Journal
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
Volume 59, Issue 11, Pages 9813-9824Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TGRS.2020.3044958
Keywords
Feature extraction; Object detection; Task analysis; Modulation; Data models; Semantics; Acceleration; Attention mechanism; deep learning; feature fusion; infrared small target; local contrast
Categories
Funding
- National Natural Science Foundation of China [61573183]
- Open Project Program of the National Laboratory of Pattern Recognition (NLPR) [201900029]
- Nanjing University of Aeronautics and Astronautics PhD Short-Term Visiting Scholar Project [180104DF03]
- Excellent Chinese and Foreign Youth Exchange Program of China Association for Science and Technology, China Scholarship Council [201806830039]
Ask authors/readers for more resources
In this paper, a novel model-driven deep network is proposed for infrared small target detection, which combines discriminative networks and conventional model-driven methods. By designing a feature map cyclic shift scheme and incorporating bottom-up attentional modulation, the network is able to encode long-range contextual interactions and preserve small target features effectively. Experimental results demonstrate that the proposed network outperforms other model-driven methods and deep networks, showing a performance boost in small target detection.
To mitigate the issue of minimal intrinsic features for pure data-driven methods, in this article, we propose a novel model-driven deep network for infrared small target detection, which combines discriminative networks and conventional model-driven methods to make use of both labeled data and the domain knowledge. By designing a feature map cyclic shift scheme, we modularize a conventional local contrast measure method as a depthwise parameterless nonlinear feature refinement layer in an end-to-end network, which encodes relatively long-range contextual interactions with clear physical interpretability. To highlight and preserve the small target features, we also exploit a bottom-up attentional modulation integrating the smaller scale subtle details of low-level features into high-level features of deeper layers. We conduct detailed ablation studies with varying network depths to empirically verify the effectiveness and efficiency of the design of each component in our network architecture. We also compare the performance of our network against other model-driven methods and deep networks on the open SIRST data set as well. The results suggest that our network yields a performance boost over its competitors.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available