Edge-Aided Multiscale Context Network for Infrared Small Target Detection
Xinyu Ma, Yang Li
Abstract
Infrared small target detection technology has been widely used in various fields. Infrared small target usually shows the characteristics of low gray values and a lack of texture information in complex scenes, which makes it difficult to accurately segment the boundary of infrared small target. To solve this problem, we propose an edge-aided multiscale context network (EAMCNet) in this letter. In the proposed network, we design a two-stream architecture for target detection that considers edge information as a separate processing branch, the edge detection stream processes information in parallel to the target segmentation stream. To emphasize features for different scale of targets, we introduce a gate-based multiscale context information extraction (GMCIE) module to regulate contextual features transmission. Finally, edge features and semantic features are fused by feature fusion module to make full use of their complementarity. Experiments on the SIRST dataset show that the proposed method can achieve excellent performances compared with the state-of-the-art methods.