基于参数核图割的纱线毛羽图像处理方法

摘要

为了准确计算纱线毛羽特征参数,克服图割法中人为随机选择阈值造成的消极影响,提出了一种基于参数核图割的图像处理纱线毛羽算法.首先,采用局部最大类间方差法获取纱线二值图像,并与纱芯二值图像作逻辑运算,提取毛羽进行特征分析.然后,选取两种不同支数的纱线,以USTER CLASSIMAT 5测试结果为标准,比较参数核图割法与区域生长法、线性近似纱芯边界法对纱芯分割图像处理方法的优劣.结果表明,参数核图割法耗时少、误分点少,毛羽检测算法准确率高达95.5%,远优于另两种算法,也证明了选取不同的纱芯分割方法将直接影响毛羽特征分析的准确性.%In order to calculate yarn hairiness characteristic parameter accurately,and overcome the negative impact from selecting threshold artificially in graph cut method,image processing algorithm of yarn hairiness based on parametric kernel graph cut method is proposed.Firstly, the yarn core binary image is acquired via the step of yarn core segmentation by parametric ker-nel graph cut method,and then the logical operation is carried out between the yarn binary im-age and the yarn core binary image to extract hairiness for characteristic analysis.Then,the method is compared with two image processing methods previously used for yarn core segmen-tation in terms of two kinds of yarn samples.The experimental results show that the method a-chieves the least time and minimum number of classification error,the proposed algorithm is more precise and accurate,and reaches up to 95.5%,compared with the other two methods in the standard of test results of USTER CLASSIMAT 5 .Meanwhile,it is proved that the accu-racy of hairiness characteristic analysis is directly influenced by yarn core segmentation method.