Human vision-inspired detection for infrared small targets under scaly light background on sea surface

Abstract

The automatic detection and search of small targets by the infrared guidance system is challenged by the complex sea background environment, particularly the dense scaly light generated by the reflection of sunlight by the sea clutter and the appearance of boundaries. To address this issue, inspired by the significant sensitivity, classification perception, reverse exclusion, and association prediction characteristics of the human visual system, this paper proposes a human vision-inspired detection model for infrared small targets under scaly light background on the sea surface. Firstly, the local energy factor and multi-scale patch contrast measure are fused to construct a significance analysis model and enhance the contrast of the target region of interest. Then, the image is classified into different regions according to the detection of grayscale fluctuations and mutation points in the background, and a region self-classification perception model is constructed for adaptive segmentation of scene images. Next, the prior information of the target gradient is used to construct a reverse exclusion analysis model for suppressing the residual background of the boundary. Subsequently, by using the fluctuation law and characteristic difference of the spatial energy and size features of the target and the scaly light background in the time domain, the correlation prediction analysis model is constructed to further suppress the residual background. Finally, through the combination of the above four visual feature models, a new detector with good adaptability is proposed for small targets under scaly light background on the sea surface. The experimental results indicate that the average detection rate of our detector is 95.26%, and the number of false alarms in a single frame is reduced by 93.77% compared with the typical detectors.