摘要
探讨采用数字图像处理的羊毛与羊绒纤维识别效果。首先采集纹理细节和形状轮廓增强的羊毛与羊绒纤维图像,将交叉纤维处理分割成单根纤维。分别提取纤维的形态特征和纹理特征,然后基于支持向量机模型,根据有限样本信息的学习精度和学习能力进行羊毛与羊绒纤维识别。最终,羊毛与羊绒纤维识别正确率达到93.1%。认为:采用数字图像处理提取纤维特征能较好地识别羊毛与羊绒纤维。
The identification of wool and cashmere fiber based on digital image processing was discussed. Firstly, the texture details and contour shape of wool fiber and cashmere fibers images were collected. Decussating fibers were treated and cut into single fibers by treatment. The morphology and textural characteristics were ex- tracted separately. Then, based on support vector machine model, the identification of wool and cashmere was pre- ceded according to the learning accuracy and learning ability for limited samples. Finally, the accuracy rate for wool and cashmere was reached 93.1 %. It is considered that wool and cashmere fiber can be well identified by ex- tracting fiber characteristics with digital image processing.
出处
《棉纺织技术》
CAS
北大核心
2018年第2期1-4,共4页
Cotton Textile Technology
基金
湖北工业大学高层次人才启动基金项目(BSD2012002)
关键词
数字图像处理
纤维识别
交叉纤维
鳞片
灰度共生矩
支持向量机
Digital Image Processing, Fiber Identification, Decussating Fiber, Scale, Gray Level Co-occur-rence Matrix, Support Vector Machine