基于多尺度SVD的HMM的人脸识别方法  

Face Recognition Method Based on Multi-Scale SVD Coefficient Along with HMM

作　　者：杨露[1] 何静媛[1] 

机构地区：[1]重庆大学计算机学院,重庆400044

出　　处：《现代计算机》2017年第6期106-110,共5页Modern Computer

摘　　要：提出使用多尺度SVD作为图像的特征提取以及HMM作为分类器的人脸识别方法。首先,对图像进行多尺度的划分,然后计算每个图像块的SVD,该图像的多尺度SVD特征向量便由图像块的SVD组成,最后使用HMM进行人脸识别。多尺度SVD能反映整幅图像的全局特征和图像在多种尺度下的局部特征,所以它的鉴别信息更全面。使用的7状态的隐马尔科夫模型,覆盖更多的人脸细节,有助于提高人脸识别效率。实验结果表明,该方法的人脸识别效率较高。Proposes a method of face recognition using multi-scale SVD for image feature extraction and the HMM classifier method. First of all, segments image into multi-scale image blocks. And then calculates the SVD for each image block, the multi-scale SVD feature vector of image is composed of the SVD of all image block. Finally, uses HMM for face recognition. In addition to reflect the whole global features of the image, multi-scale SVD feature vectors also reflects the local characteristics with multiple scales of the image, so its identification information is more comprehensive. After that, uses 7 state hidden Markov model, coverage for more details on face, which can help to improve the efficiency of face recognition. The experimental results show that rate of face recognition is considerably high.

关 键 词：奇异值分解 人脸识别 隐马尔科夫模型 多尺度 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]