目标红外多波段特性仿真与处理

发布时间：2018-03-24 11:40

本文选题：点目标　切入点：红外多波段　出处：《中国科学院研究生院（光电技术研究所）》2015年硕士论文

【摘要】：目标的检测与识别问题是各个国家一直以来致力于解决的关键问题。伴随着空间探测技术的飞速发展,很多国家已能实现目标的准确识别,但是识别效能不高。所以为了缩短识别时间,他们都希望在尽可能远的距离外就能识别出真正的目标。此时,由于距离探测器较远,目标呈斑点状,无形状信息可言,传统的图像检测、识别和跟踪技术不再有效。而且实际情况中,目标周围还伴随有很多和目标特征类似的假目标。对于距离较远的目标,可见光探测和雷达探测已失去优势,利用红外探测技术可以有效地区分真假目标。然而由于光电技术的日趋成熟,假目标已经可以和真目标做得非常相似,使得区分真假目标变得越来越困难。所以,使用单一波段的红外传感器得到的辐射信息已不能准确地区分两者。因此,为了提高对点目标的检测识别能力,很多国家开始研究目标的红外多波段特性,综合利用真假目标多个波段的辐射信息进行区分识别。本文立足以此,研究点目标和伴随假目标在红外波段的辐射特征以及两者在辐射强度和辐射波谱方面的差异,并从中提取有用的识别特征、设计算法进行分类识别。论文的研究工作主要包括以下几个方面:(1)目标的温度特性:建立了真假目标的模型;分析了目标的时空关系和内外受热情况;构建并求解了目标的热平衡方程;比较了目标和三种假目标的温度特性。(2)目标的红外辐射特性:计算了理想情况下探测器接收到的目标辐射;讨论了目标反射的外界辐射对探测器接收到的目标总辐射的影响;分析了大气传输对目标红外辐射的衰减作用;研究了不同的材料发射率对目标红外辐射的影响。(3)目标的波段特性分析:分析并比较了探测器接收到的真假目标辐射;采用控制变量法讨论了太阳位置、轨道形状、轨道倾角、表壳厚度比、表面涂层、壳体材料和内热等因素对真假目标辐射特性及其差异特征的影响;基于真假目标的差异特征将研究波段进行了波段划分。(4)目标的多波段处理:基于真假目标的波段辐射特征和波段辐射变化率特征,提出了两种基于人工统计的识别算法;采用费歇尔判别分析法和反向传播神经网络法实现了真假目标的有效识别;比较了分别使用单波段、双波段和三波段特征数据的识别算法。
[Abstract]:Detection and recognition of targets in all countries has always been committed to solve the key problems. With the rapid development of space exploration technology, accurate identification of many countries has been able to achieve the goal, but the recognition efficiency is not high. So in order to reduce the recognition time, they all hope in the distance will be able to identify the real target as far as possible. At this time, due to the distance far detector, the target was patchy, no shape information at all, the traditional image detection, recognition and tracking technology is no longer valid. But in the actual situation, around the target was accompanied by a lot of false targets and target feature similar. For distant targets, visible light detection and radar detection has lost the advantage, using infrared detection technology can effectively distinguish true and false targets. However, due to the photoelectric technology matures, the false target already and the real target doing very similar, The target becomes more and more difficult to distinguish between true and false information. Therefore, radiation infrared sensor using a single band obtained cannot distinguish between the two. Therefore, in order to improve the detection capability of the target, many countries began to infrared multi band characteristics of target, distinguishing the radiation information utilization target of multi band based on this, the research target and radiation characteristics with false targets in the infrared band and the difference in radiation intensity and radiation spectrum of the feature extraction and useful in design of algorithms for classification and recognition. The main research work includes the following aspects: (1) the temperature characteristic of the target: to establish the target model; analyze the target time and be heated; the heat balance equation is constructed and solved the goal; more goals and The temperature characteristics of three kinds of false target. (2) the infrared radiation characteristics of target: target radiation received under the ideal condition of the detector is calculated; the influence of reflection of the external radiation target radiation on the receiving detector discussion; analysis of the atmospheric transmission of infrared radiation attenuation were studied; the material emissivity effect on target infrared radiation. (3) analysis of band characteristics of targets: to analyze and compare the detector received target radiation; using control variable method to discuss the position of the sun, orbital inclination, shell thickness ratio, surface coating, effect of casing material and heat radiation characteristics and other factors on the different characteristics of target; target characteristics of different bands were based on band division. (4) multi band processing target: the band and the band radiation characteristics of target based on Web Based on the characteristics of firing rate, two recognition algorithms based on artificial statistics are proposed. Fischer discriminant analysis and back propagation neural network are used to recognize the true and false targets effectively. The recognition algorithms using single band, dual band and three band characteristic data are compared.

【学位授予单位】：中国科学院研究生院（光电技术研究所）
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：O434.3

【相似文献】