Abstract: In order to improve the problem that the performance of video face recognition system is largely influenced by the non-restraint environment, the HMM model is extended by adding the state centre parameter C, and the mapping algorithm on state space is based on facial features, which make the pattern distances smaller, the class distances larger, and improve the system’s robustness. The experimental results have indicated that this model improves the recognition accuracy and anti-jamming to the video frequency sequence.

Key words: Extended C HMM, video facial recognition, state space mapping

摘要： 为改善视频人脸识别系统性能受无约束环境影响较大的问题，通过引入状态中心参数C扩展隐马尔可夫模型，对人脸面部特征做状态空间隐射，使类内差别变小，类间差别变大，提高系统的鲁棒性。实验结果表明，优化后的模型提高了对视频序列识别的准确性和抗干扰性。

关键词: 扩展C型HMM模型, 视频人脸识别, 状态空间隐射