Introduction to statistical pattern recognition (2nd ed.)
Introduction to statistical pattern recognition (2nd ed.)
Using Discriminant Eigenfeatures for Image Retrieval
IEEE Transactions on Pattern Analysis and Machine Intelligence
Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection
IEEE Transactions on Pattern Analysis and Machine Intelligence
Two-Dimensional PCA: A New Approach to Appearance-Based Face Representation and Recognition
IEEE Transactions on Pattern Analysis and Machine Intelligence
Compact Representation of Multidimensional Data Using Tensor Rank-One Decomposition
ICPR '04 Proceedings of the Pattern Recognition, 17th International Conference on (ICPR'04) Volume 1 - Volume 01
| Hi-index | 0.00 |
Traditional 1D vector based FDA algorithm is popular used in face image retrieval. In FDA, data is represented by 1D vector, which is converted from image matrix. Usually, this conversion makes the number of examples less than that of data dimension, which will give rise to small sample problem. To overcome this problem, 2D matrix based algorithm is proposed, in which the within-class scatter matrix is derived directly from matrix. In the existing matrix based algorithms, IMPCA and GLRAM don’t utilize discriminant information between classes. Although TDLDA goes further, yet it is solved by iterative steps. Here we propose a new matrix based technique: IMFDA. It not only takes the advantage of discriminant information between classes, but also can be solved as a generalized eigenvalue problem. Experiments on ORL face database show that the new algorithm is more efficient than IMPCA, GLRAM and TDLDA with lower test error and shorter running time.