基于无人机红外光谱技术的影像数据匹配方法研究

无人机加载红外光谱载荷对区域内影像进行获取现已成为遥感领域一种重要的技术手段，可通过对携带位置信息的影像进行分类提取，得到植被盖度、温度指数等一系列因子指标。利用FREE BIRD(自由鸟)小型低空无人机系统挂载Tetracam红外相机(310万像素)对新疆玛纳斯县一河道进行影像获取。无人机飞行面积约为20.5 km2，为了得到更加精确的植被、温度等因子，需要对无人机红外影像进行配准，通过优化SIFT匹配参数和RANSAC粗差剔除后，获取了可靠的匹配结果，即经过算法匹配后的影像与原影像进行了误差比对，能够满足后期的应用需要, 这也是本文的创新点之一。将影像进行配准后进行二维影像拼接，将多张红外影像按照航向重叠度不低于60%，旁向重叠度不低于50%的概率进行拼接，得到拼接后的红外影像图。另外比较了SIFT和SUFT两种算法，利用优化的SIFT算法及改进的FLIR传感器获取1 600张热红外影像，利用地面同步测量数据对拼接后的红外影像进行算法匹配，并利用ENVI(完整的遥感图像处理平台)软件进行温度及植被盖度的影像反演，得到了研究区域的单一影像及红外影像的温度反演图及植被反演图。通过对两种算法的对比得到更加优化的算法模型，并对该模型进行回归分析和精度检验，得到该模型的相关系数R2为0.767，匹配精度为81.51%，模型精度较高。本模型的建立对日后无人机红外影像的配准及提取反演奠定了理论和实践基础。

Research on Image Data Matching Method Based on Infrared Spectrum Technology of UAV

UAV loading infrared/near-infrared spectroscopy on regional image acquisition load has become an important field of remote sensing technology, through classifying the position information of the portable image, and getting the vegetation cover, temperature index and a series of factors. In this paper, we used FREE BIRD low altitude unmanned aerial vehicle (UAV) to mount Tetracam- infrared camera (3 million 100 thousand pixels) to get the image of a river in Xinjiang, Manasi. In order to get more accurate vegetation temperature and other factors, we needed UAV infrared/near infrared image registration, through the optimization of SIFT, detection of outliers and RANSAC parameters, to obtain reliable matching results. After the matching algorithm of the image ,the original image of the error ratio were below 60%, which was one of the innovations of this paper to meet the needs of the application. After registering the images, the images were spliced, and the infrared images were spliced according to the degree of overlap of the course of not less than 60%, while the probability of the adjacent overlap was not less than 50%. At the same time, this paper compared the SIFT and SUFT two kinds of algorithms, using FLIR sensor SIFT algorithm and improved optimization to obtain 1 600 thermal infrared image matching and image inversion of ground by utilization of synchronous measurement data. We used ENVI software to carry out the inversion of vegetation coverage temperature inversion and inversion vegetation map to get the single image and the infrared image of the study area. The algorithm model is more optimized through the comparison of the two algorithms, while the model of regression analysis and test of accuracy, the correlation coefficient R2 of the model is 0.767, and accuracy is 81.51% with higher model precision. This model provides theoretical and practical basis for the registration and extraction of inversion of UAV infrared image.