利用Radon函数变换对木材纹理方向自动检测的研究(英文)

本文提出一种利用计算机自动检测木材纹理方向的新方法。四种Matlab 图像变换函数被尝试用于木材纹理形状的检测。通过比较发现BWMORPH 是最适于检测木材这类中弱纹理的函数,并提取生成了木材纹理骨骼线图像;再对木材纹理骨骼线图像进行Radon变换,得到0°~180°范围内每一角度上的纹理线在投影变换域的积分值,并绘制出纹理线积分值随角度变化的二维曲线图以反映木材纹理角度上的变化规律。进而分析了国内40 个树种的纹理方向曲线图以及它们以针叶、阔叶树材和径向、弦向切面作区别的分类统计规律。结果显示,根据Radon 变换图及其纹理曲线图所反映的木材纹理的方向性规律与人们平常对木材纹理的印象相吻合。这也证明了此种新方法的有效性以及它的应用潜力。图7参6。

Auto detection of wood texture orientation by Radon transform

A novel and efficient approach for detecting wood texture orientation by computer was presented. Four Matlab functions were tried to describe the relative position and orientation of wood texture pixels, to detect texture shape and to create skeletal lines image of wood texture, and BWMORPH function was found the best one. Then by Radon transform, it generated a signature composed of 180 values, each value summing up the size of texture lines that are shaped along that angle, and a two dimensional curve plot was drawn to represent the texture orientation of wood. Furthermore, it analyzed texture orientations of forty species as well as their general statistic laws, classified by softwood, hardwood, radial section and tangential section, and the results showed that texture orientation laws described by Radon trans- form plot and their extracting datum were in accord with the impression of wood texture that people possessed in daily life, which con- firmed the validity of this new approach and their appealing utilization potentials.