A speech synthesis system is designed for automatically recognizing visual speech generated by the speech impaired and present a new communication approach for the speech-impaired people. In order to acquire more parameters of lip contours, a new model is made up, which can extract the degree of pouting from it. At the same time, the differential coefficients of some parameters are calculated to describe dynamic characteristic of the lip contours. Movement detection and morphological processing are used to extract mouth area and parameters of lip contours from the image sequence.