基于机器视觉的生丝抱合性能检测方法

针对现有生丝抱合性能人工检测精准性差,判断标准无客观量化指标等问题,提出一种基于机器视觉的生丝抱合性能检测方法。首先对采集的生丝图像进行二值化处理、干扰信息去除、图像填充、生丝边缘检测,获得单像素生丝边缘特征。然后通过计算生丝径向上、下边缘像素点的直线距离,得到生丝直径相对变化值,根据生丝直径变化的轴向长度确定开裂区域。最后利用开裂区域大于6 mm时对应的生丝抱合摩擦次数表征生丝抱合性能。结果表明,通过该检测方法测得的200组生丝的直径值与显微镜测得的直径值相比,误差均在5%以内,可满足生丝抱合性能检测需求。

Detection method of cohesive performance of raw silk based on machine vision

Aiming at the problem of poor precision of artificial detection of existing raw silk cohesiveness and no objective quantitative indicators, a method based on machine vision for detecting the cohesive performance of raw silk was proposed. Firstly, the collected raw silk images were subjected to binarization processing, interference information removal, image filling and raw silk edge detection, and single-pixel raw silk edge feature was obtained. Then, by calculating the linear distance between the upper and lower edge points of the raw silk, the relative change of the diameter of the raw silk was obtained, and the cracked area was determined according to the axial length of the change of the diameter of the raw silk. Finally, the cohesive performance of the raw silk was characterized by the times of raw silk cohesion frictions corresponding to the cracked area greater than 6 mm. The experimental results show that the diameter values of the 200 sets of raw silk measured by the detection method are compared with the diameter values measured by the microscope, and the errors are all within 5%, which satisfies the requirement of raw silk cohesion performance detection.