【摘要】 针对汽车变速箱齿轮生产过程中采用传统方法测量几何参数时自动化水平差、检测效率低的问题，开发了一种基于视觉原理的汽车变速箱齿轮几何参数测量系统，提高了检测效率与自动化程度。首先，通过图像采集平台获取待测图像，并通过图像预处理剥离齿轮图像与背景信息；然后，面对测量过程中齿轮复杂曲面的光照干扰问题，提出了一种改进的卡尺工具测量法，基于Deriche算法对齿廓进行亚像素级边缘提取，并利用卡尺工具获取边缘点信息，并将其输入到Huber算法中进行边缘拟合，计算得到齿轮几何参数；最后，搭建了汽车变速箱齿轮几何参数测量系统，使用C#编程显示测量结果，通过将测量结果与影像测量仪测量结果进行实验对比可知，汽车变速箱齿轮几何参数测量系统测量精度可达0.02 mm,满足现场精度要求，平均检测时间为2.86 s,相比人工方法，检测效率和自动化程度均得到提高，符合实际生产现场5秒/件的节拍需求，适用于生产现场的快速检测。更多还原

【Abstract】 To address the issues of low automation and poor efficiency associated with traditional methods for measuring geometric parameters of automotive transmission gears during production, a vision-based measurement system of geometric parameters of automobile transmission gear had been developed. This system significantly enhances detection efficiency and the degree of automation. The process began with an image acquisition platform capturing images of the target gear, followed by image preprocessing to separate the gear from the background. Subsequently, to overcome illumination interference caused by the complex reflective surfaces of the gear during measurement, an improved caliper tool measurement method was proposed. This method utilized the Deriche algorithm for sub-pixel edge extraction of the tooth profile. The caliper tool was then employed to acquire edge point information, which was input into the Huber algorithm for edge fitting, enabling the calculation of the gear′s geometric parameters. Finally, a dedicated measurement system for automotive transmission gear geometric parameters was constructed, with measurement results displayed via programming in C#. Experimental comparisons with results from an optical measuring instrument demonstrate that the proposed system achieves a measurement accuracy of 0.02 mm, meeting on-site precision requirements. The average detection time is 2.86 seconds, representing a significant improvement in both detection efficiency and automation level compared to manual methods. This performance aligns with the practical production cycle demand of 5 seconds per part, making the system highly suitable for rapid on-site detection.更多还原