探讨一种主要针对椒盐噪声的图像去噪算法。该方法首先对图像进行固定块分割(例如8×8)。对于每个分割块进行Huber曲面拟合,由于Huber线性回归对椒盐噪声具有很好的鲁棒性,该算法从而达到图像去噪的目的。实验结果表明,对椒盐噪声图像而言,该算法与传统最小二乘方法相比,去噪效果非常明显。
An image denoising algorithm for salt pepper noise is discussed.The method divides the image into uniformed blocks with size of 8×8.Then each block is fitted using Huber surface regression.Since the Huber linear regression model is very robust against the salt pepper noise,the discussed method can restore the noisy image effectively.Experimental results show that,compared with traditional method based on lest squares principle,the method can achieve better restored image quality for noisy images corrupted by salt pepper noise.