基于机器视觉的毛纱管聚类方法

Clustering method for bobbin based on machine vision

根据纺纱厂毛纱管分类的实际需要,采用机器视觉处理方式,对摄像头实时捕获的每帧毛纱管图像进行HSV颜色空间转换、滤波降噪等预处理。依据色调及饱和度2个颜色指标对毛纱管颜色信息进行提取和分析,与毛纱管形状、面积等信息一起作为分类的重要特征,并基于SVM算法设计一种能将毛纱管按照残纱含量、颜色等信息进行聚类的方法,为毛纺企业的残纱管理及纱管管理提供自动化的方式,降低人工成本,提高企业生产管理水平。

In compliance with the practical requirement for bobbin, a method of color clustering was developed, in which the machine vision was adopted to capture the real-time bobbin＇s figure, and then it was reprocessed by HSV system and filtering noise reduction, next abstracted and analyzed according to the hue and saturation of the figure. The parameters of bobbin＇s figure, as well as the shape and the area of bobbin, were the essential characters for sorting. Consequently, a method for bobbin sorting was developed on the basis of SVM algorithm. This method would supply an automatic way to manage the bobbins in textile mills, decreasing the labor cost and increasing the level of manufacturing management effectively.