TY  - JOUR
T1  - Development of an Integrated System for Real-time EO/IR Image Restoration and Object Recognition in Battlefield Environments
AU  - Lee, Kwangil 
AU  - Kim, Jaehwan 
AU  - Lee, Chang-Eun 
JO  - The Journal of Korean Institute of Communications and Information Sciences
PY  - 2026
DA  - 2026/1/1
DO  - 10.7840/kics.2026.51.1.184
KW  - Real-time
KW  - Object restoration
KW  - Object recognition
KW  - EO/IR images
KW  - Transformer
KW  - Battlefield virtual
environment simulator
KW  - Battlefield environment dataset
AB  - Modern military technology demands real-time situational awareness capabilities in highly dynamic and
unpredictable battlefield environments. In this paper, we propose an integrated image restoration and recognition
system that restores and recognizes battlefield objects occluded by obstacles in real time. The proposed system
consists of a restoration module based on the decoupled spatial-temporal transformer (DSTT) and a recognition
module based on RT-DETR. To achieve spatiotemporal continuity and fast inference speed of the restoration
module, the online restoration and memory bank techniques are applied. In addition, the recognition module is
optimized using TensorRT to enable high-speed processing. Finally, a dataset is constructed using a battlefield
virtual environment simulator, and experimental results are presented to verify the feasibility of the proposed
system by showing improved recognition accuracy and inference speed of more than 20 fps, which are the
main goals of this paper.