佳木斯大学学报（自然科学版）
佳木斯大學學報（自然科學版）
가목사대학학보（자연과학판）
JOURNAL OF JIAMUSI UNIVERSITY (NATURAL SCIENCE EDITION)
2015年
2期
258-261
,共4页
人脸识别%光照变化%相关滤波器%最优投影图像%重构图像
人臉識彆%光照變化%相關濾波器%最優投影圖像%重構圖像
인검식별%광조변화%상관려파기%최우투영도상%중구도상
face recognition%illumination variation%correlation filter%optimal projection image%refactoring image
为了减少人脸识别中由于光照变化而严重影响识别性能的问题，设计了一种基于相关滤波器的最优相关滤波算法。采用二维主成分分析法重构人脸图像，生成相关性过滤器（ OPICF ）及重构相关性滤波器（ RCF）；再将测试图像投影到二维子空间，利用重构相关性滤波器重构图像；最后，根据提前设定的光照容限阈值完成人脸的识别。在扩展YaleB人脸数据库上对提出的算法进行了有效性验证，结果表明，本文算法有更高的识别率。
為瞭減少人臉識彆中由于光照變化而嚴重影響識彆性能的問題，設計瞭一種基于相關濾波器的最優相關濾波算法。採用二維主成分分析法重構人臉圖像，生成相關性過濾器（ OPICF ）及重構相關性濾波器（ RCF）；再將測試圖像投影到二維子空間，利用重構相關性濾波器重構圖像；最後，根據提前設定的光照容限閾值完成人臉的識彆。在擴展YaleB人臉數據庫上對提齣的算法進行瞭有效性驗證，結果錶明，本文算法有更高的識彆率。
위료감소인검식별중유우광조변화이엄중영향식별성능적문제，설계료일충기우상관려파기적최우상관려파산법。채용이유주성분분석법중구인검도상，생성상관성과려기（ OPICF ）급중구상관성려파기（ RCF）；재장측시도상투영도이유자공간，이용중구상관성려파기중구도상；최후，근거제전설정적광조용한역치완성인검적식별。재확전YaleB인검수거고상대제출적산법진행료유효성험증，결과표명，본문산법유경고적식별솔。
An optimal correlation filter algorithm was proposed in face recognition by illumination varia-tion, in order to reduce the problems due to changes in illumination which seriously affects the recognition per -formance.A couple of correlation filters were generated using 2DPCA with class specified to reconstruct face ima-ges.Then, the filter was used to project testing images to two -dimensional subspace , and refactoring correlation filter was used to reconstruct images .Finally , face recognition was finished by preset lighting tolerance thresh-old .The effectiveness of proposed algorithm has been verified by experiments on YaleB databases .Experimental results show that the proposed algorithm has higher recognition rate .
