Image Deblurring via Enhanced Low-Rank Prior

ABSTRACT:

Low-rank matrix approximation has been successfully applied to numerous vision problems in recent years. In this paper, we propose a novel low-rank prior for blind image deblurring. Our key observation is that directly applying a simple low-rank model to a blurry input image significantly reduces the blur even without using any kernel information, while preserving important edge information. The same model can be used to reduce blur in the gradient map of a blurry input. Based on these properties, we introduce an enhanced prior for image deblurring by combining the low rank prior of similar patches from both the blurry image and its gradient map. We employ a weighted nuclear norm minimization method to further enhance the effectiveness of low-rank prior for image deblurring, by retaining the dominant edges and eliminating fine texture and slight edges in intermediate images, allowing for better kernel estimation. In addition, we evaluate the proposed enhanced low-rank prior for both the uniform and the non-uniform deblurring. Quantitative and qualitative experimental evaluations demonstrate that the proposed algorithm performs favorably against the state-of-the-art deblurring methods.

PROJECT OUTPUT VIDEO:

EXISTING SYSTEM:

• Fergus et al. exploit a mixture of Gaussians to fit the distribution of natural image gradients and the blur kernel estimation is obtained by variational Bayes inference.

• Shan et al. introduce a method by concatenating two piece-wise continuous functions to fit the logarithmic gradient distribution of natural images and use that for image deblurring.

• Levin et al. model the latent images by a hyper-Laplacian prior and develop an efficient marginal approximation method to estimate blur kernels.

DISADVANTAGES OF EXISTING SYSTEM:

• Query in the external dataset is computationally expensive.

• They do not carry substantially useful information for kernel estimation and often generate hallucinated high frequency contents, which complicate the subsequent kernel estimation steps

PROPOSED SYSTEM:

• In this paper, we propose a novel enhanced low rank prior for blind image deblurring. We exploit the low rank properties of both intensity and gradient maps from image patches. To regularize the solution space of latent images, we formulate the problem as a weighted nuclear norm minimization task based on low rank properties. In addition, we extend our proposed low rank prior for non-uniform image deblurring. Experimental results on two benchmark datasets demonstrate that the proposed algorithm based on the enhanced low rank prior performs favorably against the state-of-the-art deblurring methods.

The contributions of this work are summarized as follows:

• We analyze the effect of low rank matrix approximation on blind image deblurring and propose a novel algorithm using low rank prior.

• We exploit the low rank properties of both intensity and gradient maps from an image to recover the intermediate image for kernel estimation. In the proposed low rank prior, we develop a method based on weighted nuclear norm minimization to further enhance the effectiveness of low rank properties by eliminating fine texture details and small edges while preserving dominant edges.

• We extend the proposed algorithm based on low rank properties for non-uniform image deblurring caused by camera rotation.

ADVANTAGES OF PROPOSED SYSTEM:

• Performance is high
• The low rank properties of both intensity and gradient maps from image patches are exploited in the proposed algorithm.
• Proposed algorithm performs favorably against the state-of-the-art Deblurring methods.

SYSTEM REQUIREMENTS:

HARDWARE REQUIREMENTS:

• System          : Pentium Dual Core.

• Hard Disk      : 120 GB.

• Monitor          : 15’’ LED

• Input Devices : Keyboard, Mouse

• Ram                : 1GB.

SOFTWARE REQUIREMENTS:

• Operating system : Windows 7.

• Coding Language : MATLAB

• Tool                      : MATLAB R2013A

REFERENCE:

Wenqi Ren, Xiaochun Cao, Senior Member, IEEE, Jinshan Pan, Xiaojie Guo, Member, IEEE, Wangmeng Zuo, Senior Member, IEEE, and Ming-Hsuan Yang, Senior Member, IEEE, “Image Deblurring via Enhanced Low-Rank Prior”, IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 25, NO. 7, JULY 2016.