Deep-Learned Regularization and Proximal Operator for Image Compressive Sensing

Abstract

Deep learning has recently been intensively studied in the context of image compressive sensing (CS) to discover and represent complicated image structures. These approaches, however, either suffer from nonflexibility for an arbitrary sampling ratio or lack an explicit deep-learned regularization term. This paper aims to solve the CS reconstruction problem by combining the deep-learned regularization term and proximal operator. We first introduce a regularization term using a carefully designed residual-regressive net, which can measure the distance between a corrupted image and a clean image set and accurately identify to which subspace the corrupted image belongs. We then address a proximal operator with a tailored dilated residual channel attention net, which enables the learned proximal operator to map the distorted image into the clean image set. We adopt an adaptive proximal selection strategy to embed the network into the loop of the CS image reconstruction algorithm. Moreover, a self-ensemble strategy is presented to improve CS recovery performance. We further utilize state evolution to analyze the effectiveness of the designed networks. Extensive experiments also demonstrate that our method can yield superior accurate reconstruction (PSNR gain over 1 dB) compared to other competing approaches while achieving the current state-of-the-art image CS reconstruction performance. The test code is available at https://github.com/zjut-gwl/CSDRCANet.

Meta-TR: Meta-Attention Spatial Compressive Imaging Network With Swin Transformer

“…Conventionally, SCI aims to reconstruct the original highdimensional object x ∈ R n by inputting m (m << n) random measurements y ∈ R m [33]. Mathematically, the imaging process can be described as follow:…”

Robust image compression‐encryption via scrambled block bernoulli sampling with diffusion noise

“…\end{equation}$$In this work, we utilize the proximal momentum gradient descent algorithm to reconstruct the plain image from $$\tilde{y}$$ by alternating between the gradient descent step and the proximal operating step $$$$\begin{align} &v^k=\Phi ^\mathrm{T}(\Phi x^{k}-\tilde{y})+\gamma ^{k-1} v^{k-1}, \end{align}$$$$ $$$$\begin{align} &x^{k+1} = D(x^k-\alpha v^{k}), \end{align}$$$$ $$$$\begin{align} &\gamma ^k = \frac{\varepsilon ^{\mathrm{T}}(D(x^k-\alpha v^k+\varepsilon )-x^{k+1})}{m}, \end{align}$$$$where α is the step size, ε is a standard normal random vector, and D is the proximal operator for regularization term R . This splitting approach is an efficient CS reconstruction framework to exploit the plug‐and‐play prior [31, 32]. We set the initial guess x 0 as zero and step size α as 1.…”

“…where 𝛼 is the step size, 𝜀 is a standard normal random vector, and D is the proximal operator for regularization term R. This splitting approach is an efficient CS reconstruction framework to exploit the plug-and-play prior [31,32]. We set the initial guess x 0 as zero and step size 𝛼 as 1.…”

“…Six 32-bit depth secret keys are generated from the SHA256 result of the input plain image. Therefore, the overall key space size S is (2 32 ) 6 , which means that the proposed scheme is resistant to brute-force attacks.…”

A Non-Local Enhanced Network for Image Restoration

“…We can acquire non-local information in multiple ways, such as employing self non-local modules (based on nonlocal means [13]), global pooling modules (channel-wise and spatial-wise attention), multi-scale inputs, long-range connections and recurrent connections in the network. In IR approaches, for example, many networks employ global pooling to extract long-range dependencies, such as RCAN [23], [24] and RIDNet [25]. In these attention modules, they squeeze the feature map (channel-wise or spatial-wise) and generate attentive weights [26]- [28] based on the whole channel of the feature map or the spatial location of the feature map (which is used as the non-local features).…”