Study On Algorithm Of On-the-job Camera Calibration

Camera calibration has been one of important topics for photogrammetry, vision inspection, computer vision and so on. It has been useful in many practical applications such as mapping, industry controlling automatic navigation and military. Camera calibration provides a quantitative description for the corresponding transformation between 2D information of the vision image and real 3D object world. This paper presents some my researches on subpixel corner detection, camera models and calibration methods that follow in detail.1. Introduce detailly the theory and real procession of the camera imaging. After analyzing all projecting models and imaging relations, this thesis adopts the most applied perspective-imaging model. At the same time, the real-imaging procession and all relations about transforming coordinates are introduced.2. Based on the defect of Harris detection algorithm, a subpixel corner detection method is given. Using iterative method and the corner property that any vector from the true corner to a pixel point in the corner neighborhood is always orthogonal to the gradient vector of the image at the point we obtained corners subpixel coordinates whose precision precedes.0.01 pixel. This solves the problem how to achieve the control points coordinates with subpixel accuracy on camera calibration.3. In this paper the traditional method is the key point and two calibration algorithms are chosen, Tsai algorithm and Zhang algorithm. Based on the two algorithms, propose two improved algorithms below:(1) Linear approach for camera parameter calibration: It only requires a coplanar target without camera's motion. The method is simply and accurate highly. All key parameters of the camera are linearly derived through step-by-step decomposition algorithm. This not only overcomes the instability and iteration of nonlinear problems, but also implements the calibration of partial intrinsic-parameters which the other linear methods fails to do. (2) Improved camera calibration from Zhang's method: It imports tangential distortion, the initial parameters are solved by using the points near the image plane center. As the distortion is very little near the image plane center, the solved initial parameters can be very close to exact ones, and then solve the camera parameters accurately by Levenberg-Marquardt Algorithm.Compared with Tsai algorithm and Zhang algorithm, the proposed methods are more accurate and they work better robustness.