Lip-reading is used for capturing information of a speaker through his/her lips' movement. this technology is wildly used in speech recognition , ID recognition , Human-Computer intelligent interface and Multimedia system.As lip-reading's fundamental component ,lip-detecting is used for locating lips fleetly and exactly. It's accuracy will affect the whole lip-reading system. Because of the chromatic characters of lips, Recent years ,a lot of work which made use of the chromatic characters to detect lips was carried out ,and some achievement was accomplished. but ,there is no perfect algorithm.In order to locate the lip picture exactly in a lip-reading system, we investigated the chroma of skin and lip colors on a chromatic picture of human face.Furthermore, we proposed a lip detecting method, which uses the distribution of R, G, B colors in skin and lip colors. This method bases on the distribution difference of G, B colors and R, G colors, and amplifies this difference to determine human lip in a picture. We also validated the method by comparingour results with those from "Red exclusion" method. The comparing results show that the validity, robustness and the support of different human complexion is improved dramatically while the complexity of this method does not increase remarkably. Finally ,we use HDL to implement the proposed method ,the results show that the designed system reaches requirements. |