基于贝叶斯方法的山羊绒与细羊毛的鉴别

An identification method for cashmere and fine wool based on Bayes model

根据细羊毛与山羊绒的鳞片形状与结构特征的不同,提出智能识别2类纤维的方法。通过CCD系统获取2类纤维的灰度图像,采用图像技术将灰度图像处理成单像素宽度的二值图,从二值图中提取描述2类纤维鳞片形状特征的4个比对指标。在样本数据库上基于4个比对指标的统计假设建立辨识细羊毛与山羊绒纤维的贝叶斯分类模型。仿真结果表明:该模型具有较好的纤维鉴别能力,对山羊绒纤维的识别准确度达到83%,对细羊毛则达到90%;并且随着参数的增加,模型有进一步提高鉴别精度的可能。

Cashmere and fine wool have different scale structures, which is a major reference distinguishing them. A method is proposed to distinguish these two type fibers from each other intelligently. After the 8-bit grayscale images of fibers are captured by CCD camera, they are converted to the binary image only having one pixel wide edge using image technology. Then, four primarily comparable parameters describing the scale shape of fibers are extracted from these binary images and the database including them is established. A Bayes classification model, which is used to distinguish cashmere from fine wool, is developed on the basis of statistic hypothesis of four parameters based on the database. The simulation results show that the model is effective in distinguishing cashmere from fine wool, where the classification accuracy rate is 83 percent for cashmere and 90 percent for fine wool, and probably the classification accuracy rate increases with increasing parameters.