CAJ | 학술논문

张量主成分分析法（ TPCA ）用于人脸特征提取，克服了传统的基于统计特征的特征提取方法会破坏图像原始结构的问题；而源图像经过非下采样剪切波变换后得到了 k 个大小相同但尺度不同的带通图像，具有良好的时频分析特征。为了更好地提取人脸识别特征，提出了非下采样剪切波融合 TPCA 的人脸特征提取算法，该算法先对源图像进行非下采样剪切波变换得到4个子代图像，再对子代图像进行 TPCA 特征提取得到特征集，实现人脸的高效识别。实验结果表明，该算法明显优于原有的单一算法。
장량주성분분석법（ TPCA ）용우인검특정제취，극복료전통적기우통계특정적특정제취방법회파배도상원시결구적문제；이원도상경과비하채양전절파변환후득도료 k 개대소상동단척도불동적대통도상，구유량호적시빈분석특정。위료경호지제취인검식별특정，제출료비하채양전절파융합 TPCA 적인검특정제취산법，해산법선대원도상진행비하채양전절파변환득도4개자대도상，재대자대도상진행 TPCA 특정제취득도특정집，실현인검적고효식별。실험결과표명，해산법명현우우원유적단일산법。
Compared with the PCA method, TPCA for face recognition has a higher recognition rate. The source images through non-subsampled shearlet transform can get band pass images that have better feature of image with the same size and different scale. In order to extract better general face features, this paper proposes an algorithm based on non-subsampled shearlet transform and TPCA. Non-subsampled shearlet transform is firstly used and then TPCA is used to extract the feature of subband images , and the efficient recognition of face images can be realized. Experiment results show that the recognition rate of the proposed algorithm is higher than the original algorithm.