Research On Calibration Technology For Cameras With Large FOV Based On Virtual Planar Target

Stereo vision measurement systems are widely used in industry,agriculture,national defense and other fields,and their key step is camera calibration,which means the accuracy of calibration directly determines that of entire measurement system.Therefore,researches on high precision camera calibration technology have important significance for improving the performance of the vision measurement system.In measurement experiments of helicopter rotor blades movement parameters based on the stereo vision system,it was found that the accuracy of traditional high precision calibration methods is not ideal in a large field of view(FOV).Based on this,by analyzing influence mechanism of the calibration accuracy in large FOV,the camera calibration technology based on a virtual large planar target(VLPT)has been studied,and its influencing factors have been analyzed systematically in this paper.The main work contents and corrected results are as follows:(1)An experimental verification system of camera calibration methods for large FOV has been established.According to verification requirements of the calibration methods,the experimental verification system is composed of hardware platform,software platform and target design.The hardware platform includes visual module and data processing module.The software platform consists of image acquisition module,camera calibration module and calibration precision test module.With 8*8 orbicular calibrated points,anchor point and directed point,a kind of planar target has been designed by considering type of the target,type of the feature points,number of the calibrated points,positioning of the targets and sorting of the calibrated points.The experimental verification system provides the basic guarantee for validating the effectiveness of the proposed method and analyzing the influencing factors of the calibration accuracy.(2)A camera calibration method for large FOV based on VLPT was proposed.Firstly,a number of calibrated images are obtained by utilizing several mutually independent small targets(MIST).Secondly,for each calibrated image,by positioning the center of feature points,each MIST is found and calibrated points on it are sorted.Thirdly,by setting a virtual plane on which virtual points corresponding to calibrated points are found,one MIST can be connected to another one.In this way,a VLPT of this calibrated image is constructed.Next,camera parameters are calculated by using the obtained VLPTs,then they are optimized through nonlinear optimization algorithm.By comparing the proposed method and the traditional calibration methods,the results shows that calibration accuracy of the proposed method is roughly the same as that of the large-target method,and obviously higher than that of the small-target method.(3)Influencing aspects of calibration accuracy of the VLPT method have been analyzed.By studying the process of the VLPT method,the influence mechanisms of calibration precision respectively on positioning precision of the feature points,position of the target placement,size of each MIST and number of the calibrated images were analyzed.Through theoretical analysis and experimental verification,the following results were achieved: 1)positioning accuracy of the feature points has little effect on calculating the calibration parameters,but seriously affects the calculation results of 3D information of measured points.2)By selecting the placement style of “*” word,the MIST with an area of 245 mm * 245 mm,and 8 pairs of calibrated images,high calibration accuracy can be achieved.This provides a guide for the rational use of the proposed method,and further improves the calibration accuracy.