一种改进正态逆高斯分布模型的图像去噪算法

Improved image denoising algorithm based on normal inverse Gaussian distribution model

针对传统去噪算法去除含噪声较大的图像时仍有部分噪声残留的问题,基于变换域提出一种改进正态逆高斯分布的图像去噪算法。该算法在非下采样剪切波变换域,利用最优线性插值阈值函数改进正态逆高斯模型作为系数分布模型,对高频子带分解系数进行统计建模,以贝叶斯最大后验概率理论实现图像去噪。实验结果表明,对于添加不同标准差的高斯白噪声图像,该算法在有效保留图像细节和纹理信息的同时在峰值信噪比方面优于同类去噪算法。

Aiming at the problem that the traditional denoising algorithms still exist some residual noise when removing large noise fro,n noisy images, this paper proposed an improved image denoising algorithm hased on normal inverse Gaussian model. The algorithm decomposed image into frequency eoefficients in the non-suhsampled shearlet transtorm, and used the optimal linear threshoht interpolation shrink funetion to improve normal inverse Gaussian model as the coefficient distribution model. Then the algorithm built a statistic model using the high frequency subhand decomposition coefficients, and finally achieved the noise removal using Bayesian maxinmm a posterior probahilily. Experimental results show that corrupted ima-ges with additive Ganssian noise over a wide range of noise varianee. The proposed method can ettectively preserve the image details and texture information, and a state-of-the-art pertormance in terms of peak signal-to-noise ratio.