作者: Jia Xibin , Bao Xiyuan , David M W Powers , Li Yujian
DOI: 10.4156/JCIT.VOL8.ISSUE5.33
关键词:
摘要: Considering the complexity of facial expressions, this paper adopts strategy face segmentation for expression auto-recognition. On training dataset, feature-based Gabor wavelets are extracted each block. Then classification is done to realize basic recognition determine which part contributes most a given expression. A binary weight matrix assigned ‘1’ highest such rate, otherwise ’0’. At test stage, voted feature calculated by multiplying and than relative rates obtained. Finally target determined voting. The experiment results show that plays better roles in representing express PCA DCT. Based on learning optimal area wavelet modeling entire extraction, system has effect.