一种基于兴趣区提取的红外搜索系统目标实时检测算法

针对周扫红外搜索系统对空目标探测面临的图像数据量大、弱目标检测概率低、虚警率高等难点问题，提出了一种基于兴趣区(ROI)提取的目标实时检测算法。算法分析了周扫红外搜索系统获取的图像中目标与背景的特性，根据目标运动特性与灰度特性，在周扫红外搜索系统获取的整幅全景图像中快速提取目标可能存在的兴趣区；针对兴趣区内的局部目标图像切片，进一步精细检测识别，剔除虚假目标干扰。外场试验获取的实测数据目标检测结果表明，算法针对复杂低空背景下弱目标能够实现低虚警率稳健检测，已应用到了周扫红外搜索跟踪系统的工程样机研制中。

A Real-Time Target Detection Algorithm for Infrared Search and Track System Based on Region of Interest Extraction

With regard to the difficulties confronted in the aerial target detection of circumferential scan infrared search and track system, such as large image data quantity, low detection probability of weak target, high false detection rate and so on, a real-time target detection algorithm is proposed based on region of interest (ROI) extraction. It extracts the ROI of the suspected targets by quick real-time algorithm in the whole panorama image, based on the high frequency and movement characteristics of the target pixels. And then, focusing on the suspected target sliced images of ROI, it has further delicate detection and recognition to exclude those false jamming. The detection result of the test images shows, the algorithm has realized stable detection with low-rate false alarm for distant dim targets, and has been applied to the engineering sample of the panorama infrared search and tracking system.