梳棉棉网中棉结的在线视觉识别

On-line Nep Visual Recognition in Carding Web

针对国内棉纺织企业采用目光检测法检测半成品中棉结含量存在检测精度低、速度慢等问题,应用计算机视觉技术,设计棉结在线识别系统.采样得到的棉网图像,经图像预处理消除光照不匀和噪声等因素后,采用一维最大熵法结合初始分割阈值和面积阈值的方法来识别棉结,并将识别结果与人工目测结果进行对比.研究表明,本文的识别方法具有识别精度高、速度快、误检率和漏检率低等特点,符合在线检测连续性要求.

Neps from semi-product are usually detected by human visual with low detecting accuracy anti speed in many cotton spinning factories. In order to solve this problem, an on-line nep recognition system using computer vision technology is proposed. After image preprocessing for clearing illumination uneveness and image noises, neps in carding web images can be recognised using the method of maximum entropy of single dimension with an initial segmentation and square threshold value. Comparing with human visual detection, the detection method is satisfied with the demand of on-line detecting continuity and has the features of high recognition accuracy, high speed, low misdetection and omission rate.