Abstract: A face recognition system based on Apple company’s embedded operating system-iOS is proposed. Through the research of AdaBoost face detection algorithm based on Haar-like features, the real-time face detection is achieved. An improved method of face recognition based on hidden markov model is put forward. This method uses singular value compression to subtract the character of the face image, which not only reduces the amount of data storage and computational complexity but also solves the inefficient caused by processing large number of data of image in embedded system. The experiment result shows that rapid detection, high real-time performance and high recognize rate can be acquired in this method, which can be used as the base of other types of face recognition application software system in iOS platform.

Key words: iOS platform, face detection, singular value compression, hidden Markov model, face recognition

摘要： 设计了一个基于苹果公司嵌入式操作系统iOS平台下的人脸识别系统。通过对基于Haar-like特征的AdaBoost人脸检测算法的研究，实现了实时人脸检测。提出了一种改进的基于隐马尔科夫模型的人脸识别方法，此方法采用奇异值压缩抽取人脸图像特征作为观察序列，减少了数据的存储量和计算量，解决了嵌入式系统中由于图像处理数据量大造成的低效。实验结果证明，该系统检测速度快，实时性强，识别率高，可以作为iOS平台上其他类型人脸识别应用软件开发的基础。

关键词: iOS平台, 人脸检测, 奇异值压缩, 隐马尔科夫模型, 人脸识别