Blind Image Deconvolution Method Research Articles

Self-Supervised Blind Image Deconvolution via Deep Generative Ensemble Learning

Blind image deconvolution (BID) is about recovering a latent image with sharp details from its blurred observation generated by the convolution with an unknown smoothing kernel. Recently, deep generative priors from untrained neural networks (NNs) have emerged as a promising deep learning approach for BID, with the benefit of being free of external training samples. However, existing untrained-NN-based BID methods may suffer from under-deblurring or overfitting. In this paper, we propose an ensemble approach to better exploit the priors from untrained NNs for BID, which aggregates the deblurring results of multiple untrained NNs for improvement. To enjoy both the effectiveness and computational efficiency in ensemble learning, the untrained NNs are designed with a specific shared-base and multi-head architecture. In addition, a kernel-centering layer is proposed for handling the shift ambiguity among different predictions during ensemble, which also improves the robustness of kernel prediction to the setting of the kernel size parameter. Extensive experiments show that the proposed approach noticeably outperforms both exiting dataset-free methods and dataset-based methods.

An adaptive two phase blind image deconvolution algorithm for an iterative regularization model

This paper proposes a blind image deconvolution method which consists of two sequential phases, i.e., blur kernel estimation and image restoration. In the first phase, we adopt the L0-norm of image gradients and total variation (TV) to regularize the latent image and blur kernel, respectively. Then we design an alternating optimization algorithm which jointly incorporates the estimation of intermediately restored image, blur kernel and regularization parameters into account. In the second phase, we propose to take the mixture of L0-norm of image gradients and TV to regularize the latent image, and design an efficient non-blind deconvolution algorithm to achieve the restored image. Experimental results on both a benchmark image dataset and real-world blurred images show that the proposed method can effectively restore image details while suppress noise and ringing artifacts, the result is of high quality which is competitive with some state of the art methods.

Variational EM method for blur estimation using the spike-and-slab image prior

Most state-of-the-art blind image deconvolution methods rely on the Bayesian paradigm to model the deblurring problem and estimate both the blur kernel and latent image. It is customary to model the image in the filter space, where it is supposed to be sparse, and utilize convenient priors to account for this sparsity. In this paper, we propose the use of the spike-and-slab prior together with an efficient variational Expectation Maximization (EM) inference scheme to estimate the blur in the image. The spike-and-slab prior, which constitutes the gold standard in sparse machine learning, selectively shrinks irrelevant variables while mildly regularizing the relevant ones. The proposed variational Expectation Maximization algorithm is more efficient than usual Markov Chain Monte Carlo (MCMC) inference and, also, proves to be more accurate than the standard mean-field variational approximation. Additionally, all the prior model parameters are estimated by the proposed scheme. After blur estimation, a non-blind restoration method is used to obtain the actual estimation of the sharp image. We investigate the behavior of the prior in the experimental section together with a series of experiments with synthetically generated and real blurred images that validate the method's performance in comparison with state-of-the-art blind deconvolution techniques.

Designing a stable feedback control system for blind image deconvolution

Blind image deconvolution is one of the main low-level vision problems with wide applications. Many previous works manually design regularization to simultaneously estimate the latent sharp image and the blur kernel under maximum a posterior framework. However, it has been demonstrated that such joint estimation strategies may lead to the undesired trivial solution. In this paper, we present a novel perspective, using a stable feedback control system, to simulate the latent sharp image propagation. The controller of our system consists of regularization and guidance, which decide the sparsity and sharp features of latent image, respectively. Furthermore, the formational model of blind image is introduced into the feedback process to avoid the image restoration deviating from the stable point. The stability analysis of the system indicates the latent image propagation in blind deconvolution task can be efficiently estimated and controlled by cues and priors. Thus the kernel estimation used for image restoration becomes more precision. Experimental results show that our system is effective on image propagation, and can perform favorably against the state-of-the-art blind image deconvolution methods on different benchmark image sets and special blurred images.

Computational imaging through chromatic aberration corrected simple lenses

Modern camera lenses become increasingly more complex to optimize the light efficiency of the optical system. Recent research has combined single lens optics with complex post-capture correction methods based on computational photography. This study further improves lens design by correcting chromatic aberrations, after which a simple image deconvolution method is sufficient to produce a high-quality image. We initially estimate the point spread function with a blind image deconvolution method. We add Gaussian regularization as kernel prior to improve the accuracy of kernel estimation. A fast non-blind deconvolution method is then performed to recover a clear image. Experimental results show that the proposed method is at par with state-of-the-art deconvolution approaches. We found a better trade-off between lens design and digital image processing than previous work.

Comparision of Blur Detection and Segmentation Techniques

Objectives: Increasing the quality of the captured image by using different blur detection techniques and making the picture into pixels to redefine the image. Distortion identification based image integrity and verity evaluation which organizes natural scene data of image wavelet co-efficient. Methods/Statistical Analysis: To improve image quality, the different blur detection techniques used in this paper are namely blind image de-convolution, two-stage image segmentation, edge sharpness analysis, non-reference NR block, no directional high frequency. According to these techniques and their procedures, it is estimated that blind image de-convolution is best because it reduces the need for future engineering and identifies the blur type for the mixed input of image for various parameters. Findings: Images are taken around many parts and are used to store and show the information which is precise useful. But many periods the quality of the pictures that are captured is not well-intentioned. The blur detection is initiate helpful in the real life applications and are established in the areas of image segmentation, image restoration. The growth of the blur detection practices have improved the various systems to remove the blur or un-focused part from the image which is owed to imperfection of the camera or due to the de-focus of the gesture of the portion, extreme strength of light. This paper suggests the sharpness, quality image that are in out-of-focus areas. Here this paper proposes the blind image de-convolution method is finest to detect the blur image in the numerous aspects of the sections and the parameters. The outcomes of the segmentation and blur detection practices are compared based on the computational time, cost and the advantages and disadvantages that are projected in the practices. The blur image detection procedures used in this paper are Blind image de-convolution, Two stage image segmentation method,, Non-reference (NR) block, Low directional high frequency energy (for motion blur), edge sharpness analysis. Application/Improvements: Blur detection and segmentation techniques is used to eliminate blur from image source and take out the just right quality of the image using techniques that is proposed in this paper. The comparison made in this paper shows that blur detection techniques which has low computational time and Root Mean Square Error that is frequently used to calculate the differences between pixel value of the image.

Fast Bayesian blind deconvolution with Huber Super Gaussian priors

Expectation Maximization (EM) based inference has already proven to be a very powerful tool to solve blind image deconvolution (BID) problems. Unfortunately, three important problems still impede the application of EM in BID: the undesirable saddle points and local minima caused by highly nonconvex priors, the instability around zero of some of the most interesting sparsity promoting priors, and the intrinsic high computational cost of the corresponding BID algorithm. In this paper we first show how Super Gaussian priors can be made numerically tractable around zero by introducing the family of Huber Super Gaussian priors and then present a fast EM based blind deconvolution method formulated in the image space. In the proposed computational approach, image and kernel estimation are performed by using the Alternating Direction Method of Multipliers (ADMM), which allows to exploit the advantages of FFT computation. For highly nonconvex priors, we propose a Smooth ADMM (SADMM) approach to avoid poor BID estimates. Extensive experiments demonstrate that the proposed method significantly outperforms state-of-the-art BID methods in terms of quality of the reconstructions and speed.

Multiframe blind image deconvolution with split Bregman method

In the paper, a multiframe blind image deconvolution method based on total variation and framelet regularizer is proposed. An adapted version of the split Bregman method is proposed to efficiently solve the resulting minimization problems. In each iteration, four sub-problems need to be solved, one of which can be very efficiently and easily solved via fast Fourier transform implementation or closed form solution. Both simulated noisy and blurred frames and real degraded frames are used to verify the effectiveness of the proposed method. Comparative experimental results show that the proposed method can efficiently remove the blur and noises and restore high quality sharp image.

Sparse Bayesian blind image deconvolution with parameter estimation

In this article, we propose a novel blind image deconvolution method developed within the Bayesian framework. We concentrate on the restoration of blurred photographs taken by commercial cameras to show its effectiveness. The proposed method is based on a non-convex l p quasi norm with 0<p<1 that is used for the image, and a total variation (TV) based prior that is utilized for the blur. Bayesian inference is carried out by utilizing bounds for both the image and blur priors using a majorization-minimization principle. Maximum a posteriori estimates of the unknown image, blur and model parameters are calculated. Experimental results (i.e., restorations of more than 30 blurred photographs) are presented to demonstrate the advantage of the proposed method compared to existing ones.

ABC optimized neural network model for image deblurring with its FPGA implementation

Image deblurring is indispensable to many image processing applications. In this paper, we try to improve radiological images degraded during acquisition and processing. An autoregressive moving average (ARMA) model, used for nonlinearly degraded image deconvolution, is identified using a neural network (NN). The NN training is improved using a novel swarm optimization algorithm called Artificial Bees Colony (ABC), inspired from the foraging intelligence of honey bees. The ABC has the advantage of employing fewer control parameters compared to other swarm optimization algorithms. Both estimated image and blur function are identified through this representation. The optimized ARMA-NN model is then implemented on a Xilinx reconfigurable field-programmable gate array (FPGA) using hardware description language: VHDL. This VHDL code is tested on the rapid prototyping platform named ML505 based on a Virtex5-LXT FPGA chip of Xilinx. Simulation results using some test and real images are presented to sustain the applicability of this approach compared to the standard blind image deconvolution (BID) method that maximizes the likelihood using an iterative process. A statistical comparison is concluded based on performance evaluation using seven recent image quality metrics.

Blind deconvolution of quadratic time-frequency representations of musical signals for reverberation feature extraction.

An acoustic space is uniquely characterized by its reverberant behavior. Due to the complexity of the multiple reflections and diffraction of sound in an enclosure, currently available system identification algorithms cannot effectively estimate the impulse response of a concert hall simply from musical recordings in that space. This paper reports the use of blind image deconvolution methods to construct echo patterns from quadratic time-frequency representations of reverberant recordings of music. First, a quadratic time-frequency analysis is performed to decompose the musical signal into its constituent harmonic components. Quadratic time-frequency analysis methods are known to give enhanced resolution in the time-frequency plane in comparison to conventional Fourier analysis. Reverberant features then appear as blur in the time-frequency plane, which can be estimated by employing the methods of blind deconvolution of this “image” of the sound. The proposed algorithm retrieves both the blur pattern, which corresponds to the reverberation echogram, and the direct acoustic signal. By choosing the time-frequency frame scale and smoothing window, a multi-resolution analysis of the underlying acoustic impulse response is obtained. Various quadratic time-frequency analysis methods are evaluated and their relative performance is reported. The proposed methods are also compared to existing dereverberation algorithms.

A New Approach of Blind Image Deconvolution Method without Artifact along the Blurring Paths

A New Approach of Blind Image Deconvolution Method without Artifact along the Blurring Paths

Robust defocus blur identification in the context of blind image quality assessment

A defocus blur metric for use in blind image quality assessment is proposed. Blind image deconvolution methods are used to determine the metric. Existing direct deconvolution methods based on the cepstrum, bicepstrum and on a spectral subtraction technique are compared across 210 images. A variation of the spectral subtraction method, based on a power spectrum surface of revolution, is proposed and is found to compare favourably with existing direct deconvolution methods for defocus blur identification. The method is found to be especially useful when distinguishing between in-focus and out-of-focus images.

Single frame blind image deconvolution by non-negative sparse matrix factorization

Novel approach to single frame multichannel blind image deconvolution has been formulated recently as non-negative matrix factorization problem with sparseness constraints imposed on the unknown mixing vector that accounts for the case of non-sparse source image. Unlike most of the blind image deconvolution algorithms, the novel approach assumed no a priori knowledge about the blurring kernel and original image. Our contributions in this paper are: (i) we have formulated generalized non-negative matrix factorization approach to blind image deconvolution with sparseness constraints imposed on either unknown mixing vector or unknown source image; (ii) the criteria are established to distinguish whether unknown source image was sparse or not as well as to estimate appropriate sparseness constraint from degraded image itself, thus making the proposed approach completely unsupervised; (iii) an extensive experimental performance evaluation of the non-negative matrix factorization algorithm is presented on the images degraded by the blur caused by the photon sieve, out-of-focus blur with sparse and non-sparse images and blur caused by atmospheric turbulence. The algorithm is compared with the state-of-the-art single frame blind image deconvolution algorithms such as blind Richardson–Lucy algorithm and single frame multichannel independent component analysis based algorithm and non-blind image restoration algorithms such as multiplicative algebraic restoration technique and Van-Cittert algorithms. It has been experimentally demonstrated that proposed algorithm outperforms mentioned non-blind and blind image deconvolution methods.

Blind image deconvolution via dispersion minimization

In linear image restoration, the point spread function of the degrading system is assumed known even though this information is usually not available in real applications. As a result, both blur identification and image restoration must be performed from the observed noisy blurred image. This paper presents a computationally simple iterative blind image deconvolution method which is based on non-linear adaptive filtering. The new method is applicable to minimum as well as mixed phase blurs. The noisy blurred image is assumed to be the output of a two-dimensional linear shift-invariant system with an unknown point spread function contaminated by an additive noise. The method passes the noisy blurred image through a two-dimensional finite impulse response adaptive filter whose parameters are updated by minimizing the dispersion. When convergence occurs, the adaptive filter provides an approximate inverse of the point spread function. Moreover, its output is an estimate of the unobserved true image. Experimental results are provided.

A soft double regularization approach to parametric blind image deconvolution

This paper proposes a blind image deconvolution scheme based on soft integration of parametric blur structures. Conventional blind image deconvolution methods encounter a difficult dilemma of either imposing stringent and inflexible preconditions on the problem formulation or experiencing poor restoration results due to lack of information. This paper attempts to address this issue by assessing the relevance of parametric blur information, and incorporating the knowledge into the parametric double regularization (PDR) scheme. The PDR method assumes that the actual blur satisfies up to a certain degree of parametric structure, as there are many well-known parametric blurs in practical applications. Further, it can be tailored flexibly to include other blur types if some prior parametric knowledge of the blur is available. A manifold soft parametric modeling technique is proposed to generate the blur manifolds, and estimate the fuzzy blur structure. The PDR scheme involves the development of the meaningful cost function, the estimation of blur support and structure, and the optimization of the cost function. Experimental results show that it is effective in restoring degraded images under different environments.