TFCD-Net: Target and False Alarm Collaborative Detection Network for Infrared Imagery

Siying Cao,Zhi Li,Jiakun Deng,Yi’an Huang,Zhenming Peng

DOI: https://doi.org/10.3390/rs16101758

IF: 5

2024-05-16

Remote Sensing

Abstract:Infrared small target detection (ISTD) plays a crucial role in both civilian and military applications. Detecting small targets against dense cluttered backgrounds remains a challenging task, requiring the collaboration of false alarm source elimination and target detection. Existing approaches mainly focus on modeling targets while often overlooking false alarm sources. To address this limitation, we propose a Target and False Alarm Collaborative Detection Network to leverage the information provided by false alarm sources and the background. Firstly, we introduce a False Alarm Source Estimation Block (FEB) that estimates potential interferences present in the background by extracting features at multiple scales and using gradual upsampling for feature fusion. Subsequently, we propose a framework that employs multiple FEBs to eliminate false alarm sources across different scales. Finally, a Target Segmentation Block (TSB) is introduced to accurately segment the targets and produce the final detection result. Experiments conducted on public datasets show that our model achieves the highest and second-highest scores for the IoU, Pd, and AUC and the lowest Fa among the DNN methods. These results demonstrate that our model accurately segments targets while effectively extracting false alarm sources, which can be used for further studies.

environmental sciences,imaging science & photographic technology,remote sensing,geosciences, multidisciplinary

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the challenges faced when detecting small targets in infrared images. Specifically, these challenges include small - target detection in complex backgrounds (usually smaller than 9×9 pixels or accounting for less than 0.15% of the field of view), and how to effectively distinguish real targets from false alarm sources in the background. Existing methods mainly focus on modeling targets and often overlook the processing of false alarm sources. This leads to poor detection results of small targets in dense and cluttered backgrounds, because these backgrounds often contain elements that reflect solar radiation, whose patterns are similar to those of targets, increasing the detection difficulty.

To address these challenges, the authors propose a new framework named TFCD - Net (Target and False Alarm Collaborative Detection Network). This network estimates potential false alarm sources by introducing a specialized False Alarm Source Estimation Block (FEB), and performs feature fusion through multi - scale progressive up - sampling to eliminate false alarm sources at different scales. Subsequently, the Target Segmentation Block (TSB) is used to accurately segment targets and generate the final detection results. This method not only improves the accuracy of target detection, but also effectively extracts information about false alarm sources, which can be used for further research.

The key contributions of the paper include:
1. Proposing a framework that models targets and false alarm sources simultaneously, aiming to solve the detection challenges in complex and cluttered backgrounds while maintaining the interpretability of the model.
2. Introducing a specialized FEB to estimate potential false alarm sources. By integrating multiple FEBs in the network, false alarm sources can be gradually estimated and eliminated at multiple scales, which not only improves the accuracy of the method, but can also be used as a pre - processing module to suppress false alarm sources in other ISTD algorithms.
3. Extensive experiments on public datasets have verified the effectiveness of this model. It can not only accurately detect targets, but also produce multi - scale false alarm source estimation results. These estimation results are not just accidental by - products, but can be used to construct a false alarm source dataset, contributing to further research in this field.