Infrared Maritime Target Detection Based on Iterative Corner and Edge Weights in Tensor Decomposition

Abstract

High-precision infrared maritime target detection plays an important role in early warning, monitoring, search, and rescue. The methods of decomposing the original image into low-rank background components and sparse target components show favorable detection performance. However, the strong edge interference is also sparse and may be mistakenly taken as target components, resulting in a large number of false alarms and reducing detection accuracy. To solve the problem, we propose an iterative corner and edge weights method based on tensor decomposition. The original image is decomposed into the background component, target component, and additional strong edge interference component. The corner strength is designed as the weight of the target component, and the edge strength is designed as the weight of the interference component in order to separate the target and interference more accurately. And the two weights are designed to be updated during each iteration of the model solution to reduce the impact of initial imprecise weights on detection results. Compared with 8 advanced baseline methods in 10 datasets, the proposed method demonstrates outstanding results and shows engineering application prospects.