Robust Super-resolution Reconstruction Research Articles

A robust super-resolution reconstruction model of turbulent flow data based on deep learning

A new super-resolution model, namely the turbulence volumetric super-resolution (TVSR) model, is developed based on convolutional neural network (CNN) to reconstruct three-dimensional high-resolution turbulent flow field data from low-resolution data. Direct numerical simulation (DNS) and corresponding filtered DNS (FDNS) data of homogeneous isotropic turbulence at various Reynolds numbers are used to train the TVSR model. The proposed model is a modification of Liu et al. (2020), aiming to provide an improved generalization capability of the super-resolution model. For this purpose, we propose a patchwise training strategy in consideration of the property of turbulence that the velocity correlation between two points diminishes as the separation becomes sufficiently large. Furthermore, data at various Reynolds numbers are combined together to train the model. In comparison with existing models, the present TVSR model shows a better generalization capability in two aspects. First, the TVSR model trained using data at low Reynolds numbers is found robust and accurate in the super-resolution reconstructions of flow fields at higher Reynolds numbers. Second, although only DNS data are used for training, the TVSR model is also robust in reconstructing high-resolution flow fields from low-resolution data obtained from large-eddy simulation (LES). This feature of the TVSR model provides a new access to obtain turbulent motions at unresolved scales in LES studies of turbulent flows.

MAP-MRF-Based Super-Resolution Reconstruction Approach for Coded Aperture Compressive Temporal Imaging

Coded Aperture Compressive Temporal Imaging (CACTI) can afford low-cost temporal super-resolution (SR), but limits are imposed by noise and compression ratio on reconstruction quality. To utilize inter-frame redundant information from multiple observations and sparsity in multi-transform domains, a robust reconstruction approach based on maximum a posteriori probability and Markov random field (MAP-MRF) model for CACTI is proposed. The proposed approach adopts a weighted 3D neighbor system (WNS) and the coordinate descent method to perform joint estimation of model parameters, to achieve the robust super-resolution reconstruction. The proposed multi-reconstruction algorithm considers both total variation (TV) and ℓ 2 , 1 norm in wavelet domain to address the minimization problem for compressive sensing, and solves it using an accelerated generalized alternating projection algorithm. The weighting coefficient for different regularizations and frames is resolved by the motion characteristics of pixels. The proposed approach can provide high visual quality in the foreground and background of a scene simultaneously and enhance the fidelity of the reconstruction results. Simulation results have verified the efficacy of our new optimization framework and the proposed reconstruction approach.

Video Enhancement Based on A Robust Hampel Iterative SRR with a General Observation Model

This paper proposes a novel robust Super Resolution Reconstruction (SRR) framework that can enhance a real complex video sequence and is applicable to any noise models. Although SRR algorithms have received considerable attention within the traditional research community, these algorithms are typically very sensitive to their assumed model of data and noise, which limits their utility. The real noise models that corrupt the measured sequence are unknown; consequently, SRR algorithms using L1 or L2 norm may degrade the image sequence rather than enhance it therefore the robust norm applicable to several noise and data models is desired in SRR algorithms. This paper proposes a SRR framework based on the stochastic regularization technique of Bayesian MAP estimation by minimizing a cost function. In order to tolerate to any noise models, the Hampel norm is used for measuring the difference between the projected estimate of the high-resolution image and each low resolution image, and removing outliers in the data. Tikhonov regularization is used to remove artifacts from the final result and improve the rate of convergence. Moreover, in order to cope with real video sequences and complex motion sequences, this paper proposes a SRR General Observation Model (GOM or a±ne block-based transform) devoted to the case of nonisometric inter-frame motion. In the experimental section, the proposed framework can enhance real complex motion sequences, such as Suzie and Foreman sequence, and con¯rm the effectiveness of our algorithm and demonstrate its superiority to other SRR algorithms based on L1 and L2 norm for several noise models (such as AWGN, Poisson noise, Salt & Pepper noise and Speckle noise) at several noise power.

A Lorentzian Stochastic Estimation for a Robust Iterative Multiframe Super-Resolution Reconstruction with Lorentzian-Tikhonov Regularization

Recently, there has been a great deal of work developing super-resolution reconstruction (SRR) algorithms. While many such algorithms have been proposed, the almost SRR estimations are based on L1 or L2 statistical norm estimation, therefore these SRR algorithms are usually very sensitive to their assumed noise model that limits their utility. The real noise models that corrupt the measure sequence are unknown; consequently, SRR algorithm using L1 or L2 norm may degrade the image sequence rather than enhance it. Therefore, the robust norm applicable to several noise and data models is desired in SRR algorithms. This paper first comprehensively reviews the SRR algorithms in this last decade and addresses their shortcomings, and latter proposes a novel robust SRR algorithm that can be applied on several noise models. The proposed SRR algorithm is based on the stochastic regularization technique of Bayesian MAP estimation by minimizing a cost function. For removing outliers in the data, the Lorentzian error norm is used for measuring the difference between the projected estimate of the high-resolution image and each low-resolution image. Moreover, Tikhonov regularization and Lorentzian-Tikhonov regularization are used to remove artifacts from the final answer and improve the rate of convergence. The experimental results confirm the effectiveness of our method and demonstrate its superiority to other super-resolution methods based on L1 and L2 norms for several noise models such as noiseless, additive white Gaussian noise (AWGN), poisson noise, salt and pepper noise, and speckle noise.