The improved multi-scale Retinex algorithm and its application in face recognition

In this paper an improved multi-scale Retinex algorithm is studied to solve the insufficiency that the traditional multi-scale Retinex algorithm is incapable of non-uniform illumination images. Before being processed by the multi-scale Retinex algorithm, the given images are preprocessed by a nonlinear transformation to make the illumination of images tends to uniform. Besides, the improved algorithm is combined with PCA and LDA algorithm and apply for face recognition to identify the face images under complex illumination. Simulations show that the improved algorithm has a better ability for various illumination, and it increases the recognition rate under complex illumination.