井下运动目标跟踪及预警方法研究

发布时间：2018-06-10 00:25

本文选题：检测 + 跟踪　；参考：《西安科技大学》2015年硕士论文

【摘要】：安全生产是煤矿安全的一个永恒话题,其中安全运输是安全生产中的一个重要组成部分。目前,国内煤矿井下都安装了视频设备,但如何充分利用这些设备提高井下机车撞人等轨道事故的预防能力有待进一步研究,因此,本文研究基于视频监控的井下运动目标跟踪及预警方法,对于提高煤矿井下轨道事故的预防能力及煤矿安全生产具有重要意义。首先,针对井下轨道上的运动目标的检测与识别问题,对井下视频图像进行预处理,采用背景减除法和帧差分法相结合的方法,检测出井下轨道上的运动目标。在此基础上将目标形状特征和纹理特征相结合,采用基于特征匹配方法对运动目标进行识别。其次,针对运动目标的跟踪问题,提出基于粒子滤波的井下多运动目标跟踪算法。在建立并初始化多目标跟踪列表的基础上,通过状态转移方程更新目标状态,计算并归一化每个粒子权值,利用目标状态方程估计输出跟踪目标的位置。该方法可有效克服前景目标与背景灰度相似、目标遮挡等情况的影响,具有较高的准确率。最后,针对运动目标的预警问题,提出基于时间安全模型的井下预警算法。该算法首先监测运动目标的速度及方向,然后依据运动目标之间不同的状态,计算出人与机车和人与轨道之间的欧式距离。将获得的距离与基于时间安全预警模型计算所得的安全警戒值进行比较,判断是否存在安全隐患,制定安全预警决策。实验结果表明,本文所提井下运动目标跟踪及预警方法利用视觉计算理论,可自动检测、识别和跟踪轨道上的运动目标,并能够有效监测目标之间的距离,做出预警提示,可提升煤矿行业信息化和智能化水平,为保证煤矿的安全生产提供新的技术手段。
[Abstract]:Safety production is an eternal topic of coal mine safety, in which safe transportation is an important part of safe production. At present, video equipment is installed in the underground coal mine in China, but how to make full use of these equipments to improve the ability of preventing railway accidents, such as locomotive collision, etc., needs further study. In this paper, the tracking and early warning method of moving targets based on video surveillance is studied, which is of great significance to improve the prevention ability of underground rail accidents and the safety of coal mine production. Firstly, aiming at the problem of detection and recognition of moving target on underground track, the video image is preprocessed and the moving target on underground track is detected by combining background subtraction and frame difference method. On the basis of this, the shape feature and texture feature of the target are combined, and the feature matching method is used to recognize the moving target. Secondly, aiming at the problem of moving target tracking, a multi-moving target tracking algorithm based on particle filter is proposed. On the basis of establishing and initializing the multi-target tracking list, the target state is updated by the state transfer equation, each particle weight is calculated and normalized, and the position of the output tracking target is estimated by the target state equation. This method can effectively overcome the influence of foreground target and background gray level, target occlusion and so on, and has high accuracy. Finally, a downhole early warning algorithm based on time safety model is proposed for the early warning of moving targets. The algorithm firstly monitors the velocity and direction of moving targets, and then calculates the Euclidean distance between human and locomotive and man and track according to the different states of moving targets. The distance obtained is compared with the security warning value calculated based on the time security early warning model to judge whether there is a potential safety hazard or not and to make a safety early warning decision. The experimental results show that the method proposed in this paper can automatically detect, identify and track moving targets on track by using visual computing theory, and can effectively monitor the distance between targets and make warning. It can improve the level of information and intelligence of coal mine industry and provide new technical means to ensure the safety of coal mine production.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TD76;TP391.41

【相似文献】