基于SIFT特征匹配和模约束的摄像机分层自标定方法

在一些计算机视觉和摄影测量任务的执行过程中,需要在线地标定摄像机参数,这就使得不依赖标定参照物的自标定成为必需,提出一种基于SIFT特征匹配和模约束的摄像机分层自标定方法。自由移动或旋转摄像机拍摄同一场景内部参数不变条件下的四幅以上图像。对每幅图像进行SIFT特征点提取,通过特征点匹配在每幅图像中分别获得对应三维场景空间同一特征点的像素坐标。进行投影标定,获得每幅图像在投影重建空间中的相机投影矩阵,以及每个特征点在投影重建空间中的三维坐标。进行仿射标定,采用模约束法确定无穷远参考平面在投影重建空间中的参数。进行度量标定,确定内参矩阵。实验表明,该方法能在线地稳定地获得摄像机内参标定结果,从而对现有的摄像机自标定方法进行了改进。

Self-Calibration of Camera Based on SIFT Matching and Modulus Constraint

In the field of computer vision and photogrammetry, it is constantly necessary to online acquire the camera parameters through self-calibration. This paper proposes an approach about self-Calibration of camera based on SIFT matching and modulus constraint. The proposed method firstly captures more than four images of the same scene in the case of constant internal parameters by freely moving and rotating the camera. After extracting SIFT features from each image and matching them among different images, the feature pixels from different images but corresponding to the same 3D feature point are acquired. Projective calibration determines the camera projective matrix of every image and the coordinates of every 3D feature point in the projective reconstruct space. Affine calibration calculates the parameters of infinity reference plane in this space by virtue of modulus constraint. Lastly metric calibration acquires internal reference matrix according to the parameters of infinity reference plane. The proposed method can be online applied to accurately and stably obtain internal parameters of camera without using the calibration reference with known 3D information. Real data has been used to test the proposed approach, and good results have been achieved, thus improving the existing camera calibration.