Abstract: A swift filtering algorithm is proposed after a brief analysis of current methods on impulse noise removal.Its main idea is as follows：first，using a critical region to pick up the uncorrupted pixels to build a matrix of signal.Then，treating the signal matrix with eight-direction adding.A restored one will present itself nearly after two-step-correction of after adding image.As a superiority of macro-processing，this algorithm is easy to program and fit for real-time processing.Experimental results indicate that the proposed algorithm works well in denoising images with a normal noise rate and saves time remarkablely.

Key words: image denoising, impulse noise, image correction

摘要： 分析了当前脉冲噪声消噪的主要方法，提出了一种滤除脉冲噪声的快速算法。其主要思想是利用阈值检测法从输入图像中提取出信号点矩阵，然后对信号点矩阵进行八向叠加，最后对叠加后的信号点矩阵进行两步校正得到滤波结果。该算法的一个主要优势就是基于图像矩阵进行整体操作，利于数据的并行处理，编程简单，实时性强。实验结果表明该算法运行快，对普通密度的脉冲噪声滤除效果明显。

关键词: 图像消噪, 脉冲噪声, 图像校正