基于最大后验概率准则的红外图像NSCT域去噪方法

提出了具有图像增强效果的基于最大后验概率准则的非下采样Contourlet变换(NSCT,Non-Subsampled Contourlet Transform)域自适应降噪算法。该算法在定图像系数和噪声系数先验为高斯分布的前提下,利用后验概率最大原理计算NSCT系数的萎缩因子。在考虑尺度和方向能量因素的基础上对萎缩因子进行了修正,并用于NSCT系数萎缩过程中。最后,通过逆变换重构出降噪和增强的图像。试验结果表明,本文方法与小波去噪方法相比,性能有明显的提升。

Denoising Algorithm Based on the MAP Rule in NSCT Domain for Infrared Image

A NSCT adaptive denoising algorithm based on maximum a posterior ( MAP) with the effect of image enhancement is presented.On the basis of the assumption that the prior distributions of the original image coefficients and the noise coefficients are both Gaussian, the shrinkage factor for NSCT co-efficients is computed by the rule of MAP.Then, the shrinkage factor is revised by considering the factors of scale level and directional energy, which is used in the shrinking process of NSCT coefficients.Final-ly, the denoised and enhanced image could be reconstructed by inverse transformation.The experimental results show that the method given by this paper is much better in performance compared with the wavelet denoising method.