Research On Shaft Wall Scanning And Mapping System Based On Machine Vision

The increase of mining depth will lead to the increase of shaft pressure,which will eventually lead to shaft wall deformation due to fatigue effect,corrosion effect,material aging and other catastrophic factors.Shaft is the "throat" in the whole coal mine production,and its stability and safety have a significant impact on the whole coal mine production.Therefore,the deformation monitoring of shaft wall is an important method to prevent the disasters caused by shaft deformation and ensure the safety of coal mine production.It has important theoretical significance and practical application value to build a high-precision and high-efficiency shaft wall scanning and mapping system and study the related technology to improve the level of shaft deformation monitoring and ensure the safety production of coal mine.This paper combines machine vision,3D reconstruction and other technologies to build a borehole scanning imaging system,and build the system to carry out engineering experiments and simulation experimentsFirstly,the paper analyzes the functional requirements of the Shaft Wall Scanning and Mapping System and designs the overall scheme.Combined with the actual working conditions,it determines the relevant parameters required by the system design,and then determines the technical implementation scheme of each part.On this basis,the system is constructed from hardware and software.The system hardware adopts the idea of modular design,and completes the parameter calculation and selection design of measurement module,control module,communication module and power module.In order to ensure the robustness,real-time performance and high efficiency of the software system,the software part is designed based on the communication principle of ROS system.The whole software follows the hierarchical design principle.All the function nodes are divided into acquisition drive layer,communication layer,data processing layer and display layer to realize the acquisition and processing of wellbore coordinate data.The upper and lower computers use the distributed architecture and network communication framework of ROS system,and combine with QT designed GUI interface,realized the image,model visualization and other human-computer interaction functions.Each function node can run independently,which is more convenient for subsequent optimization and function increase.In this paper,the key technologies of data processing,such as spot image and point cloud processing,are studied.In the image processing,the traditional Otsu method is improved and optimized to complete the effective recognition and segmentation of small spot image,and the area difference is included in the threshold selection criterion function to avoid the problem that the spot is too small to be recognized;in the calculation of spot center coordinates,the simulation spot is used to compare different algorithms to determine the appropriate high-precision algorithm.In the research of point cloud processing algorithm,the point cloud is segmented by direct filtering,which can filter out the influence of some equipment in the wellbore on wellbore measurement,significantly reduce the amount of data for subsequent point cloud processing,and reduce the complexity of subsequent segmentation calculation.At the end of the paper,the borehole scanning imaging system is studied in laboratory and field.Through the analysis of the simulation experiment in the laboratory,the point error of the data obtained by the system is calculated,and the method to effectively reduce the system error is analyzed and discussed.Through the field test,the system completes the shaft wall scanning imaging of part of the depth of the shaft.By comparing with the theoretical shaft wall model,the deformation of the well is evaluated,and the reliability and practicability of the system are verified.The thesis has 59 figures,11 tables and 80 references.