A novel face coding scheme based on Gabor Wavelet Networks and genetic algorithm

In this paper we present a face coding scheme based on Gabor wavelet networks (GWNs) and genetic algorithm (GA). Because GWNs is insensitive to variations in illumination and geometric deformation, encodes textural and geometrical information simultaneously, it is quite suitable for face coding. Gabor wavelets are found through optimization. The traditional optimization method -Levenberg Marquardt (LM) may get stuck in local minima and must select the initial parameters carefully. We use a genetic algorithm for determining the parameters of GWNs instead of LM method to make the solution stable. The experiments have demonstrated that our scheme is more effective in face coding and reconstruction.