Rethinking the CSC Model for Natural Images
This work addresses a specific bottleneck in image processing for researchers and practitioners by improving CSC model efficiency, though it is incremental in nature.
The authors tackled the underperformance of the Convolutional Sparse Coding (CSC) model in tasks like denoising by proposing a novel feed-forward network based on Bayesian insights, achieving state-of-the-art performance with significantly fewer parameters.
Sparse representation with respect to an overcomplete dictionary is often used when regularizing inverse problems in signal and image processing. In recent years, the Convolutional Sparse Coding (CSC) model, in which the dictionary consists of shift-invariant filters, has gained renewed interest. While this model has been successfully used in some image processing problems, it still falls behind traditional patch-based methods on simple tasks such as denoising. In this work we provide new insights regarding the CSC model and its capability to represent natural images, and suggest a Bayesian connection between this model and its patch-based ancestor. Armed with these observations, we suggest a novel feed-forward network that follows an MMSE approximation process to the CSC model, using strided convolutions. The performance of this supervised architecture is shown to be on par with state of the art methods while using much fewer parameters.