关键词: 关键词: 唇部特征, 尺度不变特征变换算法, 特征提取, 识别, 分类

Abstract: Abstract: Since the feature of lip has high dimension and is sensitive to scale space, a method to do feature extraction for speaker authentication based on scale-invariant feature transform (SIFT) algorithm was proposed. Firstly, a simple video frame image neat algorithm was proposed. Namely, adjusting lip motion videos with different lengths to the same length and extracting representative lip motion pictures. Then, the other algorithm based on key points of SIFT was presented, through which texture and motion features can be extracted. Integrated with principal components analysis (PCA) algorithm,, typical lip motion features can be obtained to do identity recognition. Finally, a simple classification algorithm was presented according to obtained features. Compared with the common local binary model (LBP) feature and histogram of oriental gradient (HOG) feature, experimental results show that false acceptance rate (FAR) and false rejection rate (FRR) of the proposed feature extraction algorithm peform better, proving that the whole speaker lip motion recognition algorithm is effective and can get ideal results.

Key words: Keywords:Keywords: lip feature, Scale-invariant feature transform (SIFT), feature extraction, recognition, classification