为简化自动化设备视觉定位平台的标定流程，减少对标准块的依赖并实现标定自动化，本文提出了基于模板 匹配的主动视觉自标定方法，依靠平台运动时拍摄到的图片，实现任意场景目标的识别，然后建立模型对运动位置进行标定； 给出了视觉自标定的原理、流程和容错方法，适用于多目标的情形；整个过程由粗到精分3 圈进行，根据匹配位置自动计算标 定结果，使标定更加实用化与通用化。所提方法成功应用于LED 平面固晶机中，结果表明该方法精度可达13.947μm，能满相 关自动化设备的视觉标定要求。

To simplify the visual positioning platform calibration process of automation equipment, reduce the reliance on the standard block and realize the automatic calibration, this paper presents a visual self-calibration method based on template matching, relying on the platform motion captured images to recognize the target of any scene, and then builds a mathematical model to calculate the calibration results; Gives the principle, process, and fault tolerance method of self-calibration, suitable for multi-objective situation. The whole process calculates the calibration results from coarse to fine points with 3 laps automatically, making calibration more practical and generalization. The proposed method is successfully applied to the LED die bonder. The results show that positioning accuracy can reach 13.947 um based on the method, which can meet the calibration requirements of the relevant equipment.