基于点云数据处理的激光雷达监控系统的设计与开发
摘要
近年来,随着信息技术的发展,监控系统得到了快速的发展,为国民经济的良好的运转提供了坚实的保障。区别于采用照相机、摄像机,红外等传统信息采集设备,激光雷达拥有全天候24小时且不受天气干扰等独特的优势,它能够精确快速地获取目标物体的三维空间点云数据,对场景变化更准确的判断。如今,地面激光雷达技术已经被广泛应用,多个国家成功地对建筑物遗迹和雕像进行了三维模型重建,实现了物体完整还原。与激光雷达技术的硬件发展相比,点云数据处理的技术相对滞后,点云模型还很难应用到实际的系统中,原因就是点云模型的体积等积分属性很难计算。
体积是点云模型重要的几何属性,在许多应用场景需要频繁地计算点云模型的体积。目前,就点云模型而言,体积的计算首先要重构点云曲面模型。但是使用重建曲面来计算体积的方法,把工作集中在重构算法上,需要额外的消耗大量的运算时间与空间,这样,对于只需要快速求取体积的应用系统中多出了许多工作,也不满足实时监控系统的时间要求。其次,对于有较为复杂的拓扑结构的点云模型,重建曲面可能会失败从而导致无法求取体积。
本文拟通过分析处理激光雷达所获取的三维点云数据,建立对象物体点云模型,计算模型的体积、移动位移等物理属性,按照危险状况的检测策略,对坡地塌方进行监控,设计开发一套全天候、实时预警系统,以便及时的把危险信息发布出去,减少灾害带来的损失。论文详细介绍了系统的设计方案,以及在实现中需要的应用技术和相关算法。系统采用串口编程技术获取激光雷达的数据和控制云台,并使用多线程技术提高串口通信的效率和解决多串口实时采集数据的问题。
对于点云数据,分析介绍了目前点云数据处理的一些算法,并设计了一种快速求取点云模型体积的方法,首先使用增量式算法计算点云的凸包用来近似物体,再将凸包分解成上下两个三角网格面,使用投影法分别求取它们的投影体积,它们两者之差即是所求模型体积。实验表明该算法实现简单,可快速地求解处理具有任何几何和拓扑复杂性的点云模型。系统采用模块化的设计方法,共分为信息采集模块、信息处理模块、分析预警模块和信息传输模块。
系统实现了对铁路沿线易塌方区域的监控,能够及时的判断塌方灾害,并将预警信息发送到指定的地方,同时该系统的点云模型的快速求解算法不仅避免了系统出错的可能,还提升了分析预警的速度,使得系统的实时性大大增强。
In recent years, with the development of information technology, the monitoring system has been developed by leaps and bounds, and provides a solid guarantee for the good operation of the national economy. Different with traditional information acquisition device like the camera, video camera, infrared, laser radar has a unique advantage which can work24hours a day with the weather disturbances. That means it can quickly and accurately obtain three-dimensional point cloud data of the object and judge the scene changes more accurately. Laser radar has24hours a day from the unique advantage of the weather disturbances, and so it can quickly and accurately obtain three-dimensional point cloud data of the object, the scene changes more accurate judgments.
The ground-based laser radar technology has been widely used, several countries have succeeded in building monuments and statues of the three-dimensional model reconstruction to achieve the objects in a complete and detailed reduction. Compared with the development of laser radar technology hardware, point cloud data processing technology is lagging behind and point cloud model is also difficult to apply to the actual applications, the reason is that the integral properties of the point cloud model of the volume difficult to calculate.
Volume, as the basic geometric property of objects, need to be calculated frequently in many applications. At present, volume is basically calculated through the reconstruction of object surface indirectly, increasing the number of unnecessary work. Especially, for the more complex topology, the reconstruction of surfaces may fail so that it can't get the volume.
Through the analysis of processing3D point cloud data obtained by laser radar and establishment of the point cloud model to calculate the volume and displacement to monitor the particular scene in accordance with the detection strategy of the dangerous situation, This article wants to develop a set of all-weather, real-time early monitor system, in order to timely risk information publish, reduce the losses caused by disasters.
The paper details the system design, and application in the design and implementation of technology and point cloud data processing algorithm. The system use serial programming techniques for laser radar data and control PTZ, and multi-threading technology to improve the efficiency of the serial communication and resolve the problem of multi-serial real-time data collection.
The system uses a modular design approach, which is divided into the information collection module, the information processing module, the analysis of early warning modules and information transmission module. System can monitor the area of the railway line easy to collapse and timely judge landslide disaster, and then send the warning information to the specific place. At last the fast algorithm for computing volume also can enhance the speed of the analysis.
体积是点云模型重要的几何属性,在许多应用场景需要频繁地计算点云模型的体积。目前,就点云模型而言,体积的计算首先要重构点云曲面模型。但是使用重建曲面来计算体积的方法,把工作集中在重构算法上,需要额外的消耗大量的运算时间与空间,这样,对于只需要快速求取体积的应用系统中多出了许多工作,也不满足实时监控系统的时间要求。其次,对于有较为复杂的拓扑结构的点云模型,重建曲面可能会失败从而导致无法求取体积。
本文拟通过分析处理激光雷达所获取的三维点云数据,建立对象物体点云模型,计算模型的体积、移动位移等物理属性,按照危险状况的检测策略,对坡地塌方进行监控,设计开发一套全天候、实时预警系统,以便及时的把危险信息发布出去,减少灾害带来的损失。论文详细介绍了系统的设计方案,以及在实现中需要的应用技术和相关算法。系统采用串口编程技术获取激光雷达的数据和控制云台,并使用多线程技术提高串口通信的效率和解决多串口实时采集数据的问题。
对于点云数据,分析介绍了目前点云数据处理的一些算法,并设计了一种快速求取点云模型体积的方法,首先使用增量式算法计算点云的凸包用来近似物体,再将凸包分解成上下两个三角网格面,使用投影法分别求取它们的投影体积,它们两者之差即是所求模型体积。实验表明该算法实现简单,可快速地求解处理具有任何几何和拓扑复杂性的点云模型。系统采用模块化的设计方法,共分为信息采集模块、信息处理模块、分析预警模块和信息传输模块。
系统实现了对铁路沿线易塌方区域的监控,能够及时的判断塌方灾害,并将预警信息发送到指定的地方,同时该系统的点云模型的快速求解算法不仅避免了系统出错的可能,还提升了分析预警的速度,使得系统的实时性大大增强。
In recent years, with the development of information technology, the monitoring system has been developed by leaps and bounds, and provides a solid guarantee for the good operation of the national economy. Different with traditional information acquisition device like the camera, video camera, infrared, laser radar has a unique advantage which can work24hours a day with the weather disturbances. That means it can quickly and accurately obtain three-dimensional point cloud data of the object and judge the scene changes more accurately. Laser radar has24hours a day from the unique advantage of the weather disturbances, and so it can quickly and accurately obtain three-dimensional point cloud data of the object, the scene changes more accurate judgments.
The ground-based laser radar technology has been widely used, several countries have succeeded in building monuments and statues of the three-dimensional model reconstruction to achieve the objects in a complete and detailed reduction. Compared with the development of laser radar technology hardware, point cloud data processing technology is lagging behind and point cloud model is also difficult to apply to the actual applications, the reason is that the integral properties of the point cloud model of the volume difficult to calculate.
Volume, as the basic geometric property of objects, need to be calculated frequently in many applications. At present, volume is basically calculated through the reconstruction of object surface indirectly, increasing the number of unnecessary work. Especially, for the more complex topology, the reconstruction of surfaces may fail so that it can't get the volume.
Through the analysis of processing3D point cloud data obtained by laser radar and establishment of the point cloud model to calculate the volume and displacement to monitor the particular scene in accordance with the detection strategy of the dangerous situation, This article wants to develop a set of all-weather, real-time early monitor system, in order to timely risk information publish, reduce the losses caused by disasters.
The paper details the system design, and application in the design and implementation of technology and point cloud data processing algorithm. The system use serial programming techniques for laser radar data and control PTZ, and multi-threading technology to improve the efficiency of the serial communication and resolve the problem of multi-serial real-time data collection.
The system uses a modular design approach, which is divided into the information collection module, the information processing module, the analysis of early warning modules and information transmission module. System can monitor the area of the railway line easy to collapse and timely judge landslide disaster, and then send the warning information to the specific place. At last the fast algorithm for computing volume also can enhance the speed of the analysis.
引文
[1]王玉锐.智能视觉监控系统中运动目标检测算法研究[D].山东大学,2009.
[2]Arman F., Aggarwal J.K.: Object recognition in dense range images using a CAD system as a model base. Robotics and Automation, 1990. Proceedings.,1990 IEEE International Conference on, 13-18 May 1990. vol.3,1858-1863.
[3]Delaunay, B.Sur la sphere vide, Izvestia Akademii Nauk SSSR. Otdelenie Matematicheskikh i Estestvennykh Nauk. 1934(7):793-800.
[4]Ohtake Y, Belyaev A, Seidel H. 2004. 3D scattered data approximation with adaptive compactly supported radial basis function. Proceeding of shape Modeling international 2004:31-39
[5]刘富强,卢赤班.数字视频监控系统及其应用[J]工矿自动化,2003,(3)：31-33.
[6]严杰.支持人脸检测的智能视频监控系统的设计与实现.重庆大学硕十论文.2008
[7]袁夏三维激光扫描点云数据处理及应用技术[D].南京理工大学,2006.
[8]齐秀丽.数据点云预处理技术研究[J].怀化学院学报,2010,29(5)：38-40.
[9]赵柳,马礼,杨银刚等.逆向工程中散乱点云数据精简研究[J].光电技术应用,2010,25(1)：60-63,70.
[10]肖琪.基于激光扫描数据点集的三维重建算法研究[D].华北电力大学(保定),2011.
[11]Markus Gross, Hanspeter Pfister. Point-Based Graphics[M]. Morgan Kaufmann publisher, 2007:P1-3.
[12]Kazhdan M, Bolitho M, Hoppe H. Poisson surface reconstruction[C]. Polthier K. Sheffer A. Symposium on Geometry Processing. Switzerland: The Eurographics Association. 2006:61-70.
[13]Dey T, Giesen. Hudson. Delaunay based shape reconstruction from large data[C]. Parallel and Large-Data Visualization and Graphics Proceedings. 2001:19-146.
[14]Ohtake Y, Belyaev A, Seidel H. A multi-scale approach to 3D scattered data interpolation with compactly supported basis functions. Proceeding of Shape Modeling international. 2003:153-161.
[15]Berg M D, Kreveld M V, Overmars M. Schwarzko'pf O.Computational Geometry Algorithms and Applications[M]计算儿何——算法与应用.邓俊辉 译.2005:5,252.
[16]C Bradford Barber, Univ. of Minnesota, Minneapolis, David P. Dobkin. The quick hull algorithm for convex hulls[J]. ACM Transactions on Mathematical Software. 1996, 22-4.
[17]王泉德.任意三角网格模型体积的快速精确计算方法[J].计算机工程与应用,2009,45(18)：32-34,58.
[18]张磊,史金飞,罗翔等.运动目标检测的差分图像法分析研究[J]工业仪表与自动化装置,2007,(3)：7-11,48.
[19]刘仁杰,姚星周Windows环境下多线程多串口通讯的实现[J].工业控制计算机,2011,24(5)：32-33.
[20]李勇.一个多串口多线程数据采集系统软件的设计与实现[J].微计算机信息,2006,22(16)：152-154.
[21]刘教瑜,吴美玲,谭杰等GPRS DTU的设计及研究[J].电力自动化设备,2006,26(3)：89-91.
[22]张开军,殷跃平,张作辰等.滑坡崩塌泥石流灾害详细调查规范研究[J].水文地质工程地 质,2008,35(6)：99-102.
[23]赵军,金晓峰,陈志强等.大学男生身体体积的测量与回归方程的优选研究[J].中国体育科技,2005,41(3)：132-133,136.
[2]Arman F., Aggarwal J.K.: Object recognition in dense range images using a CAD system as a model base. Robotics and Automation, 1990. Proceedings.,1990 IEEE International Conference on, 13-18 May 1990. vol.3,1858-1863.
[3]Delaunay, B.Sur la sphere vide, Izvestia Akademii Nauk SSSR. Otdelenie Matematicheskikh i Estestvennykh Nauk. 1934(7):793-800.
[4]Ohtake Y, Belyaev A, Seidel H. 2004. 3D scattered data approximation with adaptive compactly supported radial basis function. Proceeding of shape Modeling international 2004:31-39
[5]刘富强,卢赤班.数字视频监控系统及其应用[J]工矿自动化,2003,(3)：31-33.
[6]严杰.支持人脸检测的智能视频监控系统的设计与实现.重庆大学硕十论文.2008
[7]袁夏三维激光扫描点云数据处理及应用技术[D].南京理工大学,2006.
[8]齐秀丽.数据点云预处理技术研究[J].怀化学院学报,2010,29(5)：38-40.
[9]赵柳,马礼,杨银刚等.逆向工程中散乱点云数据精简研究[J].光电技术应用,2010,25(1)：60-63,70.
[10]肖琪.基于激光扫描数据点集的三维重建算法研究[D].华北电力大学(保定),2011.
[11]Markus Gross, Hanspeter Pfister. Point-Based Graphics[M]. Morgan Kaufmann publisher, 2007:P1-3.
[12]Kazhdan M, Bolitho M, Hoppe H. Poisson surface reconstruction[C]. Polthier K. Sheffer A. Symposium on Geometry Processing. Switzerland: The Eurographics Association. 2006:61-70.
[13]Dey T, Giesen. Hudson. Delaunay based shape reconstruction from large data[C]. Parallel and Large-Data Visualization and Graphics Proceedings. 2001:19-146.
[14]Ohtake Y, Belyaev A, Seidel H. A multi-scale approach to 3D scattered data interpolation with compactly supported basis functions. Proceeding of Shape Modeling international. 2003:153-161.
[15]Berg M D, Kreveld M V, Overmars M. Schwarzko'pf O.Computational Geometry Algorithms and Applications[M]计算儿何——算法与应用.邓俊辉 译.2005:5,252.
[16]C Bradford Barber, Univ. of Minnesota, Minneapolis, David P. Dobkin. The quick hull algorithm for convex hulls[J]. ACM Transactions on Mathematical Software. 1996, 22-4.
[17]王泉德.任意三角网格模型体积的快速精确计算方法[J].计算机工程与应用,2009,45(18)：32-34,58.
[18]张磊,史金飞,罗翔等.运动目标检测的差分图像法分析研究[J]工业仪表与自动化装置,2007,(3)：7-11,48.
[19]刘仁杰,姚星周Windows环境下多线程多串口通讯的实现[J].工业控制计算机,2011,24(5)：32-33.
[20]李勇.一个多串口多线程数据采集系统软件的设计与实现[J].微计算机信息,2006,22(16)：152-154.
[21]刘教瑜,吴美玲,谭杰等GPRS DTU的设计及研究[J].电力自动化设备,2006,26(3)：89-91.
[22]张开军,殷跃平,张作辰等.滑坡崩塌泥石流灾害详细调查规范研究[J].水文地质工程地 质,2008,35(6)：99-102.
[23]赵军,金晓峰,陈志强等.大学男生身体体积的测量与回归方程的优选研究[J].中国体育科技,2005,41(3)：132-133,136.