基于对称分布特性的自适应α-剪枝均值滤波

Adaptive α-trimmed mean filtering based on characteristic of noise symmetric distribution

本文提出了一种新的消除图像中混合噪声的自适应α-剪枝均值方法。该方法基于α-剪枝均值理论和噪声模型的对称特性。与基于渐进方差最小化的自适应α-剪枝均值方法相比,两者在滤波效果上接近,在滤波速度上本文方法提高了近一倍,降低了算法复杂度。对偏离对称分布模型的混合噪声模型,此方法也有很好的鲁棒性。通过实验,其有效性和速度得到验证。

A novel algorithm is proposed to restore the images corrupted by mixed noises based on the α-trimmed mean filtering and the symmetric characteristics of noise model. Compared with the adaptive alpha-trimmed mean filtering method on the basis of asymptotic variance minimization, there is comparable and similar performance between two methods, but algorithm proposed in the paper has the advantages of approximately twice filtering speed and lower algorithmic complexity. Furthermore, it also has better robustness than traditional method for the mixed noise model deviated from symmetric noise distribution, Experimental results validate the effectiveness and speed of the algorithm.