Integral Normalized Gradient Image A Novel Illumination Insensitive Representation

In this paper, we present a novel illumination insensitive image, called integral normalized gradient image (INIGI), for face recognition. Unlike previous model-based methods, which require training images or have many str constrains for implementation, the proposed representation is simple and generic. Based on the intrinsic and extrinsic factor definition, we firstly normalized the gradient with a smoothed version of input image and then integrate the result into new grayscale image. To avoid unwanted smoothing effects on step edge region, anisotropic diffusion method is introduced. Experiment results on FRGC DB prove that our new approach is very effective in improving verification rate for both holistic and local features.