Research On Technology Of Face Recognition Based On QGA-BP Neural Network

It's well known that the technology of human face recognition is important in biology feature recognition field. It's used very extensively in visiting control, judicatory, electronic business affairs. Human face recognition has become an activity research question. Though a lot of progress has been made by many researchers these years, many key problems still have to be solved in order to popularize the application of face recognition because of the complexity of face recognition.In this paper, the development and main methods of face recognition are introduced firstly, then a face feature extraction method which is based on M-FastICA is used in the paper, moreover, BP neural networks is used as recognition classifier. The simulation testing which has better recognition effect, indicates that the human face recognition method used in the paper is feasible.In aspect of human face feature extraction, how to extract human face n feature by PCA and ICA based on the overall algebra feature has been deeply researched. The results show that the human face feature which is extracted by FastICA is more effective .But for processing the online data of human face recognition , FastICA has the shortage of needing large calculation. Therefore, a kind of human face feature extraction method based on M-FastICA is designed in this paper. The M-FastICA algorithm can simplify the calculation of Jacobian in Newton iterative process. The simulation results show that the algorithm not only inherit the advantage of FastICA effective, but also reduce the iterative frequency and duration of algorithm convergence .In aspect of the human face recognition aspect, BP neural network is used as the human face classifier in the paper. Focusing on the relationship between hidden nodes, network convergence rate and face recognition rate, experimental results show that the method in the paper has some recognition effect, but the training time is much longer and the requirements of the online real-time application can't be met.In order to overcome the shortage of BP neural network classifier, the (Quantum Genetic Algorithm) QGA designed in this paper optimizes BP neural network's connection power value. The results show that, this optimization method can greatly shorten the time of the global optimal solution searched by the neural network power value. QGA is used to optimize the BP neural network's connection power value, and design the improved BP neural network classifier based on QGA in this paper. The classifier used in the ORL database on the classification of human face recognition, has made95.83% of the average recognition rate. The results show that the performance of the improved BP neural network classifier based on QGA is better than the BP neural network expression recognition classifier.At last, analysis and expectation about of human face recognition as well as QGA are described in the paper.