Abstract:

In recent years, Unmanned Aerial Vehicle (UAV) systems have played an important role in obtaining timely and accurate acquisition of geospatial information, providing an important basis for disaster assessment, emergency relief, and reconstruction. However, with the high-resolution characteristics of UAV image, it is necessary to use number of images to cover the study areas. Therefore, the traditional CPU-based image processing method cannot meet the requirements of disaster emergencies. In this paper, we propose a Compute Unified Device Architecture (CUDA)-based UAV image processing algorithm, which takes advantage of the powerful parallel processing capability of the GPU, greatly improving the efficiency of processing time. Our experiments confirm the applicability of UAV image for natural disaster emergency response capability.