一种基于变分模型的泊松-高斯噪声图像复原算法-计算机辅助设计与.pdf

第29 卷 第3 期 计算机辅助设计与图形学学报 Vol. 29 No.3 2017 年3 月 Journal of Computer-Aided Design Computer Graphics Mar. 2017 一种基于变分模型的泊松-高斯噪声图像复原算法 赵蓝飞, 王爱丽, 王 博, 吕鑫淼 (哈尔滨理工大学测控技术与仪器黑龙江省高校重点实验室 哈尔滨 150080) (zl163.com) : 为了改善已恢复图像的图像质量, 提出一种基于变分模型的图像复原算法. 首先通过多曝光图像成像算法 重建出对于光照具有线性响应的亮度图像, 并采用泊松-高斯噪声模型对图像退化过程进行建模; 然后通过分析亮度 图像的同态子块, 将泊松-高斯噪声模型的未知参数估计问题转化为求解一个线性方程组; 再根据贝叶斯公式和最大 后验概率准则推导出一个具有严格凸函数性质的变分模型, 其最优解即为对于无噪图像的最优估计; 最后利用最小 二乘法以及最速下降法分别对噪声的参数以及无噪图像进行估计. 实验结果表明, 本文算法能够在抑制噪声的同时 保证图像的细节与图像质量. ：变分模型; 图像复原; 泊松-高斯噪声模型; 相机响应函数; 最速下降法 ：TP391.41 A Variational Model for Image Denoising Corrupted by Poisson-Gaussian Noise Zhao Lanfei, Wang Aili, Wang Bo, and Lyu Xinmiao (The Higher Educational Key Laboratory for Measuring Control Technology and Instrumentations of Heilongjiang Province, Harbin University of Science and Technology, Harbin 150080) Abstract: In order to improve the quality of recovered image, this paper proposes an image denoising algo- rithm based on variational model. Firstly, multiple-exposure imaging method is used to recovery radiance map which represents the linear response of light intensity. Simultaneously, Poisson-Gaussian noise model- ing is introduced for modeling the degraded image. Afterwards, by making an analysis to the homogeneous patches of radiance map, a linear system of equations is derived to estimate the parameters of Poisson- Gaussian noise modeling. Moreover, following Bayes’s rule and maximum a posteriori, a variational model yields to strictly convex function is derived which the optimal solution is the recover image. Eventually, Unknown pa