# Image Smoothing via Unsupervised Learning

arXiv: https://arxiv.org/abs/1811.02804 · HTML: https://mlsanity.com/paper/1811.02804/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 7 Nov 2018.

## Built on by (most cited first, 12 of 12)

- [Bilevel Fast Scene Adaptation for Low-Light Image Enhancement](https://mlsanity.com/paper/2306.01343/) (2023)
- [Adversarial Imaging Pipelines](https://mlsanity.com/paper/2102.03728/) (2021)
- [Facial Demorphing via Identity Preserving Image Decomposition](https://mlsanity.com/paper/2408.10993/) (2024)
- [LineArt: A Knowledge-guided Training-free High-quality Appearance Transfer for Design Drawing with Diffusion Model](https://mlsanity.com/paper/2412.11519/) (2024)
- [Zero-Reference Lighting Estimation Diffusion Model for Low-Light Image Enhancement](https://mlsanity.com/paper/2403.02879/) (2024)
- [Contrastive Semantic-Guided Image Smoothing Network](https://mlsanity.com/paper/2209.00977/) (2022)
- [Hierarchical Image Peeling: A Flexible Scale-space Filtering Framework](https://mlsanity.com/paper/2104.01534/) (2021)
- [Pyramid Texture Filtering](https://mlsanity.com/paper/2305.06525/) (2023)
- [A Generalized Framework for Edge-preserving and Structure-preserving Image Smoothing](https://mlsanity.com/paper/2107.07058/) (2021)
- [Toward Fast, Flexible, and Robust Low-Light Image Enhancement](https://mlsanity.com/paper/2204.10137/) (2022)
- [Improved Real-time Image Smoothing with Weak Structures Preserved and High-contrast Details Removed](https://mlsanity.com/paper/2307.06298/) (2023)
- [Generative Texture Filtering](https://mlsanity.com/paper/2604.19039/) (2026)

## Abstract (arXiv)

Image smoothing represents a fundamental component of many disparate computer vision and graphics applications. In this paper, we present a unified unsupervised (label-free) learning framework that facilitates generating flexible and high-quality smoothing effects by directly learning from data using deep convolutional neural networks (CNNs). The heart of the design is the training signal as a novel energy function that includes an edge-preserving regularizer which helps maintain important yet potentially vulnerable image structures, and a spatially-adaptive Lp flattening criterion which imposes different forms of regularization onto different image regions for better smoothing quality. We implement a diverse set of image smoothing solutions employing the unified framework targeting various applications such as, image abstraction, pencil sketching, detail enhancement, texture removal and content-aware image manipulation, and obtain results comparable with or better than previous methods. Moreover, our method is extremely fast with a modern GPU (e.g, 200 fps for 1280x720 images). Our codes and model are released in https://github.com/fqnchina/ImageSmoothing.