【作者】 李彩萍；

【导师】 张英梅；

【作者基本信息】 太原理工大学 ， 控制科学与工程， 2013， 硕士

【摘要】 我国是世界上原煤产量最大的国家,也是煤矿井下水害最为严重的国家,就其危害程度而言,煤矿水害是仅次于瓦斯爆炸的煤矿第二大灾害。矿井突水有老采空区透水、断层突水和溶洞突水等,造成这些事故的原因众多,如雨季引起地表水、地下水骤增,盲目开采使老空区、采掘工作面顶底板断层水带破裂和地下溶洞打穿透水等等。矿井突水是由多种因素综合诱发的,寻求突水预测的新的实用性理论和方法一直是做好矿井安全工作的重要任务,许多专家学者在这方面做了大量的研究并取得一定的成果。本文从突水预测的基本因素出发,使用新方法获取突水量信息,从而补充和加强现有煤矿井下水害预测方法的精确性和可靠性。本文将视频监控技术应用于矿井突水信息预报监测中。重点研究了数字图像处理技术在矿井突水检测中的应用,通过在突水易发区域布置图像传感器,一方面实时显示监控区域画面,另一方面将采集到的图像采用一种快速的具有环境特点针对性的图像校正方法,对降质的灰度不均匀的突水图像进行亮度变化,增强滤波,然后进行阈值分割,识别目标区域,最后提取特征值。通过对实时采集到的实时图像进行预处理、分割和提取突水特征信息,系统实现了对突水现场自动识别,给出了一种基于图像处理的突水信息自动检测方法,改进了以往人工现场勘查的耗时、耗人力、监测危险和人为主观误差大等缺点。突水区域特征参数作为煤矿安全等级评价的依据之一,文中通过对像素点的统计,计算出区域的面积作为一组参数传递到多传感器数据融合诊断系统来判断突水的严重情况。本文采用可视化编程语言VisualC++2008结合OpenCV计算机视觉库开发了突水图像处理和分析的上位机软件,该软件集成了包括各种预处理算法在内的突水图像处理与分析的算法,利用数据库访问技术ADO访问存储在数据库中的图像特征值,用于进一步的数据分析和处理。上位机软件拥有可视化界面,可显示图像的处理及分析结果,操作方便。本文为煤矿突水检测方法提供一种新思路,并对类似环境下的图像处理任务具有借鉴意义。更多还原

【Abstract】 China is the world’s biggest producer of raw coal, is also the most serious water damage underground. Terms of the extent of the harm, the coal mine water disaster is the second largest coal mines after the gas disaster disaster. Mine water inrush generally include the mined-out areas permeable and water inrush from fault and cave water inrush etc.the cause of these incidents are many and varied, such as the rainy season caused by surface water, groundwater surge mining face in roof and floor faults hose rupture, blind exploitation of the gob area, underground cave penetrated permeable, and all of these factors together result to inrush.Seeking practical theory and method of water inrush forecast is always an important element of the work of mine safety. Many scholars and experts have done a lot of research in this area and the corresponding results. Starting from the basic factors of water inrush forecast, the attempt by the new method, to obtain new water inrush information, thereby increasing the accuracy and certainty of the existing coal mine flood forecast method.This study apply video surveillance technology to mine water bursting forecast information monitoring. Focusing on the intelligent recognition method based on digital image processing in mine water inrush and its main feature extraction, researching the application of modern computer image processing technology in the water inrush automatic detection. By arranging the image sensor in water inrush prone areas, given a water inrush automatic detection method based on image processing, For a quick grayscale the uneven image correction method, the degraded water inrush image enhancement and denoising, and then threshold segmentation, extracted linear features in the image, Finally, identifing regional by the linear characteristic from the original grayscale image.Using digital image processing techniques to real-time acquisition of images preprocessing, segmentation, extraction of water inrush characteristic, the system achieve the water inrush site automatically identify and overcome the previous manual site survey time-consuming, consumption of human, risk of detection, subjective error greater.The calculation of the the regional characteristic parameters of the water inrush is one of the basis for the evaluation of coal mine safety level. In the feature extracting portion, the paper by the statistics of the pixel point, calculate the size of the area as the feature information. Finally, the amount of extracted features as a set of parameters used in the feasibility of multi-sensor data fusion technology for coal mine flood forecast,the predicted results contrast with the actual underground scene detection results show it can be relatively satisfied with the effect, the result shows that the method feasibily, accurately and effectively.Using visual programming language Visual C++2008combine OpenCV computer vision library developed image processing and analysis software. The software integrates a the water inrush image processing and analysis algorithms, including image preprocessing and image feature extraction algorithms, Embedded a water inrush image features database management system, which can store extracted image feature value directly for further analysis an identification. The processing and analysis of the results of the image can be displayed on the screen, The system interface is friendly, easy to operate.The system for coal mine water inrush detection method provides a new idea, and image processing tasks similar environment to be learned.更多还原