高级检索
    贾道祥, 刘鹏举, 张英凯, 刘长春, 孙永明. 基于地形匹配的图像烟火定位方法研究[J]. 北京林业大学学报, 2018, 40(6): 19-29. DOI: 10.13332/j.1000-1522.20170439
    引用本文: 贾道祥, 刘鹏举, 张英凯, 刘长春, 孙永明. 基于地形匹配的图像烟火定位方法研究[J]. 北京林业大学学报, 2018, 40(6): 19-29. DOI: 10.13332/j.1000-1522.20170439
    Jia Daoxiang, Liu Pengju, Zhang Yingkai, Liu Changchun, Sun Yongming. Smoke and fire positioning method in the image based on terrain profile matching[J]. Journal of Beijing Forestry University, 2018, 40(6): 19-29. DOI: 10.13332/j.1000-1522.20170439
    Citation: Jia Daoxiang, Liu Pengju, Zhang Yingkai, Liu Changchun, Sun Yongming. Smoke and fire positioning method in the image based on terrain profile matching[J]. Journal of Beijing Forestry University, 2018, 40(6): 19-29. DOI: 10.13332/j.1000-1522.20170439

    基于地形匹配的图像烟火定位方法研究

    Smoke and fire positioning method in the image based on terrain profile matching

    • 摘要:
      目的视频监控是林火监测的主要手段,利用数字云台获取方位角、俯仰角等参数来计算林火位置是目前林火定位的主要方式,而林火定位中常用的方法主要有单点定位、双点定位和多点定位3种方法。目前已有许多研究根据光线跟踪算法,依靠视频图像与三维地形的对应关系实现了三维场景的增强表达,为基于视频图像的定位提供了新的途径。当前,林区普遍通过建立视频监控、数字云台、GIS技术三者紧密结合的软、硬件联动系统进行森林火灾的监测、决策和扑救。本文将视频图像与虚拟地形匹配原理,应用于基于GIS的林火视频监控系统,提出基于图像的林火定位方法,并对定位精度进行分析评估。
      方法利用地形匹配原理建立林火实景图像与DEM之间的坐标映射关系,针对图像烟火区域选定像元,生成在DEM中对应的区域,并计算和分析该区域中心点的坐标、欧氏距离、方位角、俯仰角以及该区域的形状、可视性、面积、跨沟谷数等特征信息,提出定位区域精度信息分析与评估流程。该流程根据分析结果将定位区域分为可定位区与不可定位区,同时提供详细的烟火位置信息。
      结果选择北京九龙山自然保护区作为研究区域,对提出的定位方法进行了初步验证,结果表明该方法能提供准确定位精度信息,有助于快速找到火源,实现快速扑救。
      结论本文基于地形图像匹配的林火定位方法,充分利用了虚拟地形与视频图像的特征匹配关系,为林火快速扑救提供丰富的位置信息。该方法的定位精度只与图像分辨率及虚拟地形与实景图像匹配精度有关,避免了传统云台定位受硬件性能指标的限制;同时该方法也适用于手机等移动设备所拍摄的图像,只需获知拍摄点坐标,就可以实现对所采集图像上目标的定位,对于基于图像的定位分析具有重要意义。

       

      Abstract:
      ObjectiveVideo monitoring system is widely used in forest fire detection, in which forest fire position is calculated by cameral coordinate and PTZ parameters such as azimuth, pitch angle. There are three common methods for forest fire positioning, i.e. single-point positioning, double-point positioning and multipoint positioning. At present, some researches have been carried out to enhance the expression of three-dimensional scene by mapping between video images and three-dimensional virtual terrain based on the ray tracing algorithm, which can be applied to smoke and fire positioning in the image. Nowadays, many forest fire monitoring systems have been built up for monitoring, decision-making and fire-fighting, which integrating the hardware and software system including video surveillance, PTZ camera and GIS system. This paper subjects to take advantage of the terrain analysis in GIS and image process technique to match video images to virtual terrain, puts forward a smoke and fire positioning method based on image, and evaluates the positioning accuracy.
      MethodFirstly, a virtual terrain was generated from DEM by the camera location and view of field, and forest fire image was mapped to the virtual terrain based on the principle of terrain profile matching. Each pixel in the image can be projected to an area on DEM as the positioning area, and the information of the positioning area can be calculated and analyzed, such as the coordinate, Euclidean distance, azimuth, angle of pitch of center point, shape, area, visibility, the number of crossing the valley of the area.Then the accuracy information of the area was analyzed and a decision was made according to visibility and number of crossing the valley, which divided the area into two classes, one can be positioned accurately and the other can not be positioned accurately.
      ResultIn this paper, Jiulong Mountain Nature Reserve in Beijing was selected as the test area, and the proposed positioning method of smoke and fire was verified. The results showed that this method can provide location information in detail, and gave an evaluation about the positioning accuracy, which can help to locate the fire source quickly and make a quick response.
      ConclusionIn this paper, the positioning method of smoke and fire makes full use of the matching relationship between virtual terrain and video images, provides abundant location information for forest fire fighting. This method performs better than the traditional PTZ positioning which depends on hardware performance specifications, its positioning accuracy is only related to the resolution of image and the matching accuracy between virtual terrain and real image. At the same time, the method is also applicable to images taken by mobile devices such as smart phones. Locating the coordinates of the shooting point, we can realize the positioning of the targets in the image, which has important significance for the image-based positioning analysis.

       

    /

    返回文章
    返回