基于机器视觉的薄片精密叠装技术

Slice Precision Stacking Technology Based on Machine Vision

针对薄片人工叠装效率低、一致性差的问题,提出一种基于机器视觉的动态环境下精密叠装技术。首先,对原始图像采用降采样、二值化、形态学等相结合的目标识别算法进行预处理以消除动态环境下与目标无关的干扰信息,增强目标特征的可检测性。其次,设计的位姿检测算法采用辐射线搜索法将薄片外圆离散化为360份进行边缘点采样以定位缺口,实现正反方向识别和位姿计算。最后,设计一种靶片标定方法以实现系统标定,通过粗调和精调姿态的步骤,引导机器人将薄片精确叠放在专用定位模具上,从而实现精密叠装。实际测试显示设计系统叠装成功率100%、精度达到0.02mm。

Aiming at the problems of low efficiency and poor consistency of stacking,a precision stacking technology based on machine vision in dynamic environment was proposed.Firstly,the original image was preprocessed by the target recognition algorithm combining down-sampling,binarization and morphology to eliminate the interference information unrelated to the target in the dynamic environment and enhance the detectability of the target features.Secondly,the posture detection algorithm was designed by using the radial search method to discretize the outer circle into 360 parts and sample the edge points to locate the notch,which realized the positive and negative direction recognition and posture calculation.Lastly,a target calibration method was designed to achieve system calibration.Through coarse and fine adjustments of the attitude,the robot was guided to stack the slice precisely onto a dedicated positioning mold,thereby realizing precise stacking.Practical tests show that the designed system achieves a 100%stacking success rate with a precision of up to 0.02 mm.