基于计算机视觉的烟丝宽度测量方法
Method for Measuring Width of Tobacco Shred Based on Computer Vision
-
摘要: 为了提高烟丝宽度的检测效率,研究了烟丝切口快速识别与宽度测量方法。采用张正友标定法对烟丝图像进行了畸变校正,使用了基于RGB(Red-Green-Blue)颜色空间的RB分量差值阈值分割法,实现了烟丝与背景图像的精确分割,提出了计算效率较高的烟丝切口位置识别算法,通过计算烟丝两侧切口位置的距离,实现了无规则摆放烟丝的宽度测量,并以简单随机抽样为理论依据,计算了该方法的最佳样本容量为30。实验结果表明,该方法的重复性限约为0.07 mm,测量误差约0.007 mm,耗时约是ISO方法的1/8。Abstract: To improve the measuring efficiency of tobacco shred's width, a method for shred's cut edge recognizing and width measuring was developed. Zhang Zhengyou calibration method was adopted to rectify image distortion, and a threshold segmenting method on the basis of the difference between R and B components was used to segment the image of shred from background accurately. An efficient algorithm for recognizing shred's cut edges was proposed, by which the width of tobacco shred placed irregularly was measured via calculating the distance between opposite cut edges. The optimal sample size of the developed method was 30 according to simple random sampling theory. The results of experiments showed that the repeatability limit of the method was about 0.07 mm with the measurement error of about 0.007 mm, and the time needed by this method was only one-eighth of that by ISO method.
下载: