研究目的
Aiming at the difficulty of infrared target detection of 'low and slow small' unmanned aerial vehicles (UAV) in complex low-altitude background.
研究成果
The proposed algorithm effectively detects 'low and slow small' UAV targets in complex low-altitude backgrounds with low false alarm rates, achieving real-time processing at 100Hz on specified hardware, and has been successfully applied in a circumferential scanning infrared search system.
研究不足
The algorithm's performance may be sensitive to target size and speed variations, and it relies on specific hardware (TI 6657 DSP) for real-time processing, which could limit applicability to other systems.
1：Experimental Design and Method Selection:
The algorithm combines spatial multiscale decomposition filtering (using Robinson filter and morphological filter) and temporal multiscale difference processing to detect UAV targets in infrared images.
2：Sample Selection and Data Sources:
Infrared image sequences of UAV targets in low-altitude and complex sky backgrounds acquired by an infrared search system (IRST).
3：List of Experimental Equipment and Materials:
TI 6657 DSP for real-time processing, mid-wave infrared images with 640*512 resolution.
4：Experimental Procedures and Operational Workflow:
The algorithm consists of three modules: spatial multiscale processing, fusion filtering target detection and confirmation, and temporal multiscale processing. Steps include multiscale decomposition, Robinson and morphological filtering, adaptive threshold detection, target confirmation based on features, and temporal difference processing with track correlation.
5：Data Analysis Methods:
Adaptive threshold segmentation, constant false alarm threshold method, and trajectory association for false alarm elimination.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
