Abstract: A more efficient two-phase salt-and-pepper noise removal method is proposed. In the first phase, adaptive median filter is used to detect noisy pixels. In the second phase, a minimization problem is solved by global Barzilai and Borwein (GBB) method. Through the improvement on edge-preserving regularization in the second phase, no regularization coefficient need to be chosen, and the parameter α of the edge-preserving potential function, φ(t)=|t|^α, can create adaptively. Comparing the results of removing noise, our method is better with much less running time, especially when the noise ratio is high.

Key words: image denoise, salt-and-pepper noise, adaptive median filter, edge-preserving potential function, regularization, GBB method

摘要： 给出了一个更为简单高效的两阶段去除图象椒盐噪声的方法。第一阶段：应用自适应中值滤波方法来检测图像中的噪声点；第二阶段：应用全局的BB方法来求解一个极小化问题。通过对第二阶段中保边正则化方法的改进，不仅不再需要进行正则化参数的选择，而且保边势函数φ(t)=|t|^α中的参数α可以自适应地产生，从而使得这个新的方法不再存在参数的选取问题。与已有方法相比，这一新方法节省了计算时间，特别对于噪声污染严重的图像，新的方法的恢复效果更佳。

关键词: 图像去噪, 椒盐噪声, 自适应中值滤波, 保边势函数, 正则化, GBB方法