Research The Algorithm Of Camera Calibration Based On The Fundamental Matrix

In recent years,with the development of digital imaging technology,CCD camera as a result of imaging speed,safety,easy to store images,inexpensive,and many other advantages,has been developing rapidly,but the CCD camera is a non-metric camera,interior orientation of the camera parameters and distortion parameters are unknown,when applied to the field of photogrammetry,it requires calibration to obtain the camera orientation elements and lens distortion parameters.The fundamental matrix is an important part of computer vision,which describes the relationship between the two images corresponding to the image point,the basic structure of the matrix is independent of the scene,only to rely on the internal parameters and relative attitude of the camera,which has same purpose of photogrammetry coplanarity condition equation. Although the relationship between the two images containing camera internal parameters,but previous research showed that directly using the fundamental matrix or coplanar equation is not suitable to compute the camera internal parameters, the solution is not stable and low accuracy.This paper takes fundamental matrix as the object of study,introducing photogrammetric basic theoretical knowledge,under the circumstance of only image information is instability for computing the camera parameters,computing the relative orientation elements through the fundamental matrix,using point projection coefficient to obtain model coordinates,adding the relative control information and finally put the above information into a complete theoretical system of self-calibration bundle adjustment,in order to overcome the traditional bundle adjustment more memory resources,the coefficient matrix is large and easy singular disadvantage,introducing sparse bundle adjustment of computer vision to obtain the camera various parameters.In the experiment,for the sake of automatically obtain images coordinates,we design a color-coded flags to obtain high-precision image point coordinate information.The final experiments showed that the proposed algorithm can effectively solve the various camera parameters,in order to verify the accuracy of the solution,comparing the algorithm of this paper with the three-dimensional control and flat field calibration results,the consequence showed that the results obtained in the method of this paper between the plane and three-dimensional control field calibration.