Lip segmentation using automatic selected initial contours based on localized active contour model

摘要: With the rapid development of artificial intelligence and increasing popularity smart devices, human-computer interaction technology has become a multimedia multimode from being computer-focused to people-centered. Among all ways interactions, using language interact with machines is most convenient efficient one. However, performance audio speech recognition systems not satisfied in noisy environment. Thus, more researchers focus their works on visual lip reading technology. By extracting movement features speakers rather than features, can get superior results when noises interferences exist. Lip segmentation plays an important role system, since result crucial final accuracy. In this paper, we propose localized active contour model-based method two initial contours combined color space. We apply illumination equalization original RGB images decrease interference uneven illumination. A space consists U component CIE-LUV sum C2 C3 components image after discrete Hartley transform. select rhombus as closed mouth, because it similar shape lip. For open utilize semi-ellipse both outer inner boundaries. After attaining each separately, merge them together obtain result. From experiment, conclude that better compared circle segment gray space, especially for mouth. An extremely obvious advantage mouth excluding internal information such teeth, black holes, tongue, introduction contour.