An improved nonlocal sparse regularization-based image deblurring via novel similarity criteria

Abstract

Image deblurring is a challenging problem in image processing, which aims to reconstruct an original high-quality image from its blurred measurement caused by various factors, for example, imperfect focusing caused by the imaging system or different depths of scene appearing commonly in our daily photos. Recently, sparse representation whose basic idea is to code an image patch as a linear combination of a few atoms chosen out from an overcomplete dictionary has shown uplifting results in image deblurring. Based on this and another heart-stirring property called nonlocal self-similarity, some researchers have developed nonlocal sparse regularization models to unify the local sparsity and the nonlocal self-similarity into a variational framework for image deblurring. In such models, the similarity evaluation for searching similar image patches is indispensable and influential in deblurring performance. Though the traditional Euclidean distance is generally a choice as a similarity metric, its application might give rise to inferior performance since it fails to capture the intrinsic structure of image patches. Consequently, in this article, based on structural similarity index and principal component analysis, we propose the nonlocal sparse regularization-based image deblurring with novel similarity criteria called structural similarity distance and principal component analysis-subspace Euclidean distance to improve the accuracy of deblurring. The structural similarity index is commonly used for assessing perceptual image quality, and principal component analysis is pervasively used in pattern recognition and dimensionality reduction. In our comprehensive experiments, the nonlocal sparse regularization-based image deblurring with our novel similarity criteria has achieved higher peak signal-to-noise and favorable consistency with subjective vision perception compared with state-of-the-art deblurring algorithms.

Blind Image Deblurring using GLCM and ElasticNet Regularization

“…Table 1 gives the PSNR and SSIM of existing and proposed deblurring method for various digital blur images of CERTH dataset. From table 1 it is observed that, the proposed method gives 16%,14%,11%,7%,5% and 2% high average PSNR compared to recently analyzed methods of Wenqi et al [29], Clemens et al [20], Wang et al [28], Hongyan et al [30], Minghua et al [2], and Dong et al [31] respectively due to high texture detail description and predictor correlation. image, it has been observed that the proposed deblurred image is clearer than the existing methods, as the elastic net is consistent in prediction and feature selection by removing staircase effect.…”

“…A low rank prior to blind image deblurring is more complex in natural images and obtains low accuracy during evaluation [29]. Non local sparse regularization L0 [28] and L0-regularized intensity for gradient prior [2] gives a competitive performance than motion tracking [20] but it provides less identification of kernel and sparsity. Group sparsity [31] based local group clustering provide better results but distortion occurs around the edges.…”

“…The work on elastic net based image deblurring offers a promising solution but is hard to identify the sparsity [30]. Hence the existing methods [29], [20], [28], [30], [2], [31] and proposed methods are experimentally analyzed for image deblurring.…”

“…The algorithm's deblurring effect improved greatly; however, the recovery results were not always ideal, especially for images with rich details [1]. The nonlocal sparse regularization-based image deblurring technique [28] improves the accuracy of deblurring and achieves better peak signal-to-noise but does not consider the high dimensional information. Group sparsity-inducing regularize is based on image gradient on L1 and L2, local group clustering provides better results but the blur kernel estimation process has too many iterations [31] and has artifacts along the edges.…”

Objective Numerical Evaluation of Diffuse, Optically Reconstructed Images Using Structural Similarity Index

“…In addition, SSIM has been repeatedly improved, with several derivative methods developed, such as gradient-based SSIM (GSSIM), three-component weighting region, four-component weighting region, complex wavelet, and an improved SSIM with a sharpness comparison (ISSIM-S) [ 31 , 32 , 33 , 34 , 35 ]. In these advanced implementations, SSIM was transformed to demonstrate reasonable performance in assessing images without a reference and can be used for image decomposition, identifying inter-patch and intra-patch similarities, and deblurring IQA [ 36 , 37 , 38 , 39 ]. SSIM is used widely to evaluate images, including medical images; therefore, this study presents four types of SSIM as a computer-based observer to assess DOT reconstructed images.…”