Face Recognition Algorithm Based on Improved BP Neural Network

Face recognition has received wide concern as a hot direction in recognition models. Due to a strong self-adaptive and mapping ability, traditional BP algorithm occupies certain advantages in face recognition, but is has the shortcomings of fast convergence speed and being easy to fall into local optimum in itself. In this paper, an improved BP neural network is proposed aiming at the deficiency of BP algorithm and obtaining features of face images through Gabor filter, and then reducing dimensions of vectors by the method of chaos algorithm and improved weight analysis. Simulation experiments show that improve algorithm in this paper has a relatively high recognition rate in face recognition.