基于统计量的加权函数图像重建方法

摘要

针对图像重建的问题,提出了一种基于统计量的加权函数图像重建方法.考虑到退化图像不仅含有高斯噪声,且含有拉普拉斯噪声,利用最大似然估计的思想估计高斯噪声和拉普拉斯噪声的方差构造基于统计量的高斯和拉普拉斯权重函数;由于在图像重建过程中,噪声分布发生变化,整合L1,L2范数,设计了一种自适应加权函数;结合双边全变差(BTV)正则化算法,设计了一种自适应加权函数图像恢复方法.实验结果表明:相比基于L1-L2混合误差模型(HEM),方法的峰值信噪比(PSNR)和结构相似度(SSIM)分别平均提高了约2.07 dB,0.02,对含有多种噪声的退化图像能够取得比较理想的结果.%Aiming at image reconstruction problem,an image reconstruction method based on weighting function with statistic is proposed.Regarding as degraded images not only have Gaussian noise but also have Laplacian noise,the idea of the maximum likelihood estimation is adoptcd to estimate the variance of Gaussian noise and Laplacian noise;Because,in reconstruction process,the distribution of noise change,the weighting function of Gussian and Laplace is constructed based on statistic,which integrates L1 and L2 norm,and design an adaptive weighting function.Combine with bilateral total variation (BTV) regularization algorithm,an adaptive image restoration method is designed by weighting function.Experimental results show that in comparision with hybrid errors model(HEM) based on L1-L2 norm,peak signal to noise ratio(PSNR) and structural similarity (SSIM) are increased by an average of about 2.07 dB,0.02,for degraded image with different noise,can achieve more satisfied result.