基于灰度差值的均值滤波算法及其在AXI中的应用  

An Improved Mean Filter Algorithm Based on Difference of Gray Level and Its Application in AXI System

作　　者：王朋[1,2] 吕卫文[1,2] 吴懿平[1,2] 安兵[1,2] 刘骏 

机构地区：[1]华中科技大学连接与电子封装中心,湖北武汉430074 [2]武汉光电国家实验室,湖北武汉430074 [3]无锡日联光电有限公司,江苏无锡214028

出　　处：《电子工艺技术》2012年第3期132-135,151,共5页Electronics Process Technology

摘　　要：在AXI系统中,图像会由于射线散射和量子起伏等因素的干扰而产生噪声,影响成像质量。为了不影响对缺陷进行判断,需要对图像进行降噪处理,同时也要尽量保持图像边缘不发生变化。为达到这一目的,本文提出一种基于灰度差值的均值滤波方法,通过设置一个阈值,计算邻域像素与中心像素灰度值之差的绝对值,若此绝对值大于阈值,则设置该邻域像素的权值为0,否则该邻域像素的权值为阈值与以上绝对值之差。通过对比实验证明该方法可以有效的地去除图像中的噪声,且对图像细节几乎没有损失。In automatic X-ray inspection system, the image may be noised by the interference of various factors such as X-ray scattering, quantum fluctuation etc. For the purpose of not affecting the judgment of defects, it is necessary to remove noise from the image, meanwhile to keep the image edges as clear as before denoising. To achieve this goal, this paper proposed a new method of mean filter algorithm based on gray scale difference. The basic theory is that： Set a threshold value; calculate the absolute value of gray level differences between neighboring pixels and the center pixel. If the absolute value is larger than the threshold, set the weight of the corresponding neighboring pixel to O. Otherwise set the weight of the pixel to the difference of the threshold and the absolute value. By denoising the image in actual production, this method showed good performance in noise reduction and keeping the image edges clear.

关 键 词：自动X射线检测 降噪 均值滤波 灰度差值 阈值 

分 类 号：TN602[电子电信—电路与系统]