Estimating Illumination Parameters in Real Space with Application to Image Relighting

A novel algorithm is presented in this paper for estimating the direction and strength of point light with the strength of ambient illumination. Existing approaches estimate these illumination parameters directly in the high dimensional image space, while we estimate the parameters in two steps: first to project the image to an orthogonal linear subspace based on spherical harmonic basis functions; then to calculate the parameters in the low dimensional subspace. The experimental results on CMU PIE database and Yale Database B showed the stability and effectiveness of this method. The resulting illumination information can help to synthesize more realistic relighting images and recognize object under variable illumination.