基于正态逆高斯模型和快速双边滤波，本文在四元数小波变换域提出一种新的图像去噪算法。首先将噪声图像进行四元数小波变换分解，其次应用快速双边滤波算法处理低频子带，高频子带采用正态逆高斯模型对其进行建模，并结合最大后验概率估计准则推导出相应的阈值函数，最后结合最优线性插值得到的阈值算法实现图像去噪。对提出的算法进行实验仿真，通过与已有优秀去噪算法比较，结果显示本文方法取得了不错的视觉效果，且在峰值信噪比和平均结构相似性上都得到一定的提高。

This paper proposes a novel image denoising algorithm based on the normal inverse Gaussian model and fast bilateral filtering in the quaternion wavelet transform domain. The quaternion wavelet transform is utilized to decompose a noised image. The fast bilateral filtering algorithm is used to deal with the low frequency sub-band coefficients. The normal inverse Gaussian model for the prior model is used to describe the distributions of the image’s high frequency coefficients. Its corresponding threshold function is then derived using Bayesian maximum a-posteriori probability estimation theory. Finally, an optimal linear interpolation thresholding algorithm is employed to guarantee a gentler thresholding effect. Experimental results show that the proposed method outperforms other existing state-of-the-art denoising methods in terms of peak signal-to-noise and mean structural similarity.