基于局部方差的织物纹理周期测量方法  

A method for measuring fabric texture period based on local variance

作　　者：蔺曾上 潘如如 周建 LIN Zengshang;PAN Ruru;ZHOU Jian(Jiangnan University,Wuxi,214122,China)

机构地区：[1]江南大学,江苏无锡214122

出　　处：《棉纺织技术》2024年第3期61-66,共6页Cotton Textile Technology

摘　　要：针对织物纹理周期不易测量的问题,提出一种基于局部方差的织物纹理周期自动提取方法。首先利用特定窗口遍历图像所有像素点,并计算其窗口内局部方差值;然后对源图像进行缩小后,再重复上一步,进而构建出图像的局部方差曲线图;最后采用自适应谷值寻找算法测量源图像的纹理周期值。对不同类型的织物周期检测的试验结果表明:基于局部方差的织物纹理周期测量方法的相对误差仅为2.6%,相比于累加DMF方法和自相关函数法,其稳定性和准确性更高。认为:该研究所提方法能够准确测量织物的纹理周期。Aiming at the difficulty in the measurement of fabric texture period,method of automatic extracting fabric texture period based on local variance was proposed.Firstly,using specific window to traverse all pixels in the image and calculate the local variance value within the window.Then,after reducing the source image,repeating the previous step to construct local variance curve of the image.Finally,an adaptive valley search algorithm was used to measure the texture period value of the source image.The experimental results of different types of fabric period detection showed that the relative error of the method based on local variance was only 2.6%,which was more stable and accurate than the cumulative DMF method and the auto-correlation function method.It is considered that the method proposed can accurately measure the texture period of fabric.

关 键 词：局部方差 织物纹理 纹理周期测量 织物纹理分类 图像处理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]