Proc. IEEE International Conference on Image Processing, Sep. 11-14, 2011, Brussels, Belgium.

Patch-based Image Deconvolution Via Joint Modeling Of Sparse Priors

Chao Jia and Brian L. Evans

Department of Electrical and Computer Engineering, Engineering Science Building, The University of Texas at Austin, Austin, TX 78712 USA
kurtjc@gmail.com - bevans@ece.utexas.edu

Paper - Slides - Software

The software is available in a rar format, which can be decompressed using 7-zip or other similar software.

Abstract

Image deconvolution aims to recover an image that has been degraded by a linear operation such as blurring during image acquisition. Deconvolution based on maximum a-posteriori (MAP) estimation requires the global prior probability of the original image. Conventional methods usually model the image priors by uniformly characterizing the statistical properties of either some forward measurements of images or the representation coefficients in frames, neglecting the local image statistics. In this paper, we adopt local sparse representation in image deconvolution. Our contributions include proposing
  1. a joint model of natural images combining sparse representation of image patches and sparse gradient priors, and
  2. an efficient iterative algorithm to infer the MAP estimate of image deconvolution using the proposed model.
Experiments indicate that the proposed method can recover the original image with high peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) index compared with state-of-the-art methods.

Figure 1: Visual comparison of House image in test setting 1.

original blurred

L0-ABS proposed

COPYRIGHT NOTICE: All the documents on this server have been submitted by their authors to scholarly journals or conferences as indicated, for the purpose of non-commercial dissemination of scientific work. The manuscripts are put on-line to facilitate this purpose. These manuscripts are copyrighted by the authors or the journals in which they were published. You may copy a manuscript for scholarly, non-commercial purposes, such as research or instruction, provided that you agree to respect these copyrights.

Last Updated 09/10/11.