摘要： 二维保局投影(2DLPP)只在图像的横向进行数据压缩，提取的特征维数较高，针对该问题，结合二维保局投影和可选的二维保局投影，提出双向压缩二维保局投影((2D)2LPP)算法。该算法从横向和纵向2个方向实施2DLPP，使图像的横向和纵向的维数都得到有效的约简。实验结果表明，(2D)2LPP在识别率和识别时间上都优于2DLPP和A2DLPP。

关键词: 保局投影, 二维保局投影, 特征提取, 人脸识别

Abstract: Aiming at the problem that Two-Dimensional Locality Preserving Projections(2DLPP) only in the horizontal direction of the image data compression and feature’s dimension is high, an Alternative Two-Dimensional Locality Preserving Projections(A2DLPP) method is given, and Two-direction data compression Two-Dimensional Locality Preserving Projections((2D)2LPP) is proposed. The algorithm executes 2DLPP separately from both horizontal and vertical directions, so that the image of the horizontal and vertical dimensions is an effective reduction. Experimental results show that both the recognition rate and recognition time of (2D)2LPP are better than 2DLPP and A2DLPP.

Key words: Locality Preserving Projections(LPP), Two-Dimensional Locality Preserving Projections(2DLPP), feature extraction, face recognition

中图分类号: 

• TP391.4