavily on soft technology which can increase algorithm complexity,we propose a method for face detection and accurate location tracking  based on Mega2560 microcomputer (MCU),VS2008 development environment and OpenCV function libraries.The proposal combines software with hardware.To realize face detection,we first use Haar-like features method to find out the facial feature rectangule in the image and generate a cascade classifier via the AdaBoost algorithm.Then we select the image display window as the origin of the coordinate and calculate the location of face.Using the linear transformation method we can acquire  face location deviation,which is converted into angular deviation and sent to the lower machine Mega2560.Finally we employ the MCU to control the camera according to the received angular deviation information,which guarantees the face staying at the origin position all the time,thus acomplishing face location and accurate tracking.Experimental results show that the proposed system has good stability and accuracy.