Abstract
Aerial infrared point target detection under nonstationary background clutter is a crucial yet challenging issue in the field of remote sensing. This paper presents a novel omnidirectional multiscale morphological method for aerial point target detection based on a dual-band model. Considering that the clutter noise conforms to the Gaussian distribution, the single-band detection model under the Neyman-Pearson (NP) criterion is established first, and then the optimal fused probability of detection under the dual-band model is deduced according to the And fusion rule. Next, the omnidirectional multiscale morphological Top-hat algorithm is proposed to extract all the possible targets distributing in every direction, and the local difference criterion is employed to eliminate the residual background edges further. The dynamic threshold-to-noise ratio (TNR) is adjusted to obtain the optimal probability of detection under the constant false alarm rate (CFAR) criterion. Finally, the dim point target is extracted after dual-band data correlation. The experimental result demonstrates that the proposed method achieves a high probability of detection and performs well with respect to suppressing complex background when compared with common algorithms. In addition, it also has the advantage of low complexity and easy implementation in real-time systems.
Highlights
Dim point target detection under complex background is a key technology in numerous fields, including infrared search and track (IRST) systems, terminal guidance, external intrusion warnings, and medical monitoring [1,2,3]
The overall detection level is determined by the to-noise ratio (TNR) of the two detectors, and the optimal fused probability of detection can be obtained by iterating TNR1 under the dual-band model
By using the iterative method proposed above, we simulate the general situation, namely, fixing the signal-to-noise ratio (SNR) of one channel, by changing the TNR of the other channel to obtain the curve of the optimal fused probability of detection and compare it with the single-band model further
Summary
Dim point target detection under complex background is a key technology in numerous fields, including infrared search and track (IRST) systems, terminal guidance, external intrusion warnings, and medical monitoring [1,2,3]. To reduce the complexity and the false alarms, the multi-label generative Markov random field (MRF) model was proposed to realize background suppression and target enhancement [18], which performs a better effect for the point target in a larger size. Yu et al [31] proved that the probability of detection based on dual-band optimization is obviously better than that of any single-band detector relying on the NP criterion, and the traditional morphological Top-hat algorithm was adopted to simulate man-made targets for detection fusion. For the dual-band model, multiscale Top-hat transform could be optimized by the omnidirectional structural element to achieve better background suppression. We propose an omnidirectional morphological filtering for point target detection based on an infrared dual-band model.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
