Design And Implementation Of Visual Mouse System Based On Opencv

With the development of image processing and recognition technology and perfect, human-computer interaction is evolving, get around. The development of human-computer interaction will not only bring more convenience also increased the work efficiency, but also changes the traditional life for disabled people, they can facilitate the use of computer, make their living conditions improve. Though the image processing technology is still need to improve and perfect, regular mouse and keyboard input device and can not be replaced, but the disabled itself can not be selective, for them is a feasible scheme for the better, so that they can use computer to communicate with the outside world, to learn to improve or entertainment to relax. Visual Mouse computer aided system based on image processing and computer vision technology to develop alternative to the mouse as the computer input so as to operate the computer.This thesis presents the design and implementation of Visual Mouse system, this system by means of a camera, the camera to capture video input, the face detection algorithm for the real-time detection of the face, the face detected by the face tracking, motion capture user, so as to control the computer screen mouse movement; use of feature detection algorithm to find the face in the eye, and monitor its status, analysis of changes in the eye state, interaction changes in user and computer. Visual Mouse system is a collection of human-computer interaction, application of information processing, face detection, eye detection, face tracking and other related technologies, the Adaboost method is detail parts of the face classifier, to improve the accuracy of face regions, Adaboost algorithm focuses on training and analysis of classifier, and on this basis to build a detector the prototype, experiments show that our detection can realize video real-time face detection, can achieve higher detection rate; eye location algorithm using ASM, ASM and most statistical learning method, including train and test(or fit) the two part, also is the shape modeling of build and shape matching accuracy and robustness of fit, the ASM algorithm has higher, as one of the classic algorithm of the image feature location; face tracking algorithm based on Camshift module high real-time face tracking, the algorithm considers the scale and direction of target tracking in the process of change. Camshift algorithm using a weight image tracking, local maximum weight corresponds to the maximum density, said the possible target area, the iterative convergence to the density of the aorta.Experimental results show that, the Visual Mouse system in this paper can effectively detect the real time face to face the camera, left, right, up, down and real-time control mouse left, right, up, down, open the closed real-time control the mouse left click, right click on the eye.