In order to solve the problem that the accuracy of infrared small target detection is difficult to improve due to the factors such as small target size and complex background,a method of infrared small target detection based on dual-stream learning framework was proposed.The segmented network was used for small target detection,and the super-resolution task was used as an auxiliary means,the shared feature attention mechanism(SFAM)was introduced to solve the feature loss problem in feature fusion and iteration.By conducting extensive experiments on 4 different scenarios on a public dataset,the proposed method scores better than other methods with an accuracy of 0.835.At the same time,the ablation study also confirmed the importance and feasibility of SFAM.
关键词
双流学习框架/红外小目标检测/超分辨率任务/共享特征注意力机制
Key words
dual-stream learning framework/infrared small target detection/super-resolution task/shared feature attention mechanism(SFAM)
维普
万方数据