Abstract:Image deblurring has been a challenging ill-posed problem in computer vision. Gaussian blur is a common model for image and signal degradation. The deep learning-based deblurring methods have attracted much attention due to their advantages over the traditional methods relying on hand-designed features. However, the existing deep learning-based deblurring techniques still cannot perform well in restoring the fine details and reconstructing the sharp edges. To address this issue, we have designed an effective end-to-end deep learning-based non-blind image deblurring algorithm. In the proposed method, a multi-stream bottom-top-bottom attention network (MBANet) with the encoder-to-decoder structure is designed to integrate low-level cues and high-level semantic information, which can facilitate extracting image features more effectively and improve the computational efficiency of the network. Moreover, the MBANet adopts a coarse-to-fine multi-scale strategy to process the input images to improve image deblurring performance. Furthermore, the global information-based fusion and reconstruction network is proposed to fuse multi-scale output maps to improve the global spatial information and recurrently refine the output deblurred image. The experiments were done on the public GoPro dataset and the realistic and dynamic scenes (REDS) dataset to evaluate the effectiveness and robustness of the proposed method. The experimental results show that the proposed method generally outperforms some traditional deburring methods and deep learning-based state-of-the-art deblurring methods such as scale-recurrent network (SRN) and denoising prior driven deep neural network (DPDNN) in terms of such quantitative indexes as peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) and human vision.

Video Image Deblurring Algorithm Based on Denoising Engine

Real-time Motion Deblurring Algorithm with Robust Noise Suppression

Ground-to-air Aircraft Infrared Image Deblurring Based on Imaging Degradation Simulation

A Robust Blind Deblurring Method for Motion Images Based on Edge Information

The Improved Deconvolution Video Restoration Technology Based on the Infrared Detection System

Hybrid Regularization Image Deblurring in the Presence of Impulsive Noise

Joint Non-Local Statistical and Wavelet Tight Frame Information-Based l0 Regularization Model for Image Deblurring

Image Deblurring and Super-Resolution by Adaptive Sparse Domain Selection and Adaptive Regularization.

Blind Infrared Remote-Sensing Image Deblurring Algorithm via Edge Composite-Gradient Feature Prior and Detail Maintenance

An Image Alternate Restoration Algorithm Based on Two Blurred/Noisy Images

Robust Two-Step Image Restoration with Adaptive Regularization Parameter

Image Restoration Based On Adaptive Dual-Domain Filtering

Densenet-Based Multi-Scale Recurrent Network For Video Restoration With Gaussian Blur

Temporal-Spatial Consistency and Sampling Neural Network Based Adaptive Hand-held Video Deblurring

Image Deblurring Using Multi-Stream Bottom-Top-Bottom Attention Network and Global Information-Based Fusion and Reconstruction Network

Projection-Based Image Restoration Via Sparse Representation and Nonlocal Regularization

A Video Deblurring Optimization Algorithm Based on Motion Detection

Learning Degradation Representations for Image Deblurring

Image Deblurring Via Combined Total Variation and Framelet

Dynamic Scene Deblurring Using Enhanced Feature Fusion and Multi - Distillation Mechanism.

Learning Deformable Kernels for Image and Video Denoising