Camera Calibration Method

Camera calibration has been one of important topics for photogrammetry, computer vision and 3D reconstruction 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. It is a premise and base for computer vision obtaining the spatial information of 3D object. Therefore, research on camera calibration method has important theoretical significance and practical value.This thesis presents some main technologies and methods related camera calibration. My researches focus on camera model, calibration method and corner detection. The detailed contents and research results can be summarized as follows:1) Camera calibration theory and method have been studied. Then, the existing corner detection algorithms are analyzed. Based on the defect of Harris detection algorithm, for it only can obtain pixel coordinates and the presence of false detections, an improved corner detection method is introduced. Firstly, using the Harris algorithm obtains the rough corner coordinates, then using the SVD method removes false corners, finally, iteratively using interpolation and Forstner operator obtains the sub-pixel corner coordinates.2) The Characteristics of existing camera imaging geometric models are analyzed, and establishes a new camera model which has extensive applicability and can better meet the actual camera model. It bases on the classical pin-hole model, but it increases three radial and two tangential distortion coefficients.3) Zhang Zhengyou's method about two-step calibration based on plane has been studied. Since it used the self-calibration theory of the absolute conic to solve intrinsic parameters, the process is complicated and tedious. A new method based on planar vanishing points is applied to improve Zhang's method. The imaging model that we established in this thesis is calibrated with this method, and it obtains a more accurate result.4) Finally, the efficiency and feasibility of my research results are verified by large experiments. On one hand, we apply it to the Zhang's data and compare its result with the result from Zhang's calibration method; On the other hand, we carry a real experiment and apply our algorithm to it. At last, we analyze and discuss the results and data in the experiments, and arrive at some helpful conclusions.