红外热成像图像反应物体温度信息, 受环境条件影响较少, 对于特定条件下的夜间安防监控、行车辅助、航运、**侦查等方面具有很强应用价值。近年来使用人工智能对图像中目标检测与识别技术发展突飞猛进, 广泛应用于各个领域。本文提出了一种结合红外热成像图像处理技术与人工智能目标识别技术的夜间目标识别方法。实时采集热成像视频进行预处理, 增强其对比度与细节, 使用基于深度学习技术的最新目标检测框架 YOLOv3对采集处理后的热成像图像中特定目标进行检测, 输出检测结果。测试结果表明, 该方法对于夜间目标识别率高、实时性强, 结合了红外热成像夜间监测和人工智能目标检测的优势, 对于夜间的目标识别、跟踪技术具有重大应用价值。

Infrared thermal images reflect object temperature information that is less affected by environmental conditions. They have strong application value for nighttime security monitoring, driving assistance, shipping, military investigation, and other aspects, under certain conditions. In recent years, artificial intelligence has been used in the development of target detection and recognition technology in imaging and various fields. This paper proposes a nighttime target detection method combining infrared thermal imaging image processing and artificial intelligence target detection. Thermal imaging videos are acquired in real time for pre-processing in order to enhance the contrast and details of the thermal images, and the latest target detection framework, YOLOv3, based on deep learning is utilized to detect specific targets in the acquired thermal images and subsequently output the detection results. The test results show that the proposed method has high recognition rate and desirable real-time performance at nighttime; it combines the advantages of infrared thermal imaging nighttime monitoring and artificial intelligence target detection. Furthermore, it has been demonstrated that tracking technology has great application in nighttime target recognition.