High-precision Real-time Measurement Method Of Dynamic Target Pose

Vision pose measurement technology is a widely used motion parameter measurement method at present.It has high measurement accuracy and can be applied to objects to be measured with fast moving speed and wide range of motion and complex and changeable measurement environments.At present,the related research mainly focuses on three aspects:the establishment of the vision measurement model,the design of the target and the calibration of the camera.Based on the analysis of the existing vision measurement models,this paper proposes a pose measurement model based on monocular vision,and explores the camera calibration method.The main work contents are as follows:(1)Construction of the monocular vision pose measurement model.At present,the commonly used vision measurement models mainly include monocular vision measurement model,binocular vision measurement model,and polyocular vision measurement model.The above three visual measurement methods all have the function of measuring the target trajectory and pose.Compared with monocular vision,binocular vision and polyocular vision measurement are both limited by the scope of the measurement site,and have the advantages of small measurement space and short measurement distance,and other defects.Monocular vision can make up for these defects and deficiencies well,and has the advantages of simple system structure and few camera calibration steps.In this paper,a monocular vision pose measurement method is proposed by designing a new cooperative target.Just fix the designed new cooperative target on the target to be measured,and use the relationship between multiple spatial feature points to measure its pose.(2)Camera calibration algorithm based on plane calibration target.A simple and fast camera calibration method is proposed for the radially distorted pinhole perspective projection model.This method separates the camera’s internal and external parameters from the camera model’s distortion coefficient under the premise that the image center point coincides with the center of the CCD or CMOS sensor,so that the camera calibration can be further performed linearly.The method avoids the error caused by nonlinear optimization,reduces the complexity of the algorithm,improves the calibration accuracy,and saves the calculation time.Firstly,the orientation distortion coefficient of the camera lens is calibrated according to the principle of cross ratio invariance of perspective projection.Then,according to the relationship of rotation transformation and translation transformation,the internal and external parameters of the camera are linearly solved by taking full advantage of the radial distortion constraint,the orthogonality of the rotation transformation and the zero product difference of the rotation matrix.(3)Camera calibration algorithm based on 3D calibration target.Aiming at the problems that nonlinear optimization algorithms are used many times in the traditional plane camera calibration method,the coordinate dimension information of the calibration target corner point is missing,and the positioning error of the image center point is large,a high-precision positioning method of the image center based on the 3D calibration target is proposed.Complete camera calibration,most of the current camera calibration algorithms need to give the coordinates of the image center point in advance,or need multiple images to complete the camera calibration.The method first calibrates the radial distortion coefficient of the camera lens according to the principle of cross ratio invariance of perspective projection.After the image is distorted and corrected,according to the rotation transformation relationship and translation relationship,the orthogonality of the rotation matrix is fully utilized to establish an equation system,and the internal and external parameters of the camera are linearly solved.Camera calibration can be done by an image with a 3D calibration target.(4)Analysis of experimental results.The measurement accuracy of the method proposed in this paper is evaluated,and the measurement results obtained through experiments are compared with the actual data to verify the effectiveness of the method.Finally,the measurement efficiency of this method is statistically analyzed and compared with Tsai’s method.This method saves about 35% in calibration time and improves at least 15% in accuracy,which verifies the high efficiency and real-time performance of this method.