数据保真项与稀疏约束项相融合的稀疏重建

发布时间：2018-04-19 11:30

本文选题：稀疏重建 + 字典学习　；参考：《光学精密工程》2017年09期

【摘要】：本文针对低光子计数成像过程中产生的泊松高斯混合噪声,提出了一种数据保真项与稀疏约束项相融合的稀疏重建方法。首先,基于泊松高斯噪声相互独立的混合噪声模型,建立了数据保真项与稀疏约束项相融合的稀疏重建目标函数;在图像块聚类的基础上,应用改进贪婪算法实现类内稀疏分解和字典更新;最后,稀疏分解和字典更新交替迭代求解干净图像。针对强烈泊松高斯噪声污染图像的重建实验显示,本文方法与对比方法相比,重建结果的PSNR值平均提升了5.5%,MSSIM值也有明显提升。这些结果表明:本文方法对具有强烈泊松高斯混合噪声的图像有较好的图像复原和噪声去除效果。
[Abstract]:In this paper, a sparse reconstruction method based on the fusion of data fidelity term and sparse constraint term is proposed for Poisson Gao Si mixed noise in low photon counting imaging.First of all, based on Poisson Gao Si noise independent mixed noise model, a sparse reconstruction objective function of data fidelity term and sparse constraint term is established, which is based on image block clustering.The improved greedy algorithm is applied to implement the intra-class sparse decomposition and dictionary updating. Finally, the sparse decomposition and dictionary update iterate alternately to solve the clean image.The experiments on the image reconstruction with strong Poisson Gao Si noise pollution show that compared with the contrast method, the average PSNR value of the reconstruction results is increased by 5.5% and the MSSIM value is also obviously increased.These results show that the proposed method has a good effect on image restoration and noise removal for images with strong Poisson Gao Si mixed noise.
【作者单位】：华南理工大学自动化科学与工程学院;华南理工大学精密电子制造装备教育部研究中心;广州大学机械与电气工程学院;
【基金】：国家自然科学基金资助项目(No.61403146) 广州市科技计划项目(No.201707010054) 中央高校基本科研业务费资助项目(No.2015ZM128)
【分类号】：TP391.41

【相似文献】