The camera calibration of binocular vision measurement based on OpenCV

This paper has built a binocular stereo vision measurement system by connecting two CMOS cameras to the IPC using digital image acquisition card. This method sets fixed points through taking images from video camera. It uses the camera model of open source computer vision library (OpenCV) to eliminate the effects of radial and tangential distortion, and then the coordinates and proportion relations between actual objects and images are obtained. First of all, taking the image of checkerboard calibration template as the input in Visual Studio 2010 development platform, this method could get the internal and external parameters of camera. Then, the method of re-projection could obtain calibration error. Finally, the calibration error is analyzed to determine if they are practical. Experiments show that: the process of calibration is efficient and fast; the total average error is within 0.2 pixels. So it completely satisfies the requirement of measurement system.