结合DEM的红边-近红外植被指数提取城市植被信息

随着生活水平的不断提高，城市植被已成为衡量城市宜居性的重要标准之一，对城市生物多样性评估和保护起到非常重要的作用。因此，合理规划城市植被是解决环境问题和提高生活质量的重要手段。因此，城市植被的提取和监测成为重中之重的任务。目前，城市植被提取一方面受到地域和物种的影响，另一方面也受到地形和建筑物阴影的影响。为解决上述问题，提出了一种结合数字高程模型(DEM)的红边-近红外植被指数模型（RENVI）。首先选取了3景经过辐射定标和大气校正的具有红边波段、且光谱和空间分辨率较高的Worldview-3遥感影像；然后，根据红边波段对于植被具有较高的敏感性，且红边范围内的光谱数据与反映植被生长状况的参数有较好的相关关系原理，采用DEM模型和红边波段光谱差异，有效去除地形和建筑物阴影；最后，在可见光波段范围内建立红边光谱-近红外光谱构建特征空间，构建了红边-近红外植被指数模型，同时与归一化植被指数（NDVI）和增强型植被指数（EVI）进行城市植被提取的定性和定量对比分析。定性分析是利用真实植被影像参考图与模型提取植被影像进行视觉分析；后者是采用用户精度、生产者精度、总体精度和Kappa系数进行量化分析。定性分析表明：NDVI和EVI提取城市植被，由于建筑和道路像元混淆在植被中，产生了错分和漏分的问题。RENVI较好地消除了阴影像元与植被像元混淆问题，能准确的提取城市植被，减少了冗余度，增加了植被指数的信息量。定量分析表明：RENVI模型较NDVI和RVI能够准确提取城市植被，3景影像总体精度分别为89%，81.4%和91.8%，Kappa系数分别为0.852 8，0.791 3和0.905 2。综上所述，该方法有效提高了城市植被提取精度，并取得了较好的提取视觉效果。

Combining the Red Edge-Near Infrared Vegetation Indexes of DEM to Extract Urban Vegetation Information

With the continuous improvement of living standards, residents’ requirements for urban vegetation are also increasing. Urban vegetation has become one of the important criteria to measure the livability of cities and plays a very important role in assessing and protecting urban biodiversity. Therefore, rational planning of urban vegetation is an important means of solving environmental problems and improve the quality of life. To sum up, monitoring urban vegetation becomes the main task, and the extraction of urban vegetation becomes the top priority. At present, the problems of urban vegetation extraction mainly focus on two aspects. Vegetation extraction is affected by region and species. On the other hand, Vegetation extraction is affected by topography and the shadow of buildings. In order to solve the above problems, this paper proposes a red edge-near infrared vegetation index model based on DEM. In this experiment, worldView-3 remote sensing images with red-edge bands and high spectral and spatial resolution after radiation calibration and atmospheric correction were first selected. Then, according to the high sensitivity of the Red Edge band to vegetation and the good correlation between the spectral data within the red edge and the parameters reflecting vegetation growth, the DEM model and the spectral difference between the red edge were adopted to remove the shadow of terrain and buildings effectively. Finally, the red-border spectrum-near-infrared spectrum is constructed based on the feature space within the visible band, and the red-border near-infrared vegetation Index model is constructed. At the same time, the urban vegetation extraction is compared and analyzed with NDVI and EVI. The analysis methods are qualitative and quantitative. The former is to extract vegetation images for visual analysis by using a real vegetation image reference map and model. The latter is a quantitative analysis using user accuracy, producer accuracy, overall accuracy and Kappa coefficient. The result of the qualitative experiment shows that the DEM model can effectively remove the shadow of buildings and terrain by combining with the different information of the red edge band between shadow and vegetation. After removing the shadows, NDVI and EVI were used to extract urban vegetation from the images, which made the buildings and road pixels confused in the vegetation, resulting in the problem of misclassification and omission. However, RENVI can effectively eliminate the confusion between shadow pixels and vegetation pixels, accurately extract urban vegetation, reduce redundancy, and increase vegetation index information. The quantitative experimental results show that the RENVI model can accurately extract urban vegetation compared with NDVI and RVI. The overall accuracy of the 3 images is 89%, 81.4% and 91.8% respectively, and the Kappa coefficient is 0.852 8, 0.791 3 and 0.905 2 respectively. In summer, this method can effectively improve the extraction precision of urban vegetation and obtain a better visual effect of extraction.