Abstract: The traditional median filters denoising while tending to alter pixels undisturbed by noise,as well as without meeting the demand of real-time image preprocessing,so a fast adaptive median filter based on crossing windows is proposed in this paper.It makes full use of the relative information of adjacent windows and its low complexity,which result in filtering more quickly,and changes window size according to noise density to enhance noise-reduction effects.The simulation results show that this method can not only reduce more impulse noise,and keep more fine details unchanged but also reduce the complexity of traditional median filter algorithm from O（D²） to O（D）,its result turns out to be more satisfactory.

Key words: crossing window, adaptive median filter, complexity, digital image processing

摘要： 传统的中值滤波存在着在有效去掉脉冲噪声的同时也破坏了图像的细节等缺陷,而且运行速度不能很好地满足实时要求。提出了一种基于十字滑动窗口快速自适应中值滤波算法;十字滑动窗口不仅计算速度较方形窗口快,而且充分利用前一窗口信息,通过两个十字相交向量退化和推进,更加快速;还依据噪声强度来调整窗口大小来提高去噪效果。实验证明该算法在去噪的同时能很好地保护图像细节,并使传统中值滤波复杂度由O（D²）降至O（D）,速度得到大幅度的提高。

关键词: 十字窗口, 自适应中值滤波, 复杂度, 数字图像处理