为了提高布匹瑕疵图像的滤波效果,针对检测过程中图像易受到多种混合噪声污染的特点,提出一种基于多尺度迭

代思想以改进AMSS 算子。将偏微分方程中的AMSS 方程以有限差分的形式离散化,依据量化尺度参数t 和尺度步长Δt ,采

用小尺度步长对污染图像进行多次迭代滤波。针对改进AMSS 算子的滤波结果,介绍了一种采用水平线算法进行效果分析

的新方法。实验结果表明,和其他算法相比,该滤波算法在滤除受未知混合噪声污染图像干扰的同时,能够更好地保留边缘

信息,从而达到更理想的平滑效果。-Focused on fabric flaw detection , an improved AMSS operator based on multi2scale and iterative

met hod was presented. The AMSS equation was discreted with finite difference , and according to quantita2

tive scale parameter t and scale step lengt h Δt , t he noise in t he object image was iteratively filtered by

small scale step lengt h , t hen aimed at t he filter result of AMSS operator , a new evaluation met hod on level

line was int roduced. The experimental result illust rates t hat t his improved operator can efficiently keep

edge information , and t he performance of filting noise is more effective t han ot her met hods for imagemixed wit h multi2noise.