Research On The Key Technologies Of Deformation Monitoring Based On Image Assisted By Multi-source Information

With the continuous development of economy and the improvement of comprehensive national strength in China,transportation infrastructure construction,water conservancy,and resource development industries is booming,such as the large-scale construction of high-speed railways,bridges,dams,etc.Meanwhile,automated installation engineering and testing projects of large-scale precision equipments are also implemented in large numbers,for instance,the synchrotron radiation accelerator or FAST,Five-hundred-meter Aperture Spherical radio Telescope,for scientific research and ship construction for national defense.In order to ensure the full lifecycle safety and operation of the above-mentioned large-scale engineering projects,reasonable monitoring is essential.Close-range photogrammetry is a branch of photogrammetry,which obtains three-dimensional space coordinates or geometric information of the measurement target by acquiring target images at close distances(ranging from a few centimeters to hundreds of meters)and possesses advantages over traditional measurement methods.However,for some measurement targets under certain circumstances like the aforementioned large-scale projects,industrial projects,narrow and long measurement targets,etc.,the on-site environment is more complicated with many obscurations,which may have a certain impact on the close-range photogrammetry camera position,shooting angle and control point layout,and consequently will directly affect the measurement accuracy and scope of application of close-range photogrammetry.Besides,the shortcomings of photogrammetry itself,such as inaccurate scale information and error accumulation as the measuring range increases,will also affect the monitoring results and lead to measurement results that cannot meet the accuracy requirements of monitoring.Therefore,the single utilization of close-range photogrammetry is often unable to meet the measurement requirements.Nowadays,a variety of sensor fusion technologies are developing rapidly.The multi-sensor fusion technic performs comprehensive processing of multiple information from multiple sensors according to specific standards and combines redundant or complementary information in space or time to derive the result with better precision and reliability.Therefore,based on the analysis of the current research status of deformation monitoring,close-range photogrammetry,and multi-sensor and photogrammetric fusion,this thesis proposes a method of integrating multi-sensor technology into photogrammetry,aiming at the current problems of using close-range photogrammetry to monitor large-scale engineering and industrial projects.The method fuses the sensors like GNSS,UWB,INS,tilt sensor,hand-held laser scanning and laser tracking to achieve the purpose of improving the work efficiency,broadening the usable range,and improving the measurement accuracy.The main contents of this article are as follows:1)The key technology of photogrammetric image processing has been researched and improved,including image preprocessing,recognition and positioning of artificial measurement target feature matching algorithms,etc.Aiming at the circular feature points commonly used in photogrammetry,the circle detection based on Hough is improved,and a new RANSAC circle detection algorithm based on center filtering is proposed.Orienting the demands of subsidence monitoring through photogrammetry,a new kind of photogrammetric settlement monitoring sign with larger coding capacity is designed,and the identification and positioning of the sign are realized.In terms of image matching,aiming at the feature points rich in feature information,the efficiency,accuracy and applicable conditions of SIFT,SURF and ORB algorithms are compared for providing references for the selection of matching algorithm in the follow-up work.In the aspect of matching with no/less feature information,based on the study of epipolar line matching,forward intersection matching and driving point matching,a new joint matching algorithm which is suitable for multi-targets on the same plane using homography matrix and forward intersection is proposed.2)Orientating the demand for low-cost and automated settlement monitoring,a method for close-range photogrammetry elevation measurement is proposed,designing a special barcode sign for close-range photogrammetry elevation measurement,and researching its recognition algorithm.Besides,an elevation monitoring system is developed with the tilt sensor used to compensate the posture of the photogrammetry camera.The feasibility of the method was proved by comparing the measurement result between the total station and close-range photogrammetry in a section of leveling route monitoring.3)Orientating the automated monitoring requirements for outdoor engineering projects,a new combination device is designed,which connects three GNSS receivers with photogrammetry cameras.The method of using three GNSS receivers to provide pose positioning for the camera is studied,and the feasibility of the method is verified through experiments.4)Aiming at the requirement of large number of coding points in automatic monitoring of indoor industrial photogrammetry projects,a method of using UWB and INS to provide pose for industrial photogrammetry cameras is proposed.The combined device of the three is calibrated for realizing camera pose calculation and completing the matching of industrial photogrammetric images with improved driving points.5)Aiming at the problem that industrial photogrammetry cannot measure the parts of the complex structure targets devoid of light reflection marks,while key industrial originals have geometric shape measurement and modeling requirements,the method of combining handheld scanner with photogrammetry technology is explored.This method uses photogrammetry points to provide positioning points for the scanner,which improves the accuracy of the handheld scanner,and completes the unification of the two benchmarks,achieving the purpose of supplementing the photogrammetry results.6)Aiming at the problem that industrial measurement errors will accumulate and measurement accuracy will decrease accordingly as the measurement range expands,this thesis investigates the principle of the laser tracker and its error source,and studies the principles and methods of using the laser tracker to lay and install the control network according to the error source.The control points placed by the laser tracker are used to provide constraints for industrial photogrammetry,which improves the accuracy and efficiency of industrial photogrammetry for large-size targets.