CAJ | 학술논문

针对压缩感知（Compressed Sensing，CS）方法需将图像矩阵转化为向量后进行特征提取，导致数据维数很大，计算复杂等缺点，提出二维离散余弦变换（2DDCT）和压缩感知（Compressed Sensing，CS）相结合的人脸识别方法。新方法首先利用2DDCT将图像变换到频域，压缩人脸图像以去掉人眼不敏感的中频分量与高频分量，这样有效降低了所需特征的维数．减少了计算量；然后通过感知算法进行特征提取得到人脸识别特征，最后运用最近邻分类器完成人脸的识别。在ORL、Yale及Feret人脸数据库的实验结果证明了该算法的有效性与稳健性。特别是在YaleB人脸数据库运用该方法得到了很好的试验结果。
침대압축감지（Compressed Sensing，CS）방법수장도상구진전화위향량후진행특정제취，도치수거유수흔대，계산복잡등결점，제출이유리산여현변환（2DDCT）화압축감지（Compressed Sensing，CS）상결합적인검식별방법。신방법수선이용2DDCT장도상변환도빈역，압축인검도상이거도인안불민감적중빈분량여고빈분량，저양유효강저료소수특정적유수．감소료계산량；연후통과감지산법진행특정제취득도인검식별특정，최후운용최근린분류기완성인검적식별。재ORL、Yale급Feret인검수거고적실험결과증명료해산법적유효성여은건성。특별시재YaleB인검수거고운용해방법득도료흔호적시험결과。
In this paper an improved face recognition algorithm is proposed based on the combination of 2D discrete cosine transform （2DDCT） and Compressed Sensing（CS）because of CS. CS first transforms an image matrix to a vector which caused high dimensionality and computational complexity. In this paper the original face image is processed by 2DDCT to reduce the character dimensions effectively. Then, the image is processed by CS to obtain the face recognition features. Finally, the nearest neighbor （NN） classifier is selected to perform face recognition. The experimental results on ORL , Yale and Feret face databases show that this method is robust and effective in the face recognition, especially in the face database YaleB.