摘要
提出了一种新的基于B样条小波变换极大模的织物多组织识别算法:首先由B样条小波对织物图像进行分解,利用分解后的经向和纬向子图像建立正常织物图像和待检测织物图像的极大模边缘图像;然后由它们的差值结果提取特征参数识别织物组织的位置并检验结果的准确性;最后由组织矩阵绘制织物的组织图.仿真结果表明:本文方法能够精细准确刻画织物组织点位置并具有效率高、稳定性能好等优点.
A method for fabric multi-weave pattern detection was presented based on B-spline wavelet transform modulus maximum.Firstly,a fabric image was decomposed by using the spline wavelet transform and two subimages that matched warp and weft texture respectively were formed.And then a local maximum modulus theory based on single-scale edge image was reconstructed from the warp and weft subimages.Secondly,the features were extracted from difference image between two images to detect the locations of fabric multi-weave pattern.Further,the fabric pattern could be obtained with the structural matrix.Simulation results showed that the proposed method was characterized by describing the fabric weave accurate position and had robustness.
出处
《北京服装学院学报（自然科学版）》
CAS
2012年第1期77-82,共6页
Journal of Beijing Institute of Fashion Technology：Natural Science Edition
关键词
B样条小波
小波极大模
织物多组织识别
B-spline wavelet
image decomposition
fabric multi-weave pattern detection
