Abstract:In this paper we study the topic of signal restoration using complexity regularization, quantifying the compression bit-cost of the signal estimate. While complexity-regularized restoration is an established concept, solid practical methods were suggested only for the Gaussian denoising task, leaving more complicated restoration problems without a generally constructive approach. Here we present practical methods for complexity-regularized restoration of signals, accommodating degradations caused by a known linear operator of an arbitrary form. Our iterative procedure, obtained using the alternating direction method of multipliers (ADMM) approach, addresses the restoration task as a sequence of simpler problems involving L2-regularized estimations and rate-distortion optimizations (considering the squared-error criterion). Further, we replace the rate-distortion optimizations with an arbitrary standardized compression technique and thereby restore the signal by leveraging underlying models designed for compression. Additionally, we propose a shift-invariant complexity regularizer, measuring the bit-cost of all the shifted forms of the estimate, extending our method to use averaging of decompressed outputs gathered from compression of shifted signals. On the theoretical side, we present an analysis of complexity-regularized restoration of a cyclo-stationary Gaussian signal from deterioration by a linear shift-invariant operator and an additive white Gaussian noise. The theory shows that optimal complexity-regularized restoration relies on an elementary restoration filter and compression spreading reconstruction quality unevenly based on the energy distribution of the degradation filter. Nicely, these ideas are realized also in the proposed practical methods. The presented experiments show good results for image deblurring and inpainting using the JPEG2000 and HEVC compression standards.

Cross-frequency spectral prediction for compressed image restoration

Sparsity optimized compressed sensing image recovery

Blind restoration of compressed degraded image based on block effect suppression

Gradient-based Dictionary Optimization for Compressive Spectral Imaging

Reconstruction of compressed spectral imaging based on global structure and spectral correlation

Image Restoration by Estimating Frequency Distribution of Local Patches

Learning-based image restoration for compressed images

Context-Based Lossless Compression of Mosaic Image with Bayer Pattern

Learning-Based Image Restoration for Compressed Image through Neighboring Embedding

Compressed Image Restoration via External-Image Assisted Band Adaptive PCA Model Learning

Image compressive-sensing recovery using structured laplacian sparsity in DCT domain and multi-hypothesis prediction

Hyperspectral Image Compression Via Cross-Channel Contrastive Learning.

Learned Lossless Compression for JPEG via Frequency-Domain Prediction

Enhancing Lossy Compression Through Cross-Field Information for Scientific Applications

High-Accuracy Total Variation for Compressed Video Sensing

End-to-end feature domain residual coding network for multispectral image compression based on interspectral prediction

Exploring Richer and More Accurate Information via Frequency Selection for Image Restoration

Restoration by Compression

Low-Rank Decomposition-Based Restoration of Compressed Images via Adaptive Noise Estimation.

Compression of multispectral image using HEVC

Correlation Estimation from Compressed Images