计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (19): 197-199.

• 图形、图像、模式识别 • 上一篇    下一篇

DCT系数重组在特征脸中的应用研究

张娜娜1,吴 翔1,马 燕2   

  1. 1.上海建桥学院 信息技术系,上海 201319
    2.上海师范大学 计算机系,上海 200234
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-07-01 发布日期:2011-07-01

Research on eigenface based on reordered DCT coefficients

ZHANG Nana1,WU Xiang1,MA Yan2   

  1. 1.Department of Computer Science,Shanghai Jianqiao College,Shanghai 201319,China
    2.Department of Computer,Shanghai Normal University,Shanghai 200234,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-07-01 Published:2011-07-01

摘要: 针对现在图像压缩主要采用DCT变换方法,提出了一种对DCT系数进行重组的方法,获取三级多分辨率子带。通过对各单独子带实施特征脸方法可得到不同的识别效果。提出在子带[S0]和子带[S1],[S2]和[S3]上进行特征脸方法识别,如两者结果不一致,则在满足一定门限条件下输入多张人脸,分析了所提出方法与原方法所需识别时间的关系。在ORL库和YALE库上的实验表明,提出方法的识别率要远高于直接采用特征脸方法。

关键词: 人脸识别, 特征脸, DCT变换

Abstract: Because current image compression schemes are mainly focused on DCT transformation,three level multiresolution subbands from reordered DCT coefficients for face image are acquired ahead of the using of eigenface method.Different results are achieved by applying the eigenface method on each single method.A method is proposed which recognizes the eigenface on subband [S0] and subband [S1],[S2] and [S3].If this paper can get different results,more face images are input under certain threshold condition.The relationship of recognition time between the proposed method and the original method is analyzed.The experimental results on ORL and YALE database show that the recognition rate of the proposed method is higher than those of directly using the eigenface method.

Key words: face recognition, eigenface, DCT transform