Recovering facial intrinsic images from a single input

According to Barrow and Tenenbaum's theory, an image can be decomposed into two images: a reflectance image and an illumination image. This midlevel description of images attracts more and more attentions recently owing to its application in computer vision, i.e. facial image processing and face recognition. However, due to its ill-posed characteristics, this decomposition remains difficult. In this paper, we concentrate on a slightly easier problem: given a simple frontal facial image and a learned near infrared image, could we recover its reflectance image? Experiments show that it is feasible and promising. Based on extensive study on hyperspectral images, skin color model and Quotient Image, we proposed a method to derive reflectance images through division operations. That is to divide visual frontal face images by learned near infrared images which are generated by super-resolution in tensor space. With the operation on grey distribution of frontal facial images, the results after division can represent the reflectance of skin, rarely bearing any illumination information. Experimental results show that our method is reasonable and promising in image synthesis, processing and face recognition.