This study investigates multispectral characteristics of an unmanned aerial vehicle (UAV) at different observation angles by experiment. The UAV and its engine are tested on the ground in the cruise state. Spectral radiation intensities at different observation angles are obtained in the infrared band of 0.9–15 μm by a spectral radiometer. Meanwhile, infrared images are captured separately by long-wavelength infrared (LWIR), mid-wavelength infrared (MWIR), and short-wavelength infrared (SWIR) cameras. Additionally, orientation maps of the radiation area and radiance are obtained. The results suggest that the spectral radiation intensity of the UAV is determined by its exhaust plume and that the main infrared emission bands occur at 2.7 μm and 4.3 μm. At observation angles in the range of 0°–90°, the radiation area of the UAV in MWIR band is greatest; however, at angles greater than 90°, the radiation area in the SWIR band is greatest. In addition, the radiance of the UAV at an angle of 0° is strongest. These conclusions can guide IR stealth technique development for UAVs.

Building segmentation is widely used in urban planning, disaster prevention, human flow monitoring and environmental monitoring. However, due to the complex landscapes and highdensity settlements, automatically characterizing building in the urban village or cities using remote sensing images is very challenging. Inspired by the rencent deep learning methods, this paper proposed a novel end-to-end building segmentation network for segmenting buildings from remote sensing images. The network includes two branches: one branch uses Widely Adaptive Spatial Pyramid (WASP) structure to extract multi-scale features, and the other branch uses a deep residual network combined with a sub-pixel up-sampling structure to enhance the detail of building boundaries. We compared our proposed method with three state-of-the-art networks: DeepLabv3+, ENet, ESPNet. Experiments were performed using the publicly available Inria Aerial Image Labelling dataset (Inria aerial dataset) and the Satellite dataset II(East Asia). The results showed that our method outperformed the other networks in the experiments, with Pixel Accuracy reaching 0.8421 and 0.8738, respectively and with mIoU reaching 0.9034 and 0.8936 respectively. Compared with the basic network, it has increased by about 25% or more. It can not only extract building footprints, but also especially small building objects.