Self-calibration of a binocular vision system based on a one-dimensional target

Abstract

This paper proposes a method of self-calibration of a binocular vision system based on a one-dimensional (1D) target. This method only needs a 1D target with two feature points and the distance between the two points is unknown. During the process of computation, the distance value can be set an arbitrary value which is near the actual distance value. Using the method proposed in this paper, we can get the parameters of the binocular vision system including internal parameters of the two cameras and the external parameters (but there exists a non-zero scale factor in the translation vector which is connected to the initial distance value we set), the distortion parameters of cameras and the three-dimensional coordinates of the two points in different positions. In this paper, we determine theoretically that the initial distance value will not influence the results, and also the results of numerical simulation and experiment example are shown to demonstrate the method. Most importantly, this method is insensitive to the initial distance value, and it is the biggest advantage. In a practical application, we can use a 1D target with unknown distance to calibrate the binocular system conveniently; also we can use this method to calibrate the camera in a large field of view with a small 1D target.