Study On The Key Technology Of The Accurate Measurement System Of Similarity Material Model Based On Moving Camera

With the development of digital photogrammetry, computer vision and digital camera technology, photogrammetry has been widely used in the experiment of geotechnical engineering. But because of the conflict between the field of view of image acquiring and measurement precision, the traditional method of digital photogrammetry cannot meet the accuracy requirement of deformation monitoring of the large-span similarity material model. To solve this confliction, this paper designed an image acquisition platform, obtaining continuous image by industrial camera which driven steady uniform motion by the servo motor; Combined with the theory of photogrammetry and computer vision, the collected images are processed and analyzed strictly; The accurate deformation of the monitoring points to the model can be calculated, and the accuracy of the deformation monitoring of the similar material model can be verified. This paper studied the related theory, key problems and solutions of the automatic precision monitoring for large-span similar material model achieved by the mobile industrial camera. The main work and research results are as follows:(1) Designed an image acquisition platform for single camera moving, driven by servo motor, controlled by computer, acquiring image by industrial camera, and achieved camera parameters controlling, image display, high quality image acquisition and real-time transmission combining with industrial camera software trigger system,(2) In this paper, the particularity and feasibility of the precision monitoring method based on industrial camera of large-span similar material model was analyzed. Combined with the theory of photogrammetry and computer vision, this paper firstly corrected the image deformation caused by the camera moving shoot according the deformation distortion parameters and tilt angle of image obtained by camera calibration and single-photo space resection, and obtained equivalent orthophoto; secondly proposed a coordinate constraint algorithm to improve the quality of image matching based on SIFT feature point extraction, euclidean distance matching,and the stitched image seamlessly are obtain combining with image registration and the gradated in-and-out amalgamation algorithm; finally the paper proposed a new method of automatic landmarks identification by combining template matching algorithm and gray threshold method to improve the robustness of automatic recognition of landmarks, and the automatic recognition and high precision positioning of the monitoring points are realized in the model combining the circular fitting method.(3) The similar material model of coal dust are mined using the methods of this paper designing acquisition platform and processing, showing that the relative position accuracy is higher than 0.14 mm and the relative displacement accuracy is higher than 0.20 mm, verifying the method feasibility that the paper proposed precision monitoring method of large-span similar material model, satisfying the accuracy requirements of monitor for similarity material model, obtaining the accurate deformation curve. Moreover, the model image of the large field of view and high resolution save more details of the model.The research of this method not only introduced the photogrammetry to the similar material model of large span, but also can improve the accuracy of the deformation monitoring of the similar material model of the middle and small scale. It provided a new monitoring method by applying the close-range photogrammetry to comprehensively study the subsidence law of similarity material model.