| Camera calibration is one of the key technologies in computer vision, which plays animportant role in the three-dimensional reconstruction process and includes determining theinternal camera parameters and internal structure of the principal point, focal length andcamera rotation and translation on external parameters.In this paper, we carried out a binocular camera calibration in-depth research and themain contents include estimated corner detection, stereo matching, basic matrix cameracalibration based on genetic algorithms, etc., specific details are as follows:Firstly we set up a platform containing a binocular camera, optics, and computerhardware, and made a high-precision calibration checkered template; Different calibrationmethods were analysed to adapt to circumstances and conditions. Extracting the corner is thefirst step of the camera calibration.The extraction accuracy will have a huge impact on thecalibration results.We use the Harris corner detection in the paper.Secondly we researched the principle of the camera model involving the basics, then weused a more classic traditional calibration algorithm-Zhang Zhengyou flat checkerboardcalibration method to contain the inside and outside parameters of binocular camera.For theshortcomings of the traditional calibration method constrained by the calibration object, thecamera self-calibration techniques are studied based on the fundamental matrix and the natureof self-calibration technology. We chose the RANSAC te solve the fundamental matrix of thetwo images.Lastly considering two nonzero singular values of the essential matrix should be equal,The paper chose the genetic algorithm optimization and LM algorithm to contain the camerainternal parameters compared to traditional calibration methods and values. Reprojectedcoordinates using the parameters and the actual coordinates were contrasted.Experimentsshow little difference between the two methods to rebuild three-dimensional effect. |
PDF Full Text Request