Blind Removal of Image Non-Linearities

H. Farid and A.C. Popescu

International Conference on Computer Vision (ICCV), Vancouver, Canada 2001

Paper (ps.gz)    Paper (pdf)    Bibtex

Most imaging devices introduce luminance and geometric non-linearities. For many applications in image processing and computer vision it is advantageous to remove these non-linearities prior to subsequent processing. This paper presents a technique for blindly removing non-linearities in the absence of any calibration information or explicit knowledge of the imaging device. The basic approach exploits the fact that a non-linearity introduces specific higher-order correlations in the frequency domain (beyond second-order). These correlations can be detected using tools from polyspectral analysis. The non-linearities can then be estimated and removed by simply minimizing these correlations