Geometric and shading correction for images of printed materials a uni .ed approach using boundary

Abstract

We present a novel approach that uses boundary interpolation to correct (1) geometric distortion and (2) shading artifacts present in images of printed materials. Unlike existing approaches, our algorithm can simultaneously correct a variety of geometric distortions, including skew, fold distortion, binder curl, and combinations of these. In addition, the same interpolation framework can estimate the intrinsic illumination component of the distorted image to correct shading artifacts.