摘要
由于织物花纹式样的多样性,传统的图像分割算法常常无法准确地识别出织物边缘轮廓。根据织物悬垂图像的边缘轮廓特点,提出采用基于梯度向量流场(GVF)的动态轮廓模型(Snake模型)来识别织物的边缘轮廓。采用贪心算法实现了Snake边缘检测模型,分别用该模型和基于边缘梯度扫描的算法识别5种典型印花织物悬垂边缘,并将2种算法得到的边缘轮廓进行了比较。实验结果表明,该边缘检测模型能准确地识别多种传统图像分割算法所无法准确识别甚至无法识别的特殊织物的边缘轮廓。
Because of the diversity of printing patterns of fabrics, traditional image segmentation arithmetic usually can not detect the drape contours of fabrics correctly. Therefor, according to features of the contours in fabric drape images, an active contour model（Snakes model）based on the gradient vector flow（GVF） field is proposed to recognize the edges of fabrics. A Snake edge detection model is implemented by greedy arithmetic, and then the drape contours of 5 typical printing fabrics detected by this model are compared with the contours extracted by edge gradient scanning. The result of the experiment shows that this edge detection model is able to precisely detect various drape contours of special printing fabrics, which can not be detected precisely, or even are undetectable, by traditional image segmentation arithmetic.
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2006年第3期8-10,共3页
Journal of Textile Research
基金
国家自然科学基金资助项目(50275139)
浙江省自然科学基金资助项目(01388-G)
关键词
织物悬垂性
图像分割
边缘检测
动态轮廓模型
梯度向量流
fabric drape performance
image segmentation
edge detection
active contour model
gradient vector flow
