The Integrating Resampler and EfficientImage Warping
Ming‐Chao Chiang, Terrance E. Boult · 1996
This paper introduces integrating resamplers as an efficient method for image warping using a class of imaging-consistent reconstruction/restoration algorithms proposed in [Boult and Wolberg-1993]. Examples are given to show the use of the integrating resamplers described herein for two low-level computer vision applications. The first is geometric correction for images degraded by radial lens distortion, demonstrated as a preprocessing step in correlation-based stereo matching. The second is warping-based data fusion for polarization computations in a small-baseline multi-camera systems. These two applications are chosen because they exemplify the kinds of applications for which the integrating resamplers and the imaging-consistent reconstruction/restoration algorithms are designed--problems that use warped intensity values (as opposed to edge structure).