摘要
棉纺工业中,异性纤维的有效分离是一项重要而又困难的工作。提出基于多尺度小波和模糊方法的棉花异纤检测算法。首先对图像进行多尺度小波变换,旨在检测各类异性纤维的线条边缘;然后把线条边缘的尺寸映射到隶属度空间,对隶属度求和,通过与阈值比较完成线条识别,既分离了异纤,又抑制了棉花图像纹理的干扰。实践证明,所提出的算法解决了异纤漏识别和棉花误识别两大难题,以快速、高效、鲁棒性好的特点,满足了工业的需要。
Efficient separation of cotton foreign fiber is an important and difffcult task in spinning industry.Cotton foreign fiber detection arithmetic based on multi-scale wavelet transform and fuzzy methods are proposed.Firstly,the image is executed through multi-scale wavelet transform in order to detect the line edge of foreign fiber.Then the size of line edge is mapped to membership space.The line edge recognition is carried out by comparing the sum of all memberships with threshold value.This method can restrain the interference of cotton image texture.The practice shows that this arithmetic solves the foreign fiber missing recognition and cotton mistaken recognition,and meets the need of industry,with fast,efficient and more robust features.
出处
《控制工程》
CSCD
北大核心
2009年第S1期173-175,179,共4页
Control Engineering of China
关键词
异性纤维
小波变换
多尺度
线条检测
隶属函数
foreign fiber
wavelet transform
multi-scale
line detection membership function