Learning Collaborative Generation Correction Modules for Blind Image Deblurring and Beyond

Abstract

Blind image deblurring plays a very important role in many vision and multimedia applications. Most existing works tend to introduce complex priors to estimate the sharp image structures for blur kernel estimation. However, it has been verified that directly optimizing these models is challenging and easy to fall into degenerate solutions. Although several experience-based heuristic inference strategies, including trained networks and designed iterations, have been developed, it is still hard to obtain theoretically guaranteed accurate solutions. In this work, a collaborative learning framework is established to address the above issues. Specifically, we first design two modules, named Generator and Corrector, to extract the intrinsic image structures from the data-driven and knowledgebased perspectives, respectively. By introducing a collaborative methodology to cascade these modules, we can strictly prove the convergence of our image propagations to a deblurring-related optimal solution. As a nontrivial byproduct, we also apply the proposed method to address other related tasks, such as image interpolation and edge-preserved smoothing. Plenty of experiments demonstrate that our method can outperform the state-of-the-art approaches on both synthetic and real datasets.

Underexposed Image Correction via Hybrid Priors Navigated Deep Propagation

“…Especially like Convolutional Neural Networks (CNNs), which has been demonstrated that CNNs can learn realistic natural image distributions from a number of images [18]. Thus, several approaches have been proposed to apply the implicit CNN priors for low-level vision tasks, such as super-resolution ( [19], [20]), deconvolution ( [21], [22]), dehazing ( [23]), and others ( [24], [25], [26]). However, since there exist highly coupled variables and complex constraints, which cause that both synthetic and realworld datasets are all obtained difficult, thus it is extremely challenging to apply CNNs to inference the decomposition models in Eq.…”

RFID multi-tag dynamic detection measurement based on conditional generative adversarial nets and YOLOv3

“…Its purpose is to restore potentially tidy image from blurry image, which is a morbid inverse problem. Many scholars have improved the image quality by regarding blur and potential sharp image as prior information (Liu et al, 2018; Wang et al, 2011), including regularization intensity prior (Dong et al, 2011; Tang et al, 2019), total variation (TV) (Jidesh and Banothu, 2017; Osher et al, 2005), mathematically driven discriminate prior (Li et al, 2018), sparse image prior (Pan et al, 2013), and so forth. Although these improve the image quality, estimated blur kernel may cause image visual artifacts.…”

Multi-Scale Grid Network for Image Deblurring With High-Frequency Guidance

“…Zhang et al [12] trained a set of CNN denoisers and integrated them into the model-based optimization method as a prior. Liu et al [13] designed two CNN modules, named Generator and Corrector, to extract the intrinsic image structures from the data-driven and knowledge-based perspectives, respectively. However, it is difficult for these methods to remove nonuniform dynamic blurs, and they are computationally inefficient due to the complex optimization process.…”