Preserving Old Memories in Vivid Detail: Human-Interactive Photo Restoration Framework

Seung-Yeon Back,Geonho Son,Dahye Jeong,Eunil Park,Simon S. Woo

2024-10-12

Abstract:Photo restoration technology enables preserving visual memories in photographs. However, physical prints are vulnerable to various forms of deterioration, ranging from physical damage to loss of image quality, etc. While restoration by human experts can improve the quality of outcomes, it often comes at a high price in terms of cost and time for restoration. In this work, we present the AI-based photo restoration framework composed of multiple stages, where each stage is tailored to enhance and restore specific types of photo damage, accelerating and automating the photo restoration process. By integrating these techniques into a unified architecture, our framework aims to offer a one-stop solution for restoring old and deteriorated photographs. Furthermore, we present a novel old photo restoration dataset because we lack a publicly available dataset for our evaluation.

Computer Vision and Pattern Recognition,Artificial Intelligence

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: how to repair and restore old photos efficiently and with high quality through automation and user interaction. Specifically, due to improper long - term preservation, old photos are easily affected in many aspects such as physical damage and image quality degradation. Although human experts can provide high - quality restoration services, their high cost and long time - consuming make it difficult for many users to afford. Therefore, this research proposes an AI - based modular deep - learning framework, aiming to accelerate and automate the restoration process of old photos, reduce the restoration cost, and improve the restoration efficiency.

### Core problems of the paper:
1. **Physical damage of old photos**: such as scratches, tears, etc.
2. **Image quality degradation**: such as blurring, increased noise, etc.
3. **Color fading**: Old photos are usually black - and - white or with color distortion.
4. **High cost and time - consuming**: The cost and time consumption of restoration by human experts are relatively large.

### Solutions:
- **Multi - stage restoration framework**: It includes four main stages - severe damage removal, noise reduction processing, face restoration, and coloring.
- **User interaction**: Allows users to adjust parameters at each stage to achieve personalized restoration.
- **Construction of a new data set**: Created a data set specifically for evaluating old - photo restoration algorithms, solving the problem of insufficient existing public data sets.

Through these methods, this research aims to provide a one - stop solution that can effectively repair and restore old photos while maintaining the visual memory and historical value of the original photos.