Camera Self-calibration And 3d Reconstruction Method

Camera calibration is the essential step of obtaining 3D information from views in the field of computer vision, which is widely used in the area of 3D reconstruction, navigation, visual supervision , etc. Therefore, the associative theoretic research has been put more and more attention in this field. The intrinsic parameters of camera are found from the feature-point -set of views by means of any one of camera self-calibration techniques. And the traditional camera calibration technique is off-line, however the camera self-calibration technique is on-line. So researching and developing excellent camera self-calibration techniques is valuable in theory and practice.Using the existent advanced theory for reference, this paper introduces several kinds of techniques of camera self-calibration with 3D reconstruction, which are based on full information of views under several different camera motion sets and the reasonable optimization criterion. It mainly includes three parts. The first part presents Wu-Hu's linear camera self-calibration and 3D reconstruction technique and our new technique. And the performance comparison of them is also investigated in detail. The second part provides three kinds of camera self-calibration and 3D reconstruction technique under one camera motion set from the matched-point sets on two orthogonal planes. Similarly, the three techniques are compared. The third part gives camera self-calibration and 3D reconstruction technique based on model from single view. At the same time, we make effective improvement to the existent algorithm. As a result, the minimal realization conditions are reduced and the process of calculation is simplified, therefore, the performance of them is better.In this paper, the classical pinhoJe imaging is adopted as the model of camera, which means the matrix of camera intrinsic parameter, has five independent arguments. In Wu-Hu's linear camera self-calibration and 3D reconstruction technique the epipole and fundamental matrix are firstly estimated from the view correspondences, and the homography matrixes of the planes at infinity under two camera motion sets are computed from them. Finally, the camera self-calibration and 3D reconstruction are completed from all the above information. The new technique undertakes the same task just from the views under only one camera motion set. It makes improvement as compared with Wu-Hu's technique, such as: it is more easily done in less time. Furthermore, it is more suitable in the case of on-line applications. The experiments with real views are done to demonstrate the performance comparison of them. The key step of camera self-calibration and 3D reconstruction technique under one camera motion set from the matched-pointsets on two orthogonal planes is calculation of homography matrix of the plane at finite and the corresponding epipole. Then, from them, the homography matrix of the plane at infinity can been solved. Finally, the camera self-calibration and 3D reconstruction are finished very easily. Based on the different methods obtaining epipole, there are three techniques. The results of simulation prove that no apparent difference exists among them. The feature-point set in the single view and their corresponding model location are necessarily known for the developed model- based camera self-calibration and 3D reconstruction technique from single view. It is the simplest technique of all techniques mentioned in this paper. Furthermore, the run-time is the least. It should be point out that the algorithm is new. And it is the generalization of the traditional 3D reconstruction algorithm based on model from calibrated single view.The paper introduces several kinds of technique of camera self-calibration and 3D reconstruction. And it makes deep research on the minimal realization conditions of them. The results of experiments demonstrate that the developed new algorithms are valuable in theory and in practice. In addition, the influence of calculation error in simulations plays negative role in the process of r...