# A flexible and accurate total variation and cascaded denoisers-based image reconstruction algorithm for hyperspectrally compressed ultrafast photography

**Authors:** Zihan Guo, Jiali Yao, Dalong Qi, Pengpeng Ding, Chengzhi Jin, Ning Xu, Zhiling Zhang, Yunhua Yao, Lianzhong Deng, Zhiyong Wang, Zhenrong Sun, Shian Zhang  
**arXiv:** [2309.02835](https://arxiv.org/abs/2309.02835)  
**Date:** 2023-09-06  
**Summary page:** https://www.summarizepaper.com/en/paper/2309.02835v1/

## Key points

- Hyperspectrally compressed ultrafast photography (HCUP) enables simultaneous temporal and spectral imaging of non-repeatable or difficult-to-repeat transient events in a single exposure
- HCUP achieves an incredibly high frame rate of tens of trillions of frames per second and a sequence depth of several hundred
- The study proposes a flexible image reconstruction algorithm called the total variation (TV) and cascaded denoisers (CD) algorithm for HCUP
- The algorithm leverages TV denoising model cascaded with advanced deep learning-based denoising models within the iterative plug-and-play alternating direction method of multipliers framework
- The proposed TV-CD algorithm significantly improves image reconstruction accuracy and quality of HCUP, capturing fine structures in complex transient scenes
- This advancement has significant implications for various fields, including capturing high-dimensional complex physical, chemical, and biological ultrafast optical scenes
- The refined image reconstruction algorithm based on TV and cascaded denoisers paves the way for practical applications of HCUP in diverse scientific domains.

## Summary

Hyperspectrally compressed ultrafast photography (HCUP) is a groundbreaking technique that enables the simultaneous temporal and spectral imaging of non-repeatable or difficult-to-repeat transient events in a single exposure. It achieves an incredibly high frame rate of tens of trillions of frames per second and a sequence depth of several hundred, revolutionizing single-shot ultrafast optical imaging. To address the poor image reconstruction quality caused by the large sequence depth and the limited fidelities of traditional reconstruction algorithms, this study proposes a flexible image reconstruction algorithm called the total variation (TV) and cascaded denoisers (CD) algorithm for HCUP. The algorithm leverages the TV denoising model cascaded with advanced deep learning-based denoising models within the iterative plug-and-play alternating direction method of multipliers framework. This approach preserves image smoothness while utilizing deep denoising networks to obtain more prior information, effectively solving the common sparsity representation problem in local similarity and motion compensation. Both simulation and experimental results demonstrate that the proposed TV-CD algorithm significantly improves the image reconstruction accuracy and quality of HCUP. It overcomes previous restrictions by capturing fine structures in complex transient scenes. This advancement has significant implications for various fields, including capturing high-dimensional complex physical, chemical, and biological ultrafast optical scenes. In conclusion, this study presents a refined image reconstruction algorithm based on TV and cascaded denoisers for hyperspectrally compressed ultrafast photography. The algorithm's effectiveness in improving image reconstruction accuracy and quality paves the way for practical applications of HCUP in diverse scientific domains.