Image Deconvolution Ringing Artifact Detection and Removal via PSF Frequency Analysis

Abstract

Abstract. We present a new method to detect and remove ringing artifacts produced by the deconvolution process in image deblurring techniques. The method takes into account non-invertible frequency components of the blur kernel used in the deconvolution. Efficient Gabor wavelets are produced for each non-invertible frequency and applied on the deblurred image to generate a set of filter responses that reveal existing ringing artifacts. The set of Gabor filters is then employed in a regularization scheme to remove the corresponding artifacts from the deblurred image. The regularization scheme minimizes the responses of the reconstructed image to these Gabor filters through an alternating algorithm in order to suppress the artifacts. As a result of these steps we are able to significantly enhance the quality of the deblurred images produced by deconvolution algorithms. Our numerical evaluations using a ringing artifact metric indicate the effectiveness of the proposed deringing method.

MPTV: Matching Pursuit-Based Total Variation Minimization for Image Deconvolution

“…In the conventional model (5), the strength of the TV regularizer is only controlled by the parameter λ. Although a large weight of the TV regularizer can suppress the potential ringing artifacts, it also over smooths the significant edges and textures in the recovered image [4], [21]. When λ is not large enough, minimizing the TV based objective cannot produce images with sparse gradients and leads to results lying in X with visible ringing.…”

“…In practice, image deconvolution results often suffer from wave-like ringing artifacts, especially in regions near strong edges. This issue may be caused by the unavoidable error in blur kernel estimation [20], the Gibbs phenomenon [48], the zero values in the frequency spectrum of the blur kernel [21] and/or the mismatch between the data and the model (e.g. non-conforming noise, saturation) [49], [50].…”

“…Thirdly, existing TV-based deconvolution methods are usually sensitive to errors or noise in k (i.e. A), and this often causes heavy ringing artifacts in the latent image x [20], [21]. For existing TV methods, it is non-trivial to achieve a balance between suppressing artifacts and preserving image details.…”

Convolutional Deblurring for Natural Imaging

“…This becomes more useful when both blur and image priors (in the blind case) could be fit into one regularization framework to address more complex formulations. The common practice is to use split variable techniques to recast the algorithms in parallel and independently update each sub-modular task [34], [36]- [38], [40]- [45]. 6) Deep Convolutional Neural Network (CNN): This is a variant of deep learning methods in which a convolutional neural network is trained to encode image features in multiple layers of decomposition.…”

“…Trained CNN model to classify blur vs clean as image prior for regularized minimization formulation Simoes [35] 2016 NB • Diagonalizing unknown convolution operator using FFT and solving via ADMM Kim [36] 2015 B • • Encode temporal/spatial coherency of dynamic scene using optical-flow/TV regularized minimization Liu [37] 2014 B • • • Estimate blur from image spectral property and feed into a regularized TV/eigenvalue minimization Mosleh [38] 2014 N • • • Encode ringing artifacts using Gabor wavelets and fit into a regularized minimization for cancelation Pan [39] 2014 N • • • Text image deblurring regularized by sparse encoding of spatial/gradient domains Pan [40] 2013 B • • Estimates the kernel and the deblurred image from a combined sparse regularization framework Kim [41] 2013 B • • • Dynamic image deconvolution using TV/Tikhonov/temporal-sparsity regularized minimization Shen [42] 2012 B • • • TV/Tikhonov regularized minimization for image deconvolution Sroubek [43] 2012 B • • 1-regularized minimization for image deconvolution Dong [44] 2011 NB • • Learn adaptive bases and use in adaptive regularized minimization for sparse reconstruction Zhang [45], [46] 2011 B • • • Sparse regulation of images via KSVD library for deconvolution and apply to facial recognition Bai [47] 2018 B • • Both kernel/image recovered via combined regularization using reweighted graph TV priors Lou [48] 2015 N • • Weighted differences of TV regularizers in 1/ 2 norms and solved by split variable technique Zhang [49] 2014 N • • • Local/non-local similarities defined by TV 1 /TV 2 and regulated by combined minimization Xu [50] 2012 B • Regulate motion by difference of depth map and deconvolve via non-convex TV minimization Chan [51] 2011 N • Deconvolve image/videos using spatial/temporal TV regularization solved by split varying technique Afonso [52] 2010 N • Deconvolve image using TV regularization solved by split varying technique Li [53] 2018 B • • Non-iterative deconvolution via combination of Wiener filters, solution by a system linear equations Bertero [54] 2010 N • Generalized Kullback-Leiblar divergence function to regularize Poisson images Cho [16] 2009 B • • Separate recovery of motion kernel and image from residual image using Tikhonov regularization Wiener [55], [56] 1949 N • Regulate image spectrum in Fourier domain with inverse kernel response Xiao …”