Multi-View Correction Camera Calibration Algorithm

Camera calibration is the key step of recovering 3D shape from 2D images and the classical problem of photogrammetry and computer vision. Calibration results will determine the outcome of three-dimensional reconstruction directly .Therefore, the research of camera calibration algorithm has an important theoretical significance and practical value.According to calibration methods, calibration camera can be divided into three categories: traditional calibration method, method based on active vision calibration and self-calibration method. Traditional calibration method is using the known structure information of calibration object to calibrate, that is by computing the corresponding relationships between the 3D points and 2D image points; The method based on Active Vision camera calibration is some of the known movement information of camera; Self-calibration method doesn't need the calibrated object, it achieves calibration by using the corresponding points of multiple images.Camera model can be divided into the linear model and the non-linear model, there are lens distortion in the non-linear model. Four reference coordinates are used in the process of camera imagining,they are the basis of quantitative description of the camera imaging process. The various camera calibration techniques generate in the development of computer vision. On the basic of Summarizing and Analysing these techniques, the paper introduces a new highly efficient calibration algorithm—Multi-View correction camera calibration algorithm.A regular black-and-white checkerboard is used as calibration object in the algorithm, the features extraction with subpixel accuracy from various views of a regular black-and-white checkerboard, the internal and external parameters of camera are estimated by mapping these feature sets into the corresponding points of the undistorted and rectified image that would be generated by an ideal pinhole digital camera with the same focal length as the camera to calibrate but devoid of lens distortion and perspective warp. The Multi-View correction algorithm is formulated as a nonlinear least-squares optimization problem where a quadratic cost function expressing the residual registration error has to be minimized. The Multi-View correction algorithm's calculation accuracy has been tested with images taken with a camera with lens distortion of a group of different and synthetic versions of these images. Finally, the comparing results, that obtained the Projection error with Bouguet's images and obtained the remaining registration error with synthetic versions of these images, confirms the high accuracy and robust of the algorithm.During the Multi-View correction camera calibration algorithm, the concept of estimating all the internal and external parameters through image rectification is unprecedented. At the same time,the external parameters of each view and the focal length of the camera are estimated from the relative homographs coefficients once the pixel aspect ratio has been computed as the zero-crossing of the 3D rotation angles. This result represents an attractive alternative in the field of computer vision.