Human Face Recognition Based on Improved D-LDA and Integrated BPNNs Algorithms
In this paper, a combination methodology of Discrete Cosine Transform (DCT) and an improved D-LDA and Neural Networks was proposed. DCT can compress the information of original signal efficiently, so the authors reduce the dimension firstly and then extract features by improved D-LDA on the low dimension space to overcome the shortages of LDA maximally. After calculating the eigenvectors and a new Fisher's criterion using improved D-LDA algorithm they proposed, the projection vectors are calculated for the training set and then used to train the neural networks for human identity. The experimental results on ORL face database show that this combined method has well performance.