Camera Calibration Study

Computer vision is a new kind of fields that developed from the base of image processing. It ranges from three-dimension reconstruction and motion analysis to camera calibration, etc. Of which the camera calibration is a very active project. We can carry out three-dimension reconstruction and motion analysis if the camera is correctly calibrated.In this paper, we introduce the relative knowledge and technique around the theory of camera calibration. And the whole paper can be divided into the following four chapters and of which the last three chapters are what the author has done.In the first chapter, we introduce linear camera model and some general knowledge and several traditional methods of camera calibration, such as Direct Linear Transformation and two-step method etc.In the second chapter, we talk about the method for camera self-calibration through translation movement of the camera. We firstly analyze the theory of eigensystem of homograghy matrix between translation visions find the relation between its eigen-subspace and intrinsic parameters of camera. Then we apply the eigen-system theory to camera self-calibration and get two methods for linear calibration. We also give its other applications.In the third chapter, we focus on the way of self-calibration by means of the image of absolute point. We firstly talk about the basic concept and property of absolute point in projective space. Then we build the constrained equations of camera intrinsic parameters by using its property and calibrate the camera. In order to obtain the image of absolute point, we design two kinds of planar model. And they are equilateral-polygon planar model and planar model that containing more than two circles. Another important content in this chapter is that we linearly calibrate the nonlinear parameters of camera based on the rectification features of lens and the fact that projective transform converts a line into a line.In the last chapter, we testify the feasibility of the methods for camera selfcalibration in previous chapters in real application with the experiment of threedimension reconstruction. We experiment with real image applying the calibrated camera parameters and using way of three-dimension reconstruction in threedimensional vision~ Experimental result shows that the methods have some feasibility in real application.