首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于磁光调制的多光谱目标识别系统的研究
引用本文:闫晓燕,秦建敏,乔记平.基于磁光调制的多光谱目标识别系统的研究[J].光谱学与光谱分析,2016,36(3):864-867.
作者姓名:闫晓燕  秦建敏  乔记平
作者单位:1. 太原理工大学物理与光电工程学院,新型传感器与智能控制教育部重点实验室,山西 太原 030024
2. 中北大学仪器与电子学院,山西 太原 030051
基金项目:国家自然科学基金项目(60378019)
摘    要:基于特征光谱的目标识别技术具有检出能力强,可分辨目标种类等优点,但也存在一定的问题,即需要事先获取背景光谱作为先验知识且要求背景光谱随时间的变化较小。由此限制了其在新环境、复杂环境中实时目标识别方面的应用。设计了一种采用磁光调制配合特征光谱分析的技术手段,使目标识别过程中无需事先获取背景谱,从而实现了一次采集获取被测目标信息的功能,相比传统的目标检测方法而言,对战场的适应能力更强,具有较好的实用意义。同时,磁光调制技术有效地抑制了背景杂散光的干扰,从而提高了目标识别概率。由于磁光调制提供了目标光谱的累加迭代信息,故即使未知背景光谱或者背景光谱变化较大时,也可以通过目标光谱的迭代信息大幅提高目标识别率。针对不同被测目标的回波光强与背景光强值进行实验分析,结果显示,三种目标对调制线偏振光的反射能力明显强于背景。采用伪装色的被测目标对可见光成像目标识别影响很大,而调制偏振型系统仍能很好地识别目标。在此基础上,对0.5~2 km范围内的目标进行多特征波长目标种类识别。采用三个特征波长时,目标识别概率在2 km左右明显降低,采用四个或五个特征波长位置时,可以实现95.0%以上的目标识别概率,同时为了降低运算量提高系统的实时检测能力,最终采用四个特征波长。

关 键 词:目标识别  磁光调制  多光谱分析  识别概率    
收稿时间:2014-04-24

Research on Multi-Spectral Target Recognition System Based on the Magneto-Optical Modulation
YAN Xiao-yan,QIN Jian-min,QIAO Ji-pin.Research on Multi-Spectral Target Recognition System Based on the Magneto-Optical Modulation[J].Spectroscopy and Spectral Analysis,2016,36(3):864-867.
Authors:YAN Xiao-yan  QIN Jian-min  QIAO Ji-pin
Institution:1. Key Laboratory of Advanced Transducers and Intelligent Control System, Ministry of Education, College of Physics and Optoelectronics, Taiyuan University of Technology, Taiyuan 030024, China2. Institute of Measuring and Controlling Technology, North University of China, Taiyuan 030051, China
Abstract:The technology of target recognition based on characteristic multi-spectrum has many advantages, such as strong detection capability and discriminating capability of target species. But there are some problems, it requires that you obtain the background spectrum as a priori knowledge, and it requires that the change of background spectrum is small with time. Thereby its application of real-time object recognition is limited in the new environment, or the complex environment. Based on magneto-optical modulation and characteristic multi-spectrum the method is designed, and the target is identified without prior access to the background spectrum. In order to achieve the function of the target information in the one acquisition time for tested, compared to conventional methods in terms of target detection, it’s adaptability is better than before on the battlefield, and it is of more practical significance. Meanwhile, the magneto-optical modulator is used to suppress the interference of stray light background, thereby improving the probability of target recognition. Since the magneto-optical modulation provides incremental iterative target spectral information, therefore, even if the unknown background spectrum or background spectrum change is large, it can significantly improve the recognition accuracy of information through an iterative target spectrum. Different test targets back shimmering light intensity and background intensity values were analyzed during experiments, results showed that three targets for linearly polarized reflectance modulation is significantly stronger than the background. And it was of great influence to visible imaging target identification when measured target used camouflage color, but the system of polarization modulation type can still recognize target well. On this basis, the target range within 0.5 km×2 km multi-wavelength characteristics of the target species were identified. When using three characteristic wavelengths, the probability of target identification significantly reduced at around 2km, when using four or five characteristic wavelength position, the probability of target identification reach up to more than 95.0%. Meanwhile, in order to reduce the calculation and improve the real-time detection capability of the system, finally, four characteristic wavelengths was selected. So the system has a certain application value.
Keywords:Target recognition  Magneto-optical modulation  Multi-spectral analysis  Probability of recognition
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《光谱学与光谱分析》浏览原始摘要信息
点击此处可从《光谱学与光谱分析》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号