一种基于机器视觉的自动标定贴屏实现方法

Implementation of Auto-Calibrated Screen Sticking Based on Machine Vision

针对手机贴屏系统中存在显示屏和触摸屏对位不准确问题,提出了一种基于机器视觉的自标定贴屏的实现方法.利用两个固定的单目摄像头对两屏进行图像采集,对两个屏幕的图像信息分别经过小波变换、Hough直线检测等算法处理之后得到两个屏幕的两对贴合点的图像坐标,利用自标定等几何方法计算出贴合点的三维坐标,得到显示屏和触摸屏之间的位置关系.此贴屏算法实验得到的数据精度高,图像误差率限制在2%以内,三维误差率限制在0.02 cm以内,可为手机自动贴屏机的伺服控制机构提供准确的数据支持.

Implementations of an auto-calibrated cellphone screen sticking method based on machine vision is proposed against the inaccurate contraposition between display screen and touch screen. Collect the images of two screen contrapoints using two fixed monocular cameras, obtain their image coordinates of the contrapoints, and define the relation between the locations of the display and the touch screen on the three-dimensional coordinates of the contrapoints by camera auto-calibration. The algorithm produces accurate data with the image error rated below 2%, and the three dimension error limited within 0. 02 cm, a provision of accurate dada support to the servo controllers of the cellphone screen protector machine. Its use helps to reduce the traditional human errors, save human resources and cut production cycles.