A Systematic Scheme For Automatic Airplane Detection From High-Resolution Remote Sensing Images

Airport and airplane are typical objects in remote sensing research field. However, there are rare methods to detect airport and airplane in a unit system. In this paper, we propose a systematic scheme for airport detection and airplane detection from high-resolution remote sensing images. The airport detection part is mainly based on the parallel line features of runway, containing six main stages: down-sampling, Frequency-Tuned (FT) saliency detection, Line Segment Detector (LSD) line detection, line growing, parallel lines detection and line clustering. The airplane detection part is mainly based on Circle Frequency Filter (CF-filter) and a Fast R-CNN deep learning model. Experimental results on 500 high-resolution remote sensing images acquired more than 95% accuracy, and the average detection time was about 14 s, which proved that the proposed system was effective and efficient.