# SDIP: Self-Reinforcement Deep Image Prior Framework for Image Processing

arXiv: https://arxiv.org/abs/2404.12142 · HTML: https://mlsanity.com/paper/2404.12142/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 45 references that are in the map.
- First posted on arXiv 17 Apr 2024.

## Builds on (most cited first, 19 of 19)

- [Deep Image Prior](https://mlsanity.com/paper/1711.10925/) (2017)
- [Deep Convolutional Neural Network for Inverse Problems in Imaging](https://mlsanity.com/paper/1611.03679/) (2016)
- [Lipschitz regularity of deep neural networks: analysis and efficient estimation](https://mlsanity.com/paper/1805.10965/) (2018)
- [The Little Engine that Could: Regularization by Denoising (RED)](https://mlsanity.com/paper/1611.02862/) (2016)
- [On instabilities of deep learning in image reconstruction - Does AI come at a cost?](https://mlsanity.com/paper/1902.05300/) (2019)
- [DeepRED: Deep Image Prior Powered by RED](https://mlsanity.com/paper/1903.10176/) (2019)
- [On hallucinations in tomographic image reconstruction](https://mlsanity.com/paper/2012.00646/) (2020)
- [Image Restoration using Total Variation Regularized Deep Image Prior](https://mlsanity.com/paper/1810.12864/) (2018)
- [Computed Tomography Reconstruction Using Deep Image Prior and Learned Reconstruction Methods](https://mlsanity.com/paper/2003.04989/) (2020)
- [Denoising and Regularization via Exploiting the Structural Bias of Convolutional Generators](https://mlsanity.com/paper/1910.14634/) (2019)
- [Rethinking Deep Image Prior for Denoising](https://mlsanity.com/paper/2108.12841/) (2021)
- [Lose The Views: Limited Angle CT Reconstruction via Implicit Sinogram Completion](https://mlsanity.com/paper/1711.10388/) (2017)
- [Regularization by architecture: A deep prior approach for inverse problems](https://mlsanity.com/paper/1812.03889/) (2018)
- [Limited-angle CT reconstruction via the L1/L2 minimization](https://mlsanity.com/paper/2006.00601/) (2020)
- [Compressive Spectral Image Reconstruction using Deep Prior and Low-Rank Tensor Representation](https://mlsanity.com/paper/2101.07424/) (2021)
- [Deep Radon Prior: A Fully Unsupervised Framework for Sparse-View CT Reconstruction](https://mlsanity.com/paper/2401.00135/) (2023)
- [Image reconstruction from limited range projections using orthogonal moments](https://mlsanity.com/paper/1403.3021/) (2014)
- [Residual Back Projection With Untrained Neural Networks](https://mlsanity.com/paper/2210.14416/) (2022)
- [A Global Constraint to Improve CT Reconstruction Under Non-Ideal Conditions](https://mlsanity.com/paper/2212.09654/) (2022)

## Built on by (most cited first, 1 of 1)

- [A Zero-Shot Deep Image Prior Framework for Denoising and Deconvolution in Fluorescence Microscopy](https://mlsanity.com/paper/2606.28431/) (2026)

## Abstract (arXiv)

Deep image prior (DIP) proposed in recent research has revealed the inherent trait of convolutional neural networks (CNN) for capturing substantial low-level image statistics priors. This framework efficiently addresses the inverse problems in image processing and has induced extensive applications in various domains. However, as the whole algorithm is initialized randomly, the DIP algorithm often lacks stability. Thus, this method still has space for further improvement. In this paper, we propose the self-reinforcement deep image prior (SDIP) as an improved version of the original DIP. We observed that the changes in the DIP networks' input and output are highly correlated during each iteration. SDIP efficiently utilizes this trait in a reinforcement learning manner, where the current iteration's output is utilized by a steering algorithm to update the network input for the next iteration, guiding the algorithm toward improved results. Experimental results across multiple applications demonstrate that our proposed SDIP framework offers improvement compared to the original DIP method and other state-of-the-art methods.