Remove The Illumination Algorithm Research For Face Recognition

Since from the face recognition technology appeared, there are already more than30years old. With the technology rapid advancing and changing, more and more people began to attention to face recognition. Howeve,up to now, the illumination change is a hard elimination of factors, investigator gradually proposed the removal of illumination effects algorithm, but the achieved effect is unsatisfactory.This paper proposed a new algorithm based on the logarithmic transformation, the WLFT(Wavelet Logarithm Fragments Transformation) algorithm, this algorithm is based on wavelet decomposition, make the low-frequency component logarithmic transformation. because the logarithmic transformation is just to increase the contrast of the image, not shaded difference between image the luminance and a high light portion is reduced, so that the image after the digital conversion will adjust the brightness of the image through the segmentation function, in order to remove the shadow. From the subjective and objective analysis, it can be seen that, the shaded area on the left side of the face image processing through WLFT algorithm removing the shadow is indeed obvious than the effect the logarithmic transformation processed. This paper proposes FII(Face Invariant Image) algorithms, this algorithm is obtained the face invariant grayscale image from a single image. Unlike the previous used for real scenes, while its application in the face. The algorithm is in the chromaticity space, calculate the the information entropy minimum grayscale, invariant image. Through the analysis of subjective, The grayscale image obtained by the algorithm in addition to the light effects, and there is no loss of image information and the feature of face.