Airport Aircraft Target Detection Based on Space Spectrum Feature Fusion

Due to the complexity of airport background, the traditional method of aircraft target detection usually brings a lot of false alarms or missed detection. In this paper, the full convolution network is used to extract spatial features, which are combined with spectral features, and the active learning method is used to select the hyperspectral image target detection algorithm of training samples. By combining the spectral characteristics of pixels and the spatial correlation between adjacent pixels, the comprehensive features which can reflect the spatial spectral joint characteristics of pixels are extracted, and the expression ability of pixel features is improved. The experimental results on multiple data sets show that the proposed method is suitable for the detection of small and weak targets with certain structural information, and has a good effect on the detection of aircraft targets in airport background.