低空无人机视频影像可量测立体模型构建技术

Measurable Stereo Model Generating Technology Based on LowAltitude UAV Video Images

提出了一种利用无人机快速获取的视频数据的可量测立体模型构建方法。首先,通过无人机飞行平台快速获取视频数据,基于改进平板摄像机智能标定技术,获取摄像头的内方位元素,结合摄像机内方位参数、视频帧率、飞行高度、速度、影像分辨率等参数实现了基于实时视频流的关键帧影像自适应自动提取算法,并对提取关键帧影像进行畸变差改正;其次,通过对校正后的影像进行光束法空中三角测量加密处理,计算视频关键帧影像的外方位元素信息。最后,利用北京地区的试验数据进行了验证分析。结果表明利用能快速实现可量测立体模型构建,具有较高的量测精度,能有效的提高应急测绘条件下的灾情地理信息获取,为灾情评估和辅助决策提供依据。

This paper proposes a method for generating measurable stereo model based on unmanned aerial vehicle （UAV） video images.Firstly,inner orientation elements of the cameras are got based on the improved plate smart camera calibration technology. Combining the inner orientation elements, video frame rate, flight altitude, flight speed, and image resolution etc, the key frame images are adaptively extracted and the distortion is corrected from the real-time video streaming. Secondly, based on distortion correction key frame images, the calculation of the video key frames image information of exterior orientation elements is conducted. Finally, the test data of Beijing area is verified, and the results show that using the video key frames can quickly generate the Measurable Stereo Model, with high measurement accuracy, and can effectively increase the surveying and mapping capability for the geographical information of the disaster under the emergency condition, to provide the basis for the disaster evaluation and auxiliary decision-making.