摘要
针对目前起毛起球图像的毛球分割方法无法将纹理信息有效消除的问题,提出一种基于小波分解及灰度共生矩阵的织物起球图像分割方法。在获取起球图像后,首先使用Radon变换对其进行倾斜矫正;然后对起球图像进行多尺度小波分解重构,得到各个子图像;对子图像的垂直分量和水平对角分量分别求取其灰度共生矩阵及特征参数,根据特征参数及其得出的数据和曲线图,分析可得毛球的层次图像;最后进行阈值分割得出毛球图像。结果表明,该方法能够有效滤除起毛起球图像的纹理,得到准确的毛球分割图像。
In order to solve the problem that the current pilling image segmentation methods are not universal enough to effectively eliminate the texture information,a new method based on wavelet decomposition and gray level co-occurrence matrix is proposed.After acquiring the pilling image,Radon transform is used to correct its tilt.Then the pilling image is decomposed and reconstructed by multi-scale wavelet transform,and each sub-image is obtained.The gray level co-occurrence matrix and its characteristic parameters are obtained for the vertical and horizontal diagonal components of the sub-image respectively.According to the characteristic parameters and the curves obtained,the hierarchical image of the hair ball can be obtained.Finally,threshold segmentation is used to get the hair ball image.The results show that this method can effectively filter the texture of fuzzing and pilling image,and get an accurate segmentation image of fuzzing and pilling.
作者
周虎
柯薇
王健恺
邓中民
Zhou Hu;Ke Wei;Wang Jiankai;Deng Zhongmin(School of Textile Science and Engineering,Wuhan Textile University,Wuhan,Hubei 430200,China)
出处
《针织工业》
北大核心
2025年第12期67-71,共5页
Knitting Industries
基金
湖北省技术创新专项资助项目(2019AAA005)。
关键词
织物起球
RADON变换
小波分解
灰度共生矩阵
图像分割
Fabric Pilling
Radon Transform
Wavelet Decomposition
Gray Level Co-occurrence Matrix
Image Segmentation
