Advertisement

Simultaneous Depth Recovery and Image Restoration

  • Subhasis Chaudhuri
  • A. N. Rajagopalan

Abstract

Given an image degraded by a linear space-variant (LSV) blur, the problem of restoring the original image is an interesting and challenging task. Space-variant image restoration is a problem of considerable importance in image processing because in realworld situations, the degradations are often space-varying. In comparison to the amount of work done on linear space-invariant image restoration [KT91,ST90], the literature records only a few results on the restoration of images degraded by LSV blurs. In [RH72], Robbins and Huang proposed an inversion procedure for LSV image restoration based on the Mellin transform. Sawchuk [Saw74] converted the spatially varying problem to a spatially invariant one using a suitable coordinate transformation. The approach is applicable to only a special class of LSV degradations that can be transformed into a linear space invariant (LSI) degradation. Frieden [Fri72] developed a restoration formula based on the principle of maximum entropy. In [AJ78], Angel and Jain employ a conjugate gradient descent method for restoration of images degraded by spatially varying PSFs. Trussel et al. propose a method in which the image is partitioned into rectangular regions, and each region is restored using a space-invariant technique, such as the MAP filter TH78a, TH78b] or the modified Landweber filter [TF92]. In [RR81], Schafer et al. present an iterative method for LSV image restoration. In [AS93], Patti et al. apply the reduced order Kalman filter for space-variant image restoration. The approach, however, has been found to be computationally expensive even for a moderate blur size. Ozkan et al. [MS94] propose the use of projections onto convex sets for space-varying image restoration. The method uses a set of deconvolution constraints that allow the use of a different PSF at each pixel. In [SB95], Koch et al. propose a multiple model-based extended Kalman filter for restoration of spatially varying blurred images. Note that in all the above methods, the space-variant blur is assumed to be known.

Keywords

Posterior Distribution Line Field Image Restoration Lena Image Intensity Process 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer Science+Business Media New York 1999

Authors and Affiliations

  • Subhasis Chaudhuri
    • 1
  • A. N. Rajagopalan
    • 1
  1. 1.Department of Electrical EngineeringIndian Institute of TechnologyPowai, BombayIndia

Personalised recommendations