废弃矿山植被覆盖度无人机遥感快速提取技术

Remote sensing rapid extraction technology for abandoned mine vegetation coverage via UAV

在废弃矿山开展植被覆盖度监测对评价废弃矿山绿化修复效益具有十分重要的意义。本研究基于小型多旋翼无人机与地面控制站搭建了一套集成稳定云台、图像采集的废弃矿区无人机航拍影像系统。以2016年已进行生态修复治理的北京市房山区西苑四队煤矿治理区为研究区进行飞行实验,无人机飞行高度为120 m,航拍影像的地面分辨率为0. 012 m。采用将无人机图像由RGB颜色空间转换为HSV颜色空间的方式,并限定H值和S值的取值范围,提出了一种准确、快速提取废弃矿山植被覆盖度的方法。同时利用ENVI软件采用监督分类方法提取了同一幅无人机航拍影像的植被覆盖度作为真实值进行提取精度评价。研究结果表明,利用无人机低空航拍技术,可以实现高分辨率遥感影像的获取,将无人机航拍影像由RGB颜色空间模式转换为HSV颜色空间,限定S≥0. 2,H≥47. 1°对图像进行阈值分割,能够快速将植被部分提取出来并计算出植被覆盖度,且该方法提取误差不超过6. 8576%。

[Background] The excessive exploitation of mineral resources has caused the ecological environment of the mining area and its surrounding areas deteriorated severely,which has adversely affected human production and life. Vegetation restoration and reconstruction is an integral part of the ecological restoration work in mining area. Vegetation coverage monitoring is critical for mine greening restoration. [Methods] This study relied on field surveys and low-altitude aerial surveys of drones conducted in July 2018 to monitor the ecological rehabilitation benefits of an abandoned coal mine,Xiyuan Fourth Team Coal Mine Treatment Area in Fangshan district around Beijing,which underwent ecological restoration and control in 2016. In this study,a set of aerial image system for unmanned aerial vehicles( UAV) in the abandoned mining area with integrated stable gimbal and image acquisition was built based on a small multi-rotor drone and a ground control station. The drone flying height was 120 m,and the ground resolution of aerial images was 0. 012 m. An accurate and fast method for extracting the vegetation coverage of the abandoned mine was proposed by converting drone images from RGB color space to HSV color space and limiting the range of H and S values. At the same time,the vegetation coverage of the same drone aerial image was extracted using the ENVI software via the supervised classification method to evaluate the extraction accuracy. [Results] The drone aerial photography was used to achieve the acquisition of high-resolution drone aerial images of centimeter-level vegetation in the abandoned mining areas,low system cost,simple maintenance and high efficiency. By converting the visible spectrum drone aerial image from the RGB color space mode to the HSV color space,the segmenting threshold value by limiting the S ≥ 0. 2 and the H ≥ 47. 1 allowed to quickly extract the vegetation part and further calculate the vegetation coverage to achieve accurate extraction with an error≤6. 857 6%,and under the same aerial photography conditions,the threshold settings of the S and H values were stable. Compared with the vegetation coverage extracted by supervised classification,the extraction result based on the HSV color space threshold segmentation method was lower,and the extraction error was getting smaller and smaller with the increase of vegetation coverage. [Conculsions]This study proposes a new method of quickly extracting vegetation coverage using drone aerial images,which provides a new idea for evaluating the ecological restoration effect of abandoned mines with high accuracy and efficiency.