数据与计算发展前沿2024，Vol.6Issue(5)：111-125,15.DOI:10.11871/jfdc.issn.2096-742X.2024.05.011
基于弹性权重巩固的视频单曝光压缩成像算法研究
Video Snapshot Compressive Imaging Based on Elastic Weight Consolidation
郑巳明 1朱明宇 2袁鑫 2杨小渝1
作者信息
- 1. 西湖大学工学院,浙江杭州 310024||中国科学院计算机网络信息中心,北京 100083
- 2. 中国科学院大学,北京 100049
- 折叠
摘要
Abstract
[Objective]This work aims to design a unified model with high robust hyperparameters,in-cluding compression ratio,modulation mask and measurement resolution,for Snapshot Com-pressive Imaging(SCI).[Methods]We train the proposed model based on Elastic Weight Con-solidation(EWC).The model is uniquely designed by combining Transformer and Convolution-al neural network architectures.Additionally,during the initialization phase,we incorporate Generalized Alternating Projection to enhance the model's robustness to different masks.[Results]Extensive ex-perimental results demonstrate that our proposed unified model can well adapt to different compression ratios,modula-tion masks,and measurement resolutions while achieving state-of-the-art results.Our model outperforms previous SO-TA algorithms in terms of PSNR and SSIM,with an average PSNR improvement of over 5 dB.[Limitations]Al-though our model outperforms previous SOTA algorithms in terms of adaptability and average PSNR,the model with EWC may not perform better than a model specifically trained for a particular single task.[Conclusions]By introduc-ing Generalized Alternating Projection and EWC,as well as the special design of the network structure,our proposed highly adaptive model provides new ideas and methods for solving compressive sensing reconstruction tasks in other complex scenarios.关键词
单曝光压缩成像/高光谱/连续学习/Transformer/3D卷积Key words
snapshot compressive imaging/hyperspectral/continual learning/transformer/3D convolution引用本文复制引用
郑巳明,朱明宇,袁鑫,杨小渝..基于弹性权重巩固的视频单曝光压缩成像算法研究[J].数据与计算发展前沿,2024,6(5):111-125,15.
