摘要
为抑制重轨弧面强反射场引起的局部亮度过大,同时保留足够多的原始图像信息.研究了线像素灰度处理方法,计算列像素均值,去均值化处理实现纵向光照不均校正;选取上边缘的n个离散点,三次B样条边缘拟合,y方向像素对齐实现边缘矫直,然后对每一行的像素进行叠加,将灰度均值作为基准,根据每一行像素灰度和基准的比值进行线性拉伸,实现横向光照不均校正.引入信息熵、均方误差、峰值信噪比、通用图像质量指数量化比较不同方法的处理效果.对比实验表明,文中方法得到的图像总体灰度值分布最均匀,图像质量评价参数最高.该方法校正了图像亮度的不均匀性,同时能保留足够多的原始图像信息.
To restrain the phenomenon of local excessive brightness caused by the strong reflection from arc of the heavy rail, and retain enough of the original image information, lines-pixel of gray pro- cessing was investigated amply, the average of column-pixel was calculated and subtracted for vertical poor illumination correction, n discrete points were used for cubic B-Spline edge fitting, y directional pixel was alignment to realize the edge straighten out, and then, every raw-pixel was added, with the gray average as benchmark, according to the ratio of every raw-pixel and benchmark to linear stretch, to realize the transverse poor illumination correction, lines-pixel of gray processing was contrasted with the other ways of image processing effects with gray entropy, mean square error（MSE）, peak signal-to-noise ratio（PSNR）, universal image quality index（UIQI）. The experiment indicates that the distribution of image gray value most uniform and image quality evaluation parameters maximum with the lines-pixel of gray processing algorithm. This method amends the image luminance heterogeneity, preserves sufficient original image information.
出处
《武汉理工大学学报（交通科学与工程版）》
2012年第5期950-953,共4页
Journal of Wuhan University of Technology（Transportation Science & Engineering）
基金
国家自然科学基金委员会与中国工程物理研究院联合基金项目资助(批准号:10976034)
关键词
图像处理
热态重轨
光照不均
同态滤波
背景均衡化
线像素灰度处理
image processing
hot heavy rail
poor illumination
homomorphic filtering
background equalization
lines-pixel of gray processing
