Optimal Gabor sub-band-based spectral kurtosis and Teager Kaiser energy for maritime target detection in SAR images

Detecting and tracking maritime targets through Synthetic Aperture Radar (SAR) images is an important application in automated maritime surveillance systems. Since these systems require processing a large amount of Ocean SAR data degraded by various types of distortions, providing an efficient method with lower computational complexity is a challenge. In addition to providing a comprehensive methodology to identify potential target regions for region proposal, this paper presents an efficient method to detect final ship targets. Initially the Optimal sub-band of the Gabor wavelet transform is selected based on BAT optimization on Fisher Cost. Furthermore, the potential regions of target are identified based on spectral kurtosis computed using selected sub-band of the Gabor wavelet transformed image. Finally, Teager Kaiser Energy operator is applied on selected sub-bands on selected tiles to detect target with perfect spatial resolution. This approach achieves 50% reduction in computational complexity with comparable False alarm rate and detection rate than the conventional methods.