基于机器视觉的圆柱齿轮表面微小裂纹检测方法

Detection Method of Micro Cracks on Cylindrical Gear Surface Based on Machine Vision

在温度或湿度的干扰下,对圆柱齿轮表面微小裂纹进行检测时,采集到的目标图像质量很不稳定,导致检测方法的适用性比较差。为了解决这一问题,提出基于机器视觉的圆柱齿轮表面微小裂纹检测方法。根据实际应用需求,选择合适的光学镜头和电荷耦合原件,将上述两种装置组合成机器视觉系统,使用发光二极管作为光源,在两侧照明方式的支持下,使用1394A采集目标图像。将图像分割为包含待测目标的前景区域和背景区域,计算类间方差和类内方差以及两者的比值函数。利用比值函数得到待测目标边缘点,连接边缘点即可得到检测目标。实验结果表明,在不同的温度和湿度环境下,该方法检测时间未超过400 ms,检测精度高于0.9,说明适用性较好。

Under the interference of temperature or humidity,the quality of the collected target image is very unstable in the detection of small cracks on the surface of cylindrical gear,which leads to the poor applicability of the detection method.In order to solve this problem,a method based on machine vision is proposed to detect small cracks on cylindrical gear surface.According to the actual application requirements,appropriate optical lens and charge coupling components are selected to combine the above two devices into a machine vision system.The light emitting diode is used as the light source,and with the support of lighting on both sides,1394A is applied to collect the target image.The image is divided into foreground region and background region containing the target to be measured,and the inter-class variance and intra-class variance as well as their ratio function are calculated.The edge points of the target to be measured are obtained by using the ratio function,and then the detection target can be obtained by connecting the edge points.The experimental results show that in the environment of different temperature and humidity,the detection time of this method is less than 400 ms,and the detection accuracy is higher than 0.9,indicating that this method has good applicability.