一种高效的高分辨率遥感影像飞机目标检测方法

This paper proposes an efficient method for airplane and airport detection in high-resolution remote sensing images based on a deep learning algorithm. Firstly, it pre-processes large scale remote sensing images secondly, it utilizes salience detection and LSD (line segment detector) method to get airport candidate regions through linear probability graph, parallel linear filtering and clustering. Thirdly, it takes advantage of CFF (circle-frequency filter) localizing airplane regions, and finally, use CNN (convolutional neural network) module to get the accurate position of each airplane and combines airport detection with airplane detection to an integrated system. The results indicate that precision of our proposed method can reach to 99%.