多目标聚集状态下的羊绒纤维自动识别

Automatic Recognition on Cashmere Fiber in the Multi-objective Aggregation State

利用反射式暗视场组合照明装置直接采集高放大倍率下羊绒和羊毛纤维集合体表面纤维图像,通过图像处理的方法整体提取同一图像内多根羊绒和羊毛纤维的4个表面特征参数,利用贝叶斯分类模型进行统计,得出纤维集合体中羊毛和羊绒的比例。实验结果表明鉴别精度达到87.5%。

The high magnification images of the surface on the fiber aggregation of cashmere and wool fiber was collected by using a reflection-type dark-field lighting unit fixture, and 4 surface characteristic parameters of the fibers by the same image were integrally extracted by using image processing methods, then the proportion of wool and cashmere was calculated by the use of Bayesian statisti- cal classification model. The results showed that the identification accuracy is up to 87.5%.