The Modular-2DPCA is an improvement and promotion of 2DPCA. Modular-2DPCA method creates the covariance matrix by blocked sub image, which make better robustness. There are many parts in a human face, with each part could own different weight in face recognition, block made the research of parts became possible. This paper directs the characteristic of block, calculating the mean value and covariance matrix for each sub image block, with which can extract the features of each part of human face more accurate. Theoretically, this method can efficacious reduce the effect of changed facial. Otherwise, this paper contains a preliminary research of sub image weights setting. Weights of parts can further raise contribution of some special parts of human face. Appropriate weights can improve the result of recognition. Experiments show that this method can efficacious improves the insignificancy of Modular 2DPCA in features extracting and raise the correct result of recognition.
[1]
Alejandro F. Frangi,et al.
Two-dimensional PCA: a new approach to appearance-based face representation and recognition
,
2004
.
[2]
Surendra Ranganath,et al.
Face recognition using transform features and neural networks
,
1997,
Pattern Recognit..
[3]
Garrison W. Cottrell,et al.
Organization of face and object recognition in modular neural network models
,
1999,
Neural Networks.
[4]
Xiao Chang,et al.
Is Two-dimensional PCA a New Technique? 1)
,
2005
.
[5]
M. Turk,et al.
Eigenfaces for Recognition
,
1991,
Journal of Cognitive Neuroscience.
[6]
David J. Kriegman,et al.
Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection
,
1996,
ECCV.
[7]
Norbert Krüger,et al.
Face Recognition by Elastic Bunch Graph Matching
,
1997,
IEEE Trans. Pattern Anal. Mach. Intell..