地面场景红外图像特征分析  

Feature analysis of ground scene infrared image*

作　　者：左峥嵘[1] 张天序[1] 施长城[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所图像信息处理与智能控制教育部重点实验室,湖北武汉430074

出　　处：《红外与激光工程》2006年第3期371-375,共5页Infrared and Laser Engineering

基　　金：国家自然科学基金重点项目(60135020);国家重点预研资助项目(413010701)

摘　　要：为了研究并掌握场景中目标的红外图像变化规律,提高自动目标识别系统的可靠性和稳定性,对实际采集得到的地面场景典型气象条件下一天24h的长波红外图像进行分析,研究了不同地物类型对应的红外图像典型统计特征随时间和气象条件变化的规律。研究结果表明,图像各区域亮度的日时变化表现为近似余弦的变化规律,标准差和熵特征表现为单峰的变化规律,并且,各区域由于亮度变化不一致而存在对比度反转的现象。In order to understand the variation rule of the infrared image features, improve the reliability and stability of the automatic target recognition system, the characteristic of longwave infrared image of the ground scene acquired in typical weather condition through the whole day was analyzed, the variation rule of typical image statistic features along with the variation of time and weather was studied and summarized, The result shows that the brightness of the image regions is corresponding to the different objects changing in cosine curve approximately, and its standard deviation and entropy feature changed in mono-apex curve. Moreover, the contrast reverses because the brightness variation is inconsistent at the different regions.

关 键 词：长波红外 红外图像 图像特征 检测识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]