Adaptive logarithm for facial illumination compensation based on homomorphic gomomorphic module
自适应同态对数光照补偿

Face detection has wide applications. But many of the detection algorithms are sensitive to illumination variation. Detection rates usually drop quickly if the illumination is too bright, too dark, or non-uniform illumination. This paper proposed a adaptive logarithm illumination compensation method based neighbour remediation technique, the method use logarithm function as primary function, therefore it's true of human visual trait; and setup the offset's coefficients of logarithm function to solve the illumination problem in varying degree, thus this method is ability of compensation of all types of image in varying degrees through adjusting the coefficients; and introduce a remediation technique based neighbour area to reduce illumination influence and enhance image details, so the resource of the image which transformed has been improved. Experimental results show that this method is effective in compensating facial images of over-lighting and over-shading, and it is beneficial to improve the facial detection rate.