Research On Key Techniques Of Computer Vision Method And Close-range Photogrammetry On Similar Simulation Of Surrounding Rock

The similar simulation experiment refers to the simulation of real geological conditions using a certain proportion of mixed materials in a laboratory scenario.The mine rock layers are scaled down by a certain proportion,and then simulated mining is performed on the model to observe the displacement after mining.The similar simulation experiment is an indispensable important experiment before underground roadway excavation in the mining area.Through this experiment,the physical deformation of the entire overlying strata in the process of roadway excavation can be obtained,which has a certain degree for the excavation work in the field.Guidance significance,the traditional deformation monitoring programs are mainly dial gauge method,dial indicator method,total station method,speckle correlation measurement method.Close-range photogrammetry is a technique that combines photogrammetry theory with image processing technology.This technology has been widely used in medical image processing,cultural relic reconstruction,and other fields,especially in recent years,the rapid development of computer vision,deep learning,computer vision The combination of theory and close-range photogrammetry theory has promoted the development of this discipline.The close-range photogrammetry has the characteristics of non-contact and does not cause disturbance to the measurement target,and is particularly suitable for the measurement of the appearance and movement state of the object.The measurement of the similarity of the roadway surrounding rock measurement speed is slow,the measurement accuracy is low,and the measurement cost is high.The combination of landscape photogrammetry,computer vision and analogous displacement monitoring experiments of roadway surrounding rocks has p roposed a new monitoring scheme.The thesis introduces the grey weighted centroid method and ellipse fitting method in the center of the high-precision positioning monitoring sign,and compares and analyzes the two methods.Finally,it uses the ellipse fitting center positioning algorithm to extract the center coordinates of the monitoring sign.Aiming at the high-precision matching of monitoring landmarks on sequence images and the matching of monitoring points with the same name at different times,a stra tegy of double-constrained stereo matching is proposed.The algorithm first uses the basic matrix to achieve the rough matching,and then implements the exact matching scheme based on the epipolar constraint to realize the stereo matching of the monitoring point of the same name.Finally,the matching point number is used to achieve the matching process of monitoring points of the same name at different times.The reliability and stability of the algorithm are verified.The problem of non-convergence of the external orientation elements of the image capturing station using conventional monolithic space resections was investigated in the case of large-angle photography in close-range photogrammetry.This paper combines the advantages of pyramid and unit quaternions,and improves the traditional algorithms for solving the image outside elements,and uses the field shooting data to verify the algorithm.Finally,the algorithm proposed in this paper is implemented by VC and open source computer vision library Open CV,and the data calculated by this system is compared with the data measured by the total station.The results show that the scheme is feasible,the solution results are accurate,the accuracy of the solution is high,and it can meet the similar analog displacement monitoring requirements of roadway surrounding rock.