单一视角下自适应阈值法的纱线毛羽识别及其应用

Identification and application of yarn hairiness using adaptive threshold method under single vision

为更准确测量纱线参数信息,针对图像背景处理和阈值分割算法对纱线图像处理后毛羽信息损失严重的问题,提出自适应灰度增强及线形区域阈值分割算法。并用自制图像采集系统获取6种不同类型的纱线样本,进行图像识别算法的准确性和有效性验证。结果表明:提出的2种算法可明显减少纱线图像信息损失,并且具有良好的鲁棒性,图像法检测的纱线毛羽长度和数量与目测法相近;实现了纱线主体与背景的灰度对比度增强,避免单一阈值导致的图像分割效果差的影响,提高纱线毛羽的识别精度和测量准确性,为后续研究纱线毛羽检测系统提供有效纱线图像分析算法。

In order to measure the yarn parameter information more accurately, the image grayscale enhancement algorithm and the linear region threshold segmentation algorithm were proposed to solve the serious loss of hairiness information after yarn image processing with the image background processing and the threshold segmentation algorithm. Using the self-built image acquisition system, six different types of yarn samples were acquired, and then the accuracy and validity of the image recognition algorithm was verified. Experimental results show that the proposed two algorithms can significantly reduce the loss of yarn image information and have good robustness. The length and number of yarn hairiness detected by the image processing method are similar to those of the visual inspection method. The grayscale contrast of the yarn and yarn image background is enhanced, and the effect of poor image segmentation due to a single threshold is avoided, thereby improving the recognition accuracy and measurement accuracy of the yarn hairiness. The research results provide an effective yarn image analysis algorithm for the subsequent development of a commercial yarn hairiness detection system.