摘要
在自动缫的生丝生产过程中,生丝平均纤度会受到各种工艺条件的影响而发生变化,需要通过对绪下茧粒数进行检测来了解生丝平均纤度。基于形态学方法中的腐蚀膨胀法和K均值算法相结合的方法,改进了K均值聚类算法,解决了蚕茧图像中出现的多个蚕茧相互粘连的问题,实现了对粘连蚕茧图像的分割及蚕茧计数。结果表明:该方法能很好地解决图像反光及蚕茧图像粘连的问题,实现蚕茧图像的分割及正确计数,为绪下茧粒数自动检测技术的研究打下了基础。
In raw silk production process of automatic reeling, the average fineness of raw silk changes due to the influence of various technological conditions. It is necessary to detect cocoon number under thread so as to know the the average fineness of raw silk. This study improves K-means clustering algorithm by combining con＇osion expansion method and K-means algorithm in morphological method, solves the problem of adhesion of multiple silkworm cocoons in silkworm cocoon image and realizes the segmentation of adhesive silkworm cocoon images and cocoon counting. The result shows that this method can solve the problem of image reflection and adhesion of silkworm cocoon images, realize segmentation of silkworm cocoon images and correct counting and lay a foundation for the research on automatic detection technology of cocoon number under thread.
出处
《丝绸》
CAS
北大核心
2014年第1期37-40,49,共5页
Journal of Silk
关键词
缫丝
蚕茧
腐蚀膨胀
K均值
茧粒数
图像识别
silk reeling
silkworm cocoon
corrosion expansion
K-means
cocoon number
image identification
