Based On U-face Model And Genetic Algorithm Face Recognition Technology Research

This article focused on two aspects of the research: Firstly, the author first roposed the concept of U-face in the recognition algorithms field, U-face is a face photographs pretreatment model, and simulation results showed that the U-face improved the recognition rate, to the author's desired effect. Secondly, the author further improved PCA facial recognition algorithm by genetic algorithm, the author based on the previous proposed a three-point improvement, experimental results showed that the improved algorithm is better than the previous. These two aspects are detailed below:1. U-face is the human face photos pretreatment model, it removes background, hair, and other factors that are not conducive to face recognition, and replaced with the standard background, only left the critical parts of the face; to some extent to eliminate that time effects on the human face; U-1 is the ideal model; U-2 is a rough model; U-3 model is more similar to the model U-1; U-4 model is an extension of U-face model, the model is essentially different from the other three models.2. Genetic algorithm can be applied to optimal selected the feature space of PCA face recognition method,the author based on the previous proposed a three-point improvement. Traditional methods to determine the feature space has some limitations and can not achieve the optimal identification. We can use genetic algorithms to improve it. Three improvements: the first is the improvement of genetic algorithm coding bits, the original N-bit if it is, and now only need N-1 position, and can achieve the same effect, reducing the time complexity and space complexity. Secondly, the initial population of the predecessors is randomly determined; the author is based on the distribution of eigenvalue and eigenvector to determine the non-random initial population method, it improved the efficiency of search. Finally, we save the maximum fitness for each generation of all chromosomes in the process of genetic algorithm; we optimal select eigenvalue and eigenvector according to variety ways after run of the algorithm.