智能测量热成像系统MRTD的应用方法分析  

Analysis of intelligent MRTD testing for thermal imaging system

作　　者：孙军月[1] 马冬梅[1] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所,吉林长春130033

出　　处：《红外与激光工程》2006年第4期401-404,共4页Infrared and Laser Engineering

摘　　要：针对已提出的基于神经网络智能测量热成像系统最小可分辨温差(MRTD)的方法,简要阐述其实现方法、理论根据、神经网络算法模型以及测试结果分析。MRTD智能测量是建立在大量主观训练数据的基础上,影响主观MRTD测试的各因素也同样影响最终智能测量的准确性。综合考虑算法模型等因素,分析了影响智能MRTD测试精度的各因素以及尽量减小其影响应采取的措施。同时,讨论了智能MRTD测量的实际意义以及将其应用到实际测量中还需进一步研究与解决的问题。Aiming at the method based on intelligent MRTD （Minimum Resolvable Temperature Difference） testing for thermal imaging system using ANN, the realized process, theory foundation, model of neural network and testing deviation were introduced briefly. The intelligent MRTD testing was based on the subjective training data, so the factors affecting the subjective nature of the observers also influenced the accuracy of intelligent testing. Considering the errors of the arithmetic and model, the main factors were discussed in detail in order to decrease the influence on the accuracy of intelligent testing. At the same time, the signification of intelligent MRTD testing and the problems of further research were also mentioned.

关 键 词：最小可分辨温差 智能MRTD测量 BP神经网络 MRTD测试精度 红外成像系统参数测试与评估 

分 类 号：TN215[电子电信—物理电子学]