An optimised multi-scale fusion method for airport detection in large-scale optical remote sensing images. (2nd April 2020)

• Record Type: Journal Article
• Authors: Yin, Shoulin; Li, Hang; Teng, Lin; Jiang, Man; Karim, Shahid
• Is Part Of: International journal of image and data fusion. Volume 11:Number 2(2020)
• Page Start: 201
• Page End: 214
• Publication Date: 2020-04-02
• Subjects: Airport detection -- discrete wavelet -- multi-scale fusion -- remote sensing -- region selection; Image processing -- Periodicals; Multisensor data fusion -- Periodicals; Multisensor data fusion; Periodicals; 621.36705
• DOI: 10.1080/19479832.2020.1727573
• Languages: English
• ISSNs: 1947-9832

Abstract

ABSTRACT: Airport detection in remote sensing images is an important process which plays a significant role in military and civil areas. Mostly, conventional algorithms have been used for airport detection from a small-scale remote sensing image and revealed the less efficient ability of searching the object from a large-scale high-resolution remote sensing image. The computational complexity of these algorithms is high and these are not useful for rapid localisation with high detection accuracy in high-resolution remote sensing images. Aiming to solve the above problems, we propose an optimised multi-scale fusion method for airport detection in large-scale optical remote sensing images. Firstly, we execute discrete wavelet multi-scale decomposition for remote sensing image and extract the multiple features of the object in each sub-band. Secondly, the fusion rule based on the optimised region selection is used to fuse the features on each scale. Meanwhile, singular-value decomposition (SVD) is utilised for fusing low-frequency and principal component analysis (PCA) is utilised to fuse the high-frequency, respectively. Thirdly, the final-fused image is acquired by weighted fusion. Finally, the selective search method is employed to detect the airport in the fused image. Experimental results show that the detection accuracy is better than the other state-of-the-art methods.