陕西师范大学学报（自然科学版）2025，Vol.53Issue(4)：94-105,12.DOI:10.15983/j.cnki.jsnu.2024320
多帧盲解卷积算法中的迭代策略与参量化方法对比
Comparative study of iterative strategies and parameterization methods in multi-frame blind deconvolution algorithms
摘要
Abstract
Multi-frame blind deconvolution(MFBD)is one of the current mainstream image restoration algorithms.It uses less frames(less than 20 frames)of degraded images to restore the high-resolution images.MFBD uses an iterative optimization method to obtain the optimal estimates of the target by minimizing the cost function.There are two main optimization strategies for the MFBD algorithm,namely the joint iteration strategy and the alternating iteration strategy.Currently,there are few publicly available research reports comparing the advantages and disadvantages of these two strategies.At the same time,the point spread function(PSF)parameterization method can also affect the restoration results.In order to obtain the optimal iteration strategies and parameterization methods,two iterative strategies are adopted with three different PSF parameterization methods based on the MFBD,to conduct comparative analysis on the restoration results under different signal-to-noise ratios and different initial value conditions through the normalized mean squared error(NMSE)and frequency spectrum curve of the result.Simulation experiments have shown that the joint iteration strategy with phase parameterization is able to handle more various complex degraded types.The mean square errors of the restoration results for three types of degraded images are 0.046,0.194 and 0.342,respectively,which is the least compared with the joint iteration strategy with other PSF parameterization.For the alternating iteration strategy,only the gray matrix parameterization can obtain acceptable results(mean square errors of 0.109,0.159,0.332,respectively),but the spectral curve indicates that there is an amplification problem of iterative noise.In summary,joint iteration with phase parameterization can obtain better restoration results and handle more complex degraded situation.关键词
图像复原/多帧图像盲解卷积/联合迭代/交替迭代/PSF参量化Key words
image restoration algorithm/multi-frame blind deconvolution/joint iteration strategy/alternating iteration strategy/PSF parameterization分类
数理科学引用本文复制引用
曾杨骞,梁永辉,刘进,杨慧哲..多帧盲解卷积算法中的迭代策略与参量化方法对比[J].陕西师范大学学报（自然科学版）,2025,53(4):94-105,12.基金项目
国家自然科学基金(62005314) （62005314）
