Research On Photogrammetry Method Based On Binocular Joint Calibration And Its Engineering Application

In recent years,China’s ultra-high-speed development in transportation and infrastructure construction has largely benefited from the efficient operation capabilities of a series of large-scale machinery and equipment.An important means of long-term stable operation of equipment.As a non-contact measurement method,photogrammetry has the advantages of high efficiency and low cost,and has been widely used in the field of mechanical structure measurement.However,the harsh measurement environment and the special structural characteristics of mechanical equipment limit the development of existing photogrammetry technology,and the measurement accuracy and efficiency cannot fully meet the data requirements of safety assessment.Based on the traditional binocular photogrammetry technology,this paper takes the multi-line intersection feature of the key components of mechanical equipment as the main goal,and makes research on the camera calibration accuracy,the automatic identification of image key points and the automatic matching of the same name points.The main content and the results are as follows:(a)A binocular camera joint calibration algorithm based on the principle of hierarchical global scanning is proposed,which effectively improves the error problem of the current binocular camera calibration and ensures the accuracy of the final measurement results.Through a detailed understanding of the basic theory of photogrammetry calibration,the theoretical sources of errors are analyzed,and the simulation solution of ideal data is used to explore the influence of the numerical error of focal length calibration on the accuracy of measurement results.The scanning calibration method and the camera calibration test in the laboratory verify the feasibility and accuracy of the algorithm.(b)Based on the unique properties of multi-line intersection points,the key dimension feature points that need to be measured in the structure of large mechanical equipment are often presented.A multi-line intersection point recognition algorithm based on improved Hough transform is designed.By improving the traditional Hough transform line recognition algorithm and setting a reasonable distance threshold,the difficult problem of line recognition caused by the non-strict collinear points in the boundary recognition points is solved,and the point information used is used to the greatest extent to ensure the line Then,according to the structural characteristics of the machine itself,a discrimination method based on the number of straight lines constituting the intersection point is proposed,and the type of all straight line intersection points is screened and used as an important point matching basis.(c)Based on the matching area division of LIDAR point cloud data,a matching method of left and right identical points based on epipolar line matching principle and intersection point type is proposed.Based on the intersection point of the straight line identified by the Hough transform,the reasonable image points obtained by screening the LIDAR point cloud inverse mapping through the distance limit are used as the division nodes to divide the matching area,and then the left and right same-name points are located in the same areain the limited matching area.The principle of epipolar line matching principle and the type of intersection point are used to efficiently and accurately match the points with the same name.Through the verification of engineering simulation experiments,the validity and accuracy of the measurement results of this method are shown.