基于机器视觉的烟丝质量特征在线量化分析

为了实时监控产线烟丝的质量特征,该文研究设计一种新型在线烟丝检测设备,其通过机械结构设计实现自动从产线获取目标烟丝,振筛分离烟丝,测量烟丝,并最终将所测烟丝送回产线。其使用的图像识别算法基于机器视觉,通过基于亮度、色度和浓度(YUV)色彩空间的阈值分割法,精确地分割烟丝与背景,并通过提取烟丝骨架,实现对不重叠烟丝的长度和宽度的测量。通过对比实验表明,算法平均精度误差在5%以内,烟丝宽度误差在0.1 mm以内。所设计的烟丝结构检测仪既能测量单根烟丝的参数,能够统计整批烟丝的结构参数,提供多维度多尺度的信息,并且可以实现在线检测分析,提高测量的效率,为制丝工艺质量的监控及精细调控提供重要基础支撑。

In order to monitor the quality characteristics of the cut tobacco in the production line in real time, this paper studies and designs a new type of online cut tobacco testing equipment, which can automatically obtain the target cut tobacco from the production line through the design of the mechanical structure, separate the cut tobacco by vibrating screen, measure the cut tobacco, and finally measure the cut tobacco. sent back to the production line. The image recognition algorithm it uses is based on machine vision. Through the threshold segmentation method based on the brightness, chroma and concentratin(YUV) color space, the cut tobacco and the background are accurately segmented, and the length and width of the non-overlapping cut tobacco can be measured by extracting the tobacco skeleton. The comparison experiments show that the average accuracy error of the algorithm is within 5%, and the tobacco width error is within 0.1 mm. The designed tobacco shredded structure detector can not only measure the parameters of a single shredded tobacco, but also count the structural parameters of the whole batch of shredded tobacco. The monitoring and fine control of silk process quality provide important basic support.