视频监控技术在煤炭工业中得到广泛应用，已成为井下安全生产和管理的重要技术手段之一。胶带运输机的运动状态监测是井下视频监测和管理的重要内容之一。研究基于视频监控的胶带运输机运动状态监测方法，对于提高煤矿安全监控和管理水平具有重要的意义。本文主要研究内容和成果如下： 研究设计了适于井下视频图像特点的图像预处理方法。该方法先采用双边滤波进行降噪，然后将多尺度Retinex方法与直方图均衡化相结合，实现了井下胶带运输机视频图像的增强。实验结果表明，图像降噪和增强效果良好，实现了低亮度区域增强及高亮度区域抑制。 本文针对胶带运输机上煤块实时检测问题，提出了一种基于改进的混合高斯模型的煤块目标检测方法。该方法首先将视频图像划分为块，以块的均值代替该块内的像素进行建模，缩减了背景建模时间，然后将背景建模状态阶段划分为两阶段，根据不同的阶段需求采用不同的学习率，加快了场景转换时背景形成速率。实验结果表明，该方法能够有效提高煤块检测的速度，采用并行系统可以满足视频检测实时性要求。 利用视觉计算理论研究了胶带运输机的运动状态监测方法。该方法先将卡尔曼滤波与最近邻法结合进行煤块目标跟踪，然后根据简化的成像模型计算煤块运行的速度，借助运输机的运行速度实现对胶带运输机的运动状态监测。实验表明，该方法能够用于胶带运输机运动状态的异常监测。 本文提出的基于视觉技术的井下胶带运输机运动监测方法可充分利用工业监控视频信息，实现对胶带运输机运动状态的实时自动监测，对煤矿监控和管理以及灾害预防具有重要意义。

Video surveillance technology has been widely applied in the coal industry and also become one of techniques of the safe production and management. The movement of coal belt conveyor is one of the important content of coal mine monitoring and management in the coal mine safety monitoring system. Researching on motion monitoring method of underground belt conveyor based on surveillance video is of great significance for improving the level of mine safety monitoring and management. The main research content and results are as follows: The pre-processing method which is suitable for underground video characteristics is studied in the paper. Firstly the bilateral filtering is adopted to reduce noise, and then the multi-scale Retinex method combined with histogram equalization is used to realize underground belt conveyor video image enhancement. Experiment results show that the noise reduction and enhance method is better, since the low-brightness part of conveyor video images is strengthened and the high part is suppressed. An improved Gaussian mixture background subtraction is presented for real-time coal target detection on the coal belt conveyor. Firstly, the video image is divided into blocks. Secondly the mean of blocks is instead of the pixels within the block for background modeling to reduce the modeling time. Thirdly, the state phase of the background modeling is divided into two stages, and then different learning rate is selected according to the different stages demand to speed up the background formation rate. Experiments show that the method can effectively improve the coal detection speed and the parallel system for detecting will meet real-time requirements. The motion monitoring method of underground belt conveyor is researched based on vision computing theory. The Kalman filter combined with the nearest neighbor method is used to achieve coal target tracking, and then the speed is calculated according to the simplified camera imaging model to monitor the motion state of the belt conveyor. Experiment results show that the method can be used for exception monitoring of the motion state of belt conveyor. The motion monitoring method of underground belt conveyor based on vision technology can take advantage of the industrial surveillance video so as to realize real-time automatic monitoring of the belt conveyor movement and is of great significance for mine monitoring and management and disaster prevention.