Atmospheric modulation transfer function in the infrared
- 格式:pdf
- 大小:3.72 MB
- 文档页数:12
fdk-aac编码原理
fdk-aac是一种开源的、高性能的AAC(Advanced Audio Coding)音频编码库。
以下是fdk-aac编码的基本原理:
1.AAC编码概述:AAC是一种先进的音频编码标准,旨在提供更高的音频质量和更低的比特率。
它采用了基于子带的编码技术,通过对音频信号进行频域分析和量化来实现高效的压缩。
2.Psychoacoustic Model(心理声学模型):AAC编码使用心理声学模型分析音频信号,模拟人耳的感知特性。
这包括对音频信号的掩蔽效应进行建模,以便更有效地分配比特率给对人耳更敏感的信号部分。
3.MDCT(Modulated Discrete Cosine Transform):AAC使用MDCT作为频域变换技术,将音频信号从时域变换到频域。
这种变换有助于提取信号的频域特征,为后续的量化和编码提供基础。
4.Quantization and Coding(量化和编码):MDCT输出的频域系数经过量化和编码,以减少数据量。
AAC使用了一系列的编码技术,如Huffman编码和熵编码,来进一步压缩数据。
5.Bit Allocation(比特分配):根据心理声学模型的分析结果,AAC对每个频带分配适当的比特率,以确保对人耳敏感的频段获得更多的比特,从而提高音质。
6.码率控制:AAC编码器通常具有码率控制功能,以确保生成的编码流满足指定的比特率要求。
这对于网络传输和存储空间的有效利用非常重要。
fdk-aac是一个高度优化的AAC编码库,它在实现这些基本原理的同时,通过一系列的技术手段和算法来提高编码效率和音频质量。
关于雾天图像的增强和复原现状:国内外已有很多研究人员将大气对成像的影响进行了分析和评估。
散射理论在分析大气的影响中发挥了巨大的作用,由于单射模型不能准确表现实际复杂介质的影响,多射模型得到了广泛的关注。
在光和大气介质交互问题的研究中,或多或少地借鉴了物理学科的知识,其中输运理论很常见,并被广泛应用于恶劣天气条件对图像退化的建模。
在研究尘雾等恶劣天气条件对成像影响模型的基础上,对退化图像进行复原处理取得了较大的进展。
当场景深度已知和大气条件的精确信息己知时,消除一幅图像的天气条件的影响已被证明是可行的;在计算机视觉方面,算法已经发展到不用已知大气及场景深度信息,通过处理至少两张在恶劣天气条件下拍摄的图像可以自动地计算出场景的结构和恢复场景的色度及对比度。
目前,对雾图进行清晰化处理的方法大体上有两类:基于模型的和非模型的算法。
其中,基于模型的算法通过了解图像退化的内在原因而进行逆运算来改善图像的对比度;非模型的算法并不要求知道图像退化原因的信息。
用非模型的方法处理图像我们称之为图像增强;用基于模型的方法处理图像我们称之为图像恢复,这类算法更加可靠,因为它们利用了图像退化的物理机理,其目的在于提高退化图像的保真度。
近年来,上述两个方面的研究都取得了较大的进展,特别是基于物理模型的复原问题,吸引了越来越多研究者的注意,以下分别就其研究现状做简要的介绍。
●图像增强处理的研究现状图像增强是图像处理的基本手段,是指按特定的需要突出一幅图像中的某些信息,同时,削弱或去除某些不需要的信息的处理方法。
图像对比度增强是图像处理领域中的一个传统的话题,同时又一直是较为活跃的研究领域,不断有新方法与新手段引入使该领域保持旺盛的生命力,并不断有新成果面世。
●图像恢复处理的研究现状近年来,不少学者在大气散射物理模型的基础上,对恶劣大气条件下的成像机理进行了深入的分析,并提出了一些新的雾天图像增强方法。
与传统图像增强相比,这些方法是建立在雾天成像的物理过程之上的,因此更具有针对性,处理效果也较为理想。
Modtran(Moderate Resolution Atmospheric Transmission)是一种用于计算大气传输的软件,广泛应用于气象学、环境科学、遥感技术等领域。
通过Modtran软件,可以精确计算地球大气层中不同气体的吸收光谱曲线,对于研究大气成分及其对光学传输的影响具有重要意义。
本文将以单位甲烷的吸收光谱曲线为例,介绍如何使用Modtran进行计算,并分析计算结果。
1. Modtran软件简介Modtran是由美国空军航空航天局(AFRL)开发的大气传输模拟软件,旨在模拟大气层中各种气体对光的传输情况。
该软件能够考虑地球大气层的温度、压力、湿度等参数,精确计算大气对各种波长光的吸收、散射、透射等过程,对于遥感技术、气象预测、大气化学等研究具有重要的应用价值。
2. 单位甲烷吸收光谱曲线计算我们需要准备计算所需的输入参数,包括大气条件、光源条件以及单位甲烷的吸收特性参数。
在Modtran软件中,我们需要输入大气层的温度、压力、湿度等参数,以及光源的波长范围、光照强度等参数。
我们还需要提供单位甲烷在各个波长下的吸收系数等数据。
这些数据可以通过实验测量、理论计算等手段获取。
在输入参数准备就绪后,我们可以启动Modtran软件进行计算。
软件会根据输入的数据,模拟大气层中单位甲烷对光的吸收过程,并输出吸收光谱曲线。
这个曲线可以直观地展示不同波长光在大气层中的吸收情况,帮助我们理解单位甲烷在大气中的光学特性。
3. 结果分析与讨论获得单位甲烷吸收光谱曲线后,我们可以对结果进行分析和讨论。
我们可以从图形上观察光谱曲线的形状和特点,了解单位甲烷在不同波长下的吸收强度。
通过比较不同波长下的吸收峰值及其相对强度,可以帮助我们研究单位甲烷对太阳辐射的吸收规律。
另外,我们还可以计算单位甲烷在某个波长范围内的吸收率、吸收截面等参数,进一步深入了解单位甲烷的光学性质。
这些参数对于大气化学研究、环境监测等方面具有重要的参考价值。
MTF 基礎第一章:MTF概論1-1 MTF的定義Modulation的定義--在這一個課程中,我們要進行的是有關鏡頭的MTF量測介紹。
MTF 的英文全名是Modulation Transfer Function,翻譯成中文就是光學調制傳遞函數,它有另外一個名稱叫做Contrast Transfer Function,也就是:對比度轉換函數。
從名稱來看,我們可以知道MTF有光學對比的概念在裡面。
現在就先來看Modulation (M)的定義:Modulation 是I的maximum減去I的minimum除以I的maximum加上I的minimum;也就是(光的最亮度減去光的最暗度)與(光的最亮度加上光的最暗度)的比值,所得出來的結果M,就是光的對比度。
我們舉例來看,假設有一個標靶,它的黑白條紋中最亮的光強度值(Imax )為1000,最暗的光強度值(Imin)為2,我們可以得到Imax - Imin =1000-2=998,Imax + Imin=1000+2=1002,這兩個值相比等於998除以1002,也就等於0.996。
所以這個測試標靶中,黑白條紋的modulation (M)的值為0.996,我們可以把它看成是測試標靶中黑白條紋的對比值。
Modulation基本上可以看成是經過歸一化的對比值。
何謂歸一化?歸一化指的是normalize,意思是它的最大值為1。
怎麼說呢?就一般投影機測試而言,對比度的定義是Imax 除以Imin ,如果以這個例子來看,Imax=1000 ,Imin =2,那麼它的對比度就是(Imax/Imin )=1000/2=500。
由此例,我們可以看出歸一化的對比度(Modulation)與一般的對比度(Contrast)的定義上的不同。
然而,相同的是--它們都是對比度的表示方式。
MTF的定義--在物空間,有物的Modulation。
在像空間,有成像的Modulation,我們知道Modulation其實代表的就是對比度(contrast)。
大气辐射传输模型及其软件∗焦斌亮 高志强 李素静 白云燕山大学信息科学与工程学院,河北 秦皇岛 066004摘 要:本文主要阐述了大气辐射传输模型在大气订正中的应用,介绍了大气辐射传输原理,详细地叙述了6S 、LOWTRAN 、MODTRAN 和 FASCODE 等模型,同时提到了在以上模型基础上发展起来的其它辐射传输模型及软件,并对相应的模型及软件的共同特点和主要区别进行了比较,认为大气辐射传输模型在当前的大气订正模型中依然是比较可靠而常用的方法。
关键词:大气订正 辐射传输 6S MODTRAN1 引 言大气订正是遥感技术的重要组成部分,主要包括大气参数估计和地表反射率反演两个方面。
如果获得了大气特性参数,进行大气订正就变得相对容易,但是获得准确的大气特性参数通常比较困难。
通常有两类方法用辐射传输方程来计算大气订正函数:一种是直接的方法,对于大气透过率函数和反射率函数,通过对模型的积分来得到;另一种是间接的方法,它不是直接计算所需要的大气订正函数,而是通过辐射传输模型输出的表观反射率,结合模型输入的参数来求解。
大气订正方法有很多,比如:基于图像特征的相对订正法、基于地面线形回归模型法、大气辐射传输模型法和复合模型法等。
它是利用电磁波在大气中的辐射传输原理建立起来的模型对遥感图像进行大气订正的方法。
其中,大气辐射传输模型(Atmospheric Radiative Transfer Model)法是较常用的大气订正方法,它用于模拟大气与地表信息之间耦合作用的结果,其过程可以描述为地表光谱信息与大气耦合以后,在遥感器上所获得的信息,其中考虑了光子与大气相互作用机理,物理意义明确,具有很高的反演精度。
2 大气辐射传输原理电磁辐射在介质中传输时,通常因其与物质的相互作用而减弱。
辐射强度的减弱主要是由物质对辐射的吸收和物质散射所造成的,有时也会因相同波长上物质的发射以及多次散射而增强,多次散射使所有其它方向的一部分辐射进入所研究的辐射方向。
MODTRAN介绍使用MODTRAN(Moderate Resolution Atmospheric Transmission)是一种常用的大气传输模型,它用于计算大气对电磁辐射的传输和吸收。
MODTRAN模型提供了一个完整的大气光谱模拟平台,可用于研究各种应用领域,如气象学、环境监测、红外/可见光传输和遥感应用等。
在本文中,我们将详细介绍MODTRAN的使用方法和其在不同领域中的应用。
首先,我们将介绍MODTRAN的基本原理。
MODTRAN是由美国空军强大的光谱计算模型,它模拟了地球大气对辐射的传输和吸收过程。
该模型基于辐射传输方程,将大气分为多个垂直层,考虑大气中的气体吸收、散射、云层、地面和大气透过率等因素。
它可以准确地计算不同波长和角度的电磁辐射的透过率、散射率和吸收率。
为了使用MODTRAN模型,首先需要提供准确的大气和地表输入参数。
这些参数包括大气柱密度、大气温度和湿度剖面、大气组分(如O3、CO2、CH4)的垂直分布、地表反射率和大气底部温度等。
MODTRAN提供了一个用户友好的界面,可以通过输入参数文件来设置这些参数。
此外,用户还可以选择辐射源的类型和光谱范围,并设定所需的输出参数。
一旦输入参数设置完毕,用户可以运行MODTRAN程序来计算大气传输模拟。
MODTRAN将计算地球表面和大气中不同波长的辐射的透过率、散射率和吸收率。
它还可以计算辐射在不同视场角度和观测高度下的光谱。
MODTRAN模型广泛应用于不同领域。
在气象学中,MODTRAN可以评估大气对太阳辐射的散射和吸收,从而帮助预测地面的能量平衡和气候变化。
在环境监测中,MODTRAN可以估计大气中的污染物的传输和扩散,从而帮助监测和控制大气污染。
在红外/可见光传输中,MODTRAN可以模拟和优化激光通信系统、红外导引系统和遥感系统的性能。
此外,MODTRAN还可以用于军事应用、地球观测、无人机导航等领域。
总结起来,MODTRAN是一个强大的大气传输模型,可用于计算大气对电磁辐射的传输和吸收。
MTF:Modulation(or Modulated? not sure) Transfer Functi on. 中文译作调制传递函数。
MTF 表现了一个镜头对所摄物体对比度(contrast)的再现能力。
光学中对于正弦条纹对比度的定义是V = (Imax - Imin) / (Imax + Imin),(这儿人们用矩形条纹,也一样)。
那么MTF就是成像前后此对比度的比:MTF = Vo / ViMTF值和条纹密度(空间频率,lp/mm)的关系是:条纹越密,MTF 越低。
条纹密到一定程度时,镜头的分辨能力达到极限,MTF值趋近于零。
这一极限情况到达得越晚,表明镜头的分辨力越高。
实际地,怎样从MTF值看一个镜头的特性呢?(1) 在较低条纹密度(<20lp/mm)时,MTF值受镜头分辨力的影响较小,这时候MTF主要表现了镜头成像的对比度的高低,说白了就是镜头“软”或“硬”的区别。
(2) 在高条纹密度(>40lp/mm)时,MTF值则主要体现镜头分辨能力的大小。
(对比度也有影响,但此时不是主要因素)。
看看photodo的例子:(1)Nikkor 35mm/1.4ais:Weighted MTF 1 0 lp/mm: 0,89Weighted MTF 20 lp/mm: 0,76Weighted MTF 40 lp/ mm: 0,52(2)Canon35mm/1.4L USM:Weighted MTF 10 lp/mm: 0,89W eighted MTF 20 lp/mm: 0,78Weighted MTF 40 lp/mm: 0,59看来Nikon头的分辨力较Canon稍低,但在低一些分辨力的情况下在10 lp/mm时MTF值却与canon持平,所以Nikon头的对比度的表现力要稍高于Canon, 也就是“硬”一些。
据说这是对无穷远的测试结果,我总觉得再对中距离做做测试才能对镜头有更全面的了解。
模型分析及测试方法计算大气调制传递函数作者:白珺等来源:《现代电子技术》2012年第01期摘要:采用基于点扩散函数的模型分析和基于测试方法的MODTRAN计算两种方法,计算了大气调制传递函数。
通过引入大气分层对调制传递函数的影响,给出一种大气折射率结构常数的综合计算方法,修正了大气调制传递函数模型,明确给出了模型中的参数及计算方法。
利用MODTRAN软件计算大气调制传递函数的模型并给出了计算方法。
以机载成像仪的工作环境为背景对模型进行了仿真分析,讨论了影响调制传递函数的因素,为评价大气对成像仪的影响提供了依据。
关键词:大气光学; 大气调制传递函数; 理论模型; MODTRAN; 机载成像仪中图分类号:TN911-34文献标识码:A文章编号:1004-373X(2012)01-0124-05Calculating atmospheric modulation transfer function with model analysis and measurementBAI Jun, YUAN Yan, SU Li-juan, SUN Cheng-(Key Laboratory of Precision Opto-mechatronics Technology of Ministry of Education, BeijingAeronautics and Astronautics, Beijing 100191, China)Abstract:Two methods of calculating atmospheric modulation transfer function (MTF) are introduced, which are model analysis based on point spread function and MODTRAN software calculation based on measured parameters. The model of atmospheric MTF was modified by introducing the influence of atmospheric stratification and a calculation about the structure constant of atmosphere refractive index fluctuations. All the parameters and calculation methods required in the model are presented. The methods for calculating these parameters are provided. A method of calculating MTF byMODTRAN is introduced. The parameters that influence the MTF are analyzed by simulation based on the working environment of airborne imager. Therefore, the influence of atmospheric environment on it can be evaluated.Keywords: atmospheric optics; atmospheric MTF; theoretical model; MODTRAN; airborne imager收稿日期:2011-10-基金项目:国家973计划资助项目(2009CB724005);长江学者和创新团队发展计划资助(IRT0705)0 引言成像仪在对目标进行探测时,来自地物的电磁波要经过大气层这一传输路径,大气湍流、大气分子及气溶胶的散射吸收等都会对最终的成像产生影响,使得图像失真,无法准确反映地物的物理特征。
非平衡态热传导的数学模型热传导是物体中热量从高温区域传递到低温区域的过程。
在大多数情况下,我们可以使用平衡态热传导方程来描述这个过程。
然而,在一些特殊情况下,物体内局部的温度梯度可能会产生显著影响,而这种情况下的热传导被称为非平衡态热传导。
为了建立非平衡态热传导的数学模型,我们需要考虑温度场的变化和其他可能的影响因素。
一个常用的模型是弛豫时间模型,该模型假设在物体内部的不同位置上,温度变化的速率与该位置的温度梯度成比例。
这意味着温度变化越快的地方,其时间尺度也会更快。
根据这个假设,我们可以得到一个描述非平衡态热传导的偏微分方程。
具体而言,我们可以使用以下方程来描述非平衡态热传导:$\frac{\partial T}{\partial t} - \alpha \nabla^2 T = 0$其中,$T$ 是温度场的函数, $t$ 是时间, $\alpha$ 是热扩散系数。
这个方程结合了时间变化和空间扩散的效应。
解这个方程需要一些数学技巧。
我们可以通过分析方程的性质来得到一些重要结论。
例如,该方程满足能量守恒定律,即热能在物体内部的总和不会改变。
这对于理解非平衡态热传导的行为至关重要。
另一个重要的问题是如何确定边界条件。
在实际应用中,我们经常需要考虑物体与外界的热交换。
例如,如果一个物体的一侧暴露在高温环境中,而另一侧暴露在低温环境中,我们需要考虑这些边界条件对温度场的影响。
这些边界条件可能包括热辐射、传热系数等。
通过适当选择和处理这些边界条件,我们可以模拟和预测非平衡态热传导的行为。
除了弛豫时间模型,还有其他一些数学模型可以用于描述非平衡态热传导。
例如,格林函数方法可以用于分析非均匀材料中的热传导。
这种方法利用了物体对外界扰动的响应来建立模型。
通过求解相应的积分方程,我们可以得到温度场的解析解。
这种方法在理论研究和实际应用中都有广泛的应用。
非平衡态热传导是一个复杂而重要的问题。
它在许多领域中都有着广泛的应用,包括材料科学、能源研究和环境工程等。
【珍藏】大气辐射传输校正模型(5S,modtran,acorn) 在遥感的实际应用中~常用很多简化的手段~如假设地面为朗伯面~排除云的存在~采用有关标准大气模式及大气气溶胶模式等~一次产生了许多不同类型的大气辐射传输模型~主要分为两类~1,采用大气的光学参数2,直接采用大气物理参数如lowtran、modtran等大气辐射近似计算模型~而且还增加了多次散射计算1. 5s模型该模型的代码模拟计算海平面上的均匀朗伯体目标的反射率~并假定大气吸收作用与散射作用可以耦合~就像吸收粒子位于散射层的上面一样~则大气上层测量的目标反射率可以表示为~海平面处朗伯体的反射率大气透过率分子、气溶胶层的内在反射率有太阳到地表再到传感器的大气透过率S为大气的反射率大气传输辐射校正模型,3 modtran该模型是由美国空军地球物理实验室研制的大气辐射模拟计算程序~在遥感领域被广泛应用于图像的大气校正。
,1lowtran7是一个光谱分辨率20cm~的大气辐射传输实用软件~它提供了6种参考大气模式的温度、气压、密度的垂直廓线~水汽、臭氧、甲烷、一氧化碳、一氧化二氮的混合比垂直廓线~其他13种微量气体的垂直廓线~城乡大气气溶胶、雾、沙尘、火山喷发物、云、雨的廓线~辐射参量,如消光系数、吸收系数、非对称因子的光谱分布,~以及地外太阳光谱。
lowtran7可以根据用户的需要~设置水平、倾斜、及垂直路径~地对空、空对地等各种探测几何形式~适用对象广泛。
lowtran7的基本算法包括透过率计算方法~多次散射处理和几何路径计算。
1, 多次散射处理lowtran 采用改进的累加法~自海平面开始向上直至大气的上界~全面考虑整层大气和地表、云层的反射贡献~逐层确定大气分层每一界面上的综合透过率、吸收率、反射率和辐射通量。
再用得到的通量计算散射源函数~用二流近似解求辐射传输方程。
2, 透过率计算该模型在单纯计算透过率或仅考虑单次散射时~采用参数化经验方法计算带平均透过率~在计算多次散射时~采用k,分布法3, 光线几何路径计算考虑了地球曲率和大气折射效应~将大气看作球面分层~逐层考虑大气折射效应由于lowtran直接使用大气物理参数~因而需要按照下列方法计算出与lowtran使用的大气物理参数相对应的大气光学参数179页 4.modtran辐射传输模型,1modtran可以计算0到50000cm的大气透过率和辐射亮度~它在440nm到无限,1,1大的波长范围精度是2cm~在22680到50000cm紫外波,200-440nm,范围的,1精度是20cm~在给定辐射传输驱动、气溶胶和云参数、光源与遥感器的几何立体对和地面光谱信息的基础上~根据辐射传输方程来计算大气的透过率以及辐射亮度。
光电侦察系统分辨能力研究分析袁涛;陈建发;潘枝峰;王合龙【摘要】分析了影响光电侦察系统分辨能力的基本链路要素,给出了一种通用性能预测模型的建立方法,建立了红外光电侦察设备分辨能力计算模型,并以最小可分辨温差(MRTD)作为最终评价指标.用一个具体实例进行了数值仿真计算,说明了模型的适用性.【期刊名称】《电光与控制》【年(卷),期】2019(026)006【总页数】5页(P85-88,91)【关键词】光电侦察系统;分辨能力;MRTD【作者】袁涛;陈建发;潘枝峰;王合龙【作者单位】海装驻武汉地区军事代表局,武汉 430060;中国航空工业集团公司洛阳电光设备研究所,河南洛阳 471000;中国航空工业集团公司洛阳电光设备研究所,河南洛阳 471000;中国航空工业集团公司洛阳电光设备研究所,河南洛阳 471000【正文语种】中文【中图分类】TN2160 引言光电侦察系统,通过利用光电传感器发现、识别、确认、监视、跟踪并定位目标,获取战场信息,具有高分辨率成像、昼夜侦察、快速获取敌方情报等特点,是提供战场支援和夺取信息优势的重要手段。
机载光电侦察系统最具代表性的为美国雷神公司研制的综合传感器套件(Integrated Sensor Suite,ISS),红外传感器瞬时视场可达11.4 μrad[1]。
随着光电侦察系统对分辨能力需求的不断提高,设计之初对系统的评估预测显得越来越重要,这关系到系统的最终性能以及分系统的指标参数,从而直接决定项目研发的规模及成本。
现有的一些通用的预测模型,如FLIR92,NVTherm和NVThermIP,在噪声、人眼积分滤波等方面做了大量细致的工作,对于通用光电系统具有很强的适用性,但对于“特殊需求”的光电系统开发,尤其项目设计之初牵涉到多个参数的权衡时,设计者往往只能依靠各自建立的预测模型支撑项目开发[2]。
基于此背景,本文首先分析了影响光电侦察设备分辨能力的各个链路,继而给出了部分设计参数的确定方法,并最终结合具体实例对系统分辨能力进行了预测,给出了一种通用性能预测模型的建立方法。
Atmospheric modulation transfer function in theinfraredKobi Buskila,Shay Towito,Elad Shmuel,Ran Levi,Natan Kopeika,Keith Krapels,Ronald G.Driggers,Richard H.Vollmerhausen,and Carl E.HalfordIn high-resolution ultranarrowfield-of-view thermal imagers,image quality over relatively long pathlengths is typically limited by atmospheric degradation,especially atmospheric blur.We report ourresults and analyses of infrared images from two sites,Fort A.P.Hill and Aberdeen Proving Ground.The images are influenced by the various atmospheric phenomena:scattering,absorption,and turbu-lence.A series of experiments with high-resolution equipment in both the3–5-and8–13-m regions atthe two locations indicate that,as in the visible,image quality is limited much more by atmosphere thanby the instrumentation for ranges even of the order of only a few kilometers.For paths close to theground,turbulence is more dominant,whereas for paths involving higher average elevation,aerosolmodulation transfer function͑MTF͒is dominant.As wavelength increases,turbulence MTF also in-creases,thus permitting aerosol MTF to become more dominant.A critical role in aerosol MTF in thethermal infrared is attributed to absorption,which noticeably decreases atmospheric transmission muchmore than in the visible,thereby reducing high-spatial-frequency aerosol MTF.These measurementsindicate that atmospheric MTF should be a basic component in imaging system design and analysis evenin the infrared,especially as higher-resolution hardware becomes available.©2004Optical Society ofAmericaOCIS codes:110.3080,110.4100,010.1290,010.1330,010.1110,290.1310.1.IntroductionIn this paper we describefield experiments in which images of infrared test targets were collected to in-vestigate atmospheric modulation transfer function ͑MTF͒in the IR.Thefield experiments were held inAugust2000at two sites:Fort A.P.Hill͑APHill͒in Virginia and Aberdeen Proving Ground͑APG͒in Maryland.Both sites are near Washington,D.C. Whereas at the APG site the surface is planar and the path elevation was close to the surface,at site APHill the surface was irregular with hills and valleys.A preliminary paper on the APG experiments was published recently.1Here we present more detailed analysis and measurements at both locations.Thetargets used in this research were4ftϫ4ft͑1.3mϫ1.3m͒aluminum panels of0.25-in.͑0.64-cm͒thick-ness.The panels were set in stands4ft above theground.Panels were either bare polished metal orwere painted with high-emissivityflat black paint.The panels were set out in the following order:twobare aluminum panels,two black panels,and twobare aluminum panels.The target panels were ar-ranged this way so that multiple edges were availablein thefield of view that could be used to determineMTFs at different spatial locations.The measuringsensors were in the IR range.In the mid-IR window ͑3–5m͒the sensor is a Merlin midwave-IR͑MWIR͒system manufactured by Indigo.The sensor was a320ϫ256staring array cryocooled camera.This isa staring array with a maximum integration time of16.7ms.Thefield of view was0.36by0.27deg.Inthe far-IR window͑8–12m͒the sensor was an In-frametrics760radiometer with30ϫoptics.Heretoo,thefield of view was similarly ultranarrow.However,this is a scanning-type system with an in-tegration time per detector of the order of a microsec-ond.Additional details can be found in Ref.1.Inboth cases hardware MTFs including optics extendout beyond15cycles͞mradϪ1.The long-distanceK.Buskila,S.Towito,E.Shmuel,R.Levi,and N.Kopeika͑kopeika@ee.bgu.ac.il͒are with the Department of Electrical andComputer Engineering,Ben-Gurion University of the Negev,Beer-Sheva84105,Israel.K.Krapels,R.G.Driggers,and R.H.Vollmerhausen are with the U.S.Army Night Vision and Elec-tronic Sensors Directorate,Fort Belvoir,Virginia22060-5806.C.E.Halford is with the University of Memphis,Memphis,Ten-nessee38152.Received4April2003;revised manuscript received2September2003;accepted26September2003.0003-6935͞04͞020471-12$15.00͞0©2004Optical Society of America10January2004͞Vol.43,No.2͞APPLIED OPTICS471horizontal imaging͑3.6km at APG and4.2km at APHill͒of the passive target panels was sampled in 12bits,256frames each time.A Fourier analysis of the received images from those distances was per-formed to determine the overall system MTF,which includes both the camera and the atmosphere.As described in Ref.1,because of the low signal-to-noise ratio,a Gaussianfit was used.In addition,a Fou-rier analysis was made on imaging received from nearby target boards at a distance of130m.The MTF,as a result of that image,was considered the camera’s MTF without the influence of the atmo-sphere.The overall system MTF͑long-range mea-surement͒was divided by the camera’s MTF͑short-range measurement͒to obtain the MTF of the atmosphere.Analyses of long-distance images were made with regard to time and space.The model for turbulence MTF,which we used forthe implementation of this experiment,is summa-rized by Kopeika2and derived previously by Good-man.3The turbulence strength is characterized by the refractive-index strength coefficient C n2.The transfer function that is attributed to the turbulence is implemented as either a long-term exposure model or a short-term exposure.Short-term exposure is defined as a time short enough to stop temporal changes͑less than several milliseconds typically͒. The long-term exposure MTF is described by2,3 MTF LEϭexp͑Ϫ57.4a5͞3C n2Ϫ1͞3R͒,(1) where a is the waveform constant,3͞8for a spherical wave and1for a plane wave;is the spatial fre-quency in cycles per radian;C n2is the index of re-fraction structure parameter in metersϪ2͞3;is the wavelength in meters;and R is the distance from the target in meters.For a short-term exposure the MTF is2,3MTF SEϭexp͕Ϫ57.4a5͞3C n2Ϫ1͞3R͓1Ϫ͑͞D͒1͞3͔͖,(2) where D is the diameter of the optics in meters andis0.5for a farfield and1for the nearfiing these models we calculated turbulence MTF based on C n2.As described in Ref.1,the values of C n2were mea-sured by two independent techniques:angle of ar-rivalfluctuations2and scintillations.2There was good agreement between C n2measurements with both techniques.To avoid saturation,the scintil-lometer measurements were over only a1-km path of the propagation path,near the sensor.As discussed in Ref.1,a short-term exposure ver-sion of turbulence MTF is used.The MWIR system maximum detector integration time of16.7ms is much closer to the short-term than the long-term exposure criterion.The far-IR͓long-wave infrared ͑LWIR͔͒system integration time per detector of the order of a microsecond clearly justifies our using Eq.͑2͒to calculate turbulence MTF.For the mid-IR,the short-term exposure justification is somewhat less clear.However,if we consider the actual MTF cal-culation,the difference between short-and long-term exposure turbulence MTFs is not that great,as can be seen in Figs.1and2.Hence,because the exposure time for the staring array mid-IR system is much closer to the short-exposure than long-exposure cri-terion,the error in this approximation is limited. For the far IR,there is no such error.Previous investigations of atmospheric MTF in the IR2,4,5included aerosol size distribution measure-ments and involved equipment with approximately 1͞10the resolution in a dry climate with high aerosol concentration,with the imager at the top of a30-m-high building.The present experiments involve much higher-resolution equipment in a humid cli-mate with the receiver approximately1.5m above the ground.Hence,the present experiments including those in Ref.1are much more appropriate to detect effects of optical turbulence in the thermal IR. Motivation for this research is the need for a de-tailed image system analysis to improve system de-sign.Often identification of the weakest MTF can permit image restoration to correct from such blur, including atmospheric blur.The results thatwe Fig.1.Short-and long-term exposure turbulence MTFs from measurements of Cn2at APG at noon for3600-m line of sight at a 4-m wavelength on22August2000at14:00.Fig.2.Short-and long-term exposure turbulence MTFs frommeasurements of Cn2at APG for a3600-m line of sight at a4-m wavelength on22August2000at19:00.472APPLIED OPTICS͞Vol.43,No.2͞10January2004provide here indicate that atmospheric blur is so dominant over relatively long path lengths that the hardware resolution is far from being realized.This has adverse effects on target acquisition.Charac-terization of such limiting atmospheric MTF and im-plementation in image restoration can noticeably improve target acquisition.22.ResultsA.Modulation Transfer Function of the Aerosols: Dependence on Wavelength and Time of DayIn this subsection we consider the influence of the aerosols on the atmospheric MTF.To determine the influence of aerosols wefind the overall͑long-range͒system MTF,divide it by the͑short-range͒camera’s MTF,and compare the resulting atmospheric MTF with turbulence MTF.If they are identical then the aerosol MTF is negligible.If they are not,the ratio between them is the aerosol MTF.The scintillation measurements of turbulence strength and the overallatmospheric MTF measurements from the edge re-sponse were simultaneous.As described above,there were two edges that dis-tinguished between high-and low-emissivity por-tions of the bar chart.Edge responses were measured from each.The gradient of each yielded a line-spread function,a Fourier transform of which yielded overall MTF.As described in Ref.1,C n2 values measured by angle-of-arrivalfluctuations from the left and right edges were sometimes differ-ent near midday.This was attributed to emitted heat from the dark panels,caused by crossing winds, that appeared in front of a light panel,thus causing changes in turbulence strength.Because we want to concentrate here on atmo-spheric MTF rather than on the emissivities of ma-terials used in a target,we use the scintillometer measurements of C n2rather than the angle-of-arrival fluctuation.As described above,edge response mea-surements from the long-range and short-range bar charts were used to obtain overall system MTF and camera MTFs for each measurement.Both the long-range and short-range bar charts were imaged simultaneously so that we could minimize any possi-ble vibration effects caused by winds.The overall atmospheric MTFs measured this way from both the right-and left-edge response functions were gener-ally quite similar.The case of greatest difference is shown in Figs.3and4.In other cases,as in Figs.5 and6,the atmospheric MTFs measured from both edge response functions are almost identical. Figures3–7represent MWIR measurements and Figs.8–10represent LWIR measurements at APG. From Figs.5–7for MWIR wavelengths during the day,we can see that the influence of the aerosols was not significant at APG.However,in Figs.3and4we can see the presence of a noticeable aerosol MTF at that time,causing a knee in the overall atmospheric MTF.This is characteristic of aerosol MTF.2This happened because clouds at that time caused a sud-den decrease in C n2.It is clear that in all cases atmospheric MTF cannot be neglected,especially closer to midday when turbulence is maximum.It is clear from Figs.3–10that,without good image res-toration2from atmospheric blur,the high resolution of the hardware͑15cycles͞mrad͒will not be realized. At APHill significant aerosol MTFs are also mea-sured in the LWIR,as can be seen in Figs.11–13. However,they were also observed in the MWIR wavelengths as can be seen in Figs.14–17.In all these cases atmospheric MTF is noticeably lower than turbulence MTF.In summary,at APG,aerosol MTF was negligible for most of the MWIR measurements͑except for Figs. 3and4͒whereas aerosol MTF was significant for all the LWIR measurements.At APHill,aerosol MTF was significant for all measurements in both the MWIR and the LWIR bands.Indeed,the shape of the overall atmospheric MTF is dominated by that of aerosol MTF,with a clearly defined knee in the MTF curve.2We attribute the difference in MWIRmea-Fig.3.MTF right edge at APG for MWIR on22August2000at11:58.Fig.4.MTF left edge at APG for MWIR on22August2000at 11:58.10January2004͞Vol.43,No.2͞APPLIED OPTICS473surements to the differences in topography at both locations.At APHill the higher average elevation gives rise to a reduction in C n 2and therefore to a rise in turbulence MTF,reducing its signi ficance and thereby allowing aerosol MTF to become more dom-inant,so much so that overall atmospheric MTF there is dominated by an aerosol MTF shape with a sharp knee at the aerosol MTF spatial-frequency cut-off.2At APG overall atmospheric MTF does not ex-hibit such a sharp knee for MWIR imaging.We attribute the dominance of aerosol MTF at LWIR wavelengths in both locations to the wave-length dependence of Eq.͑2͒,which causes turbu-lence MTF to be less signi ficant at longer wavelengths.B.Atmospheric Modulation Transfer FunctionIn this subsection we compare the atmospheric MTF from both experimented sites.Figure 18shows the MTFs of the images for all measurements.We can clearly distinguish the MTFs at APHill because theyare wider than those at APG despite the shorter path length at APG ͑3600m at APG,4235m at APHill ͒.The explanation is that the APG site is closer to the ground than the APHill site,which causes C n 2to increase and MTF to decrease ͑see Table 1͒.1.Spatial EffectsAnother comparison that can be performed between the areas is to look at the whole image rather than only at the target boards,as can be seen in Figs.19͑a ͒and 19͑b ͒.At site APG we can see the whole target board,but it is blurry with gradual edge spread func-tion transitions ͑and a correspondingly narrow MTF function ͒.At site APHill we can see interference on the right side of the target and a water puddle or a different element that blocks transitions from the ground to the atmosphere ͑in front of the target im-age ͒.The white target boards on the left side have vanished from the image because of equal energy emission of the surroundings.In spite of that,the right-edge transition is still sharp andimmediate.Fig.5.MTF left edge at APG for MWIR on 22August 2000at14:08.Fig.6.MTF right edge at APG for MWIR on 22August 2000at14:08.Fig.7.MTF left edge at APG for MWIR on 22August 200at16:04.Fig.8.MTF left edge at APG for LWIR on 22August 2000at 12:10.474APPLIED OPTICS ͞Vol.43,No.2͞10January 2004Fig.9.MTF left edge at APG for LWIR on22August2000at15:04.Fig.10.MTF right edge at APG for LWIR on22August2000at18:00.10January2004͞Vol.43,No.2͞APPLIED OPTICS475Fig.11.MTF atmospheric left edge at APHill for LWIR on16August2000at13:01.Fig.12.MTF atmospheric right edge at APHill for LWIR on16August2000at15:19. 476APPLIED OPTICS͞Vol.43,No.2͞10January2004Fig.13.MTF atmospheric right edge at APHill for LWIR on16August2000at17:53.Fig.14.MTF atmospheric left edge at APHill for MWIR on16August2000at12:49.10January2004͞Vol.43,No.2͞APPLIED OPTICS477Fig.15.MTF atmospheric right edge at APHill for MWIR on16August2000at12:49.Fig.16.MTF atmospheric right edge at APHill for MWIR on16August2000at14:31. 478APPLIED OPTICS͞Vol.43,No.2͞10January2004The conclusions for the overall system MTF of the image are also relevant to the atmospheric MTF be-cause the atmospheric MTF can be obtained when we divide the overall system MTF by the camera ’s MTF ͑that is constant ͒.In some of the images that we have analyzed,from both sides of the image we found a difference between the MTF of the right side compared with that of the left side.We can see this difference in the LWIR images in Figs.14and 15and 20and 21.The con-trast between the high-and low-emissivity boards is usually better on the right side than on the left,con-forming to the wider and higher MTFs for the rightedge on most graphs in which MTFs for both edges are displayed.Exceptions are Figs.3and 4for MWIR at APG.2.Temporal EffectsTemporal effects are also shown in Figs.20and 21.The worst images are those near midday when tur-bulence is strongest.C.ErrorsSources of error include the short-term exposure as-sumption for MWIR turbulence MTF that,as shown in Figs.1and 2,is usually small compared to the difference between turbulence and atmospheric MTFs for the MWIR measurements,and even more so for the LWIR measurements.This error is small because the time exposure for the MWIR equipment is much closer to that required for small exposure than for long exposure.For the LWIR equipment,the exposure is so short that there is no error intheFig.17.MTF atmospheric left edge at APHill for MWIR on 16August 2000at17:51.Fig.18.MTF comparison at APHill and APG for LWIR.Table 1.C n 2Data at APG on 22August 2000and APHill on16August 2000Time ͑h ͒APG ͑10Ϫ14͒APHill ͑10Ϫ14͒11:009 1.9713:0015 1.6514:0014 1.916:0011 2.618:00 4.4 1.2320:001.20.8610January 2004͞Vol.43,No.2͞APPLIED OPTICS 479short exposure modeled here,as described at the end of Section 1.Another source of error is the fact that the scintil-lometer measurements of C n 2were only over a por-tion of the line of sight ͑from a camera extending 1km toward the distant target ͒rather than the entire line of sight.However,it is the turbulence in the region closer to the camera that dominates the image plane turbulence.2Therefore differences between C n 2at distances longer than a kilometer and those mea-sured within a kilometer from the cameras have a relatively small effect on the value of C n 2affecting the images.Another source of error is the noise or uncertainty in measuring turbulence MTF,especially at high spa-tial frequencies where noise can become a problem.Examination of Table 2and Fig.10in Ref.1indicates that,at low values of the turbulence MTF at high spatial frequencies,such errors were limited to ap-proximately 20%,or approximately Ϯ10%of the tur-bulence MTF values plotted here.Another source of error is the validity of the as-sumption that turbulence and aerosol MTFs are in-dependent of one another.A statistical regression coef ficient model with which to predict the refractive-index structure coef ficient according to weather was carried out in Ref.6.Indeed,it was found that high aerosol loading of the atmosphere is associated with noticeably increased turbulence strength,attributed to increased atmospheric warming derived from ab-sorption of light by aerosols.This means that,strictly speaking,aerosol and turbulence phenomena are not completely independent of one another.However,it was found that this occurs infrequently and when it does,the statistical signi ficance is small,being of the order of 10–15%.This means that to a first approximation the multiplication of turbulence MTF by aerosol MTF to compose atmospheric MTF is reasonable.Accordingly,designating aerosol MTF as MTF A and overall atmospheric MTF by M A ,MTF SE MTF A ϭM A .(3)The errors in measuring overall atmospheric MTF are rather small,being based on direct measurement of the line-spread function rather than time-varying scintillations.Therefore differentiating Eq.͑3͒yields⌬MTF A ͞MTF A ϭϪ⌬MTF SE ͞MTF SE ,(4)where ⌬designates the errors.The minus sign in-dicates that if turbulence MTF is underestimated then aerosol MTF is overestimated,and vice versa.Because the turbulence MTF error magnitude is es-timated at Ϯ10%,then the same holds for the aerosol MTF error magnitude as well.The above errors are too small to change the conclusions here,described in Section3.Fig.19.͑a ͒Image from APG,LWIR on 22August 2000at 15:03.͑b ͒Image from APHill,LWIR on 16August 2000at14:12.Fig.20.Images from ͑a ͒APG on 22August 2000and ͑b ͒APHill on 16August 2000for LWIR for different hours of theday.Fig.21.͑a ͒APG for LWIR on 24August 2000at 09:05.͑b ͒APG for LWIR on 23August 2000at 12:07.͑c ͒APG for LWIR on 22August 2000at 19:04.͑d ͒APHill for LWIR on 22August 2000at 11:00.480APPLIED OPTICS ͞Vol.43,No.2͞10January 20043.Discussion and ConclusionsWith the advent of high-resolution IR equipment,these experiments indicate that,for ranges of the order of even a few kilometers,image resolution is dominated much more by the atmosphere than by the equipment,even in the thermal IR.Atmospheric MTF should therefore become a basic component in thermal imaging system design and analysis.Both wavelength and elevation have critical roles.In Fig.18we can see that the MTF graphs at APHill are wider than those at APG despite the longer distance at APHill ͑4235m as compared with 3600m ͒.The explanation is that the APG site is at a lower eleva-tion ͑1.5m ͒than the APHill site,which causes C n 2to increase and the MTF to narrow.Data for C n 2are presented in Table 1and Fig.22.The smaller values of C n 2at APHill give rise to more noticeable aerosol MTFs in both the MWIR and the LWIR spectral regions,whereas at APG aerosol MTF is noticeable primarily at LWIR and less so at MWIR.The reason for the latter is probably that,for the same value of C n 2,the wavelength dependence of turbulence MTF ͓Eqs.͑1͒and ͑2͔͒causes it to be higher at longer wavelengths,thus permitting a more noticeable role for aerosol blur at longer wavelengths.It is interest-ing to consider the clear signi ficance of aerosol MTF at both locations for LWIR radiation.Because of the long wavelengths,scattering is diminished in this wavelength region.However,aerosol MTF is also strongly dependent on absorption by aerosols and at-mospheric molecules,which are also particulates.2,7The idea is that image transmission through particu-lates involves both a sharp unscattered light image and a blurred scattered light image.By conservation of energy,interaction of a photon with a particulate can involve either scattering or absorption,but not both.Hence it is the clear unscattered image that is more likely to be attenuated by absorption rather than the blurred scattered image because the latter has been scattered.2,7Therefore it cannot be absorbed.This implies that aerosol MTF amplitude is decreasedso that its high-spatial-frequency asymptotic value is approximately equal to atmospheric transmission where atmospheric transmission is governed by both absorption and scattering,as derived in detail else-where.7In this way absorption degrades image qual-ity.In the LWIR region,there is signi ficant absorption especially by both H 2O and CO 2,so the aerosol MTF is affected strongly by absorption.The weather during these experiments in August in the Washington,D.C.area was rather humid at times.For example,at APG on 22August,the relative hu-midity until 8a.m.was close to 100%.It decreased to around 59%at 3p.m.,remained around there until approximately 6:30p.m.,rose to approximately 85%by 8p.m.,and then rose to approximately 95%by 3a.m.23August.Relative humidity began to decrease again to approximately 64%around noon 23August.Periods of high humidity tend to increase aerosol MTF more in the LWIR wavelength area because of the effects of atmospheric absorption on aerosol MTF,in addition to the increase in aerosol size generated by an increase in humidity because of adsorption and ab-sorption of moisture by the aerosols.Temperatures during the night of 22August were of the order of 16°–17°C,rising to 30°C around 3–4p.m.They decreased to around 21°C early in the morning of 23August,and then rose to around 31°C around mid-day.Higher temperatures tend to increase optical turbulence.2The weather data for APHill showed similar tendencies,only the numbers were slightly dif-ferent.The figures shown here are consistent with the weather,with turbulence MTFs being narrowest in the midafternoons when temperature was at a maxi-mum and aerosol MTFs being least prominent at that time ͑when humidity was at its lowest ͒.This is con-sistent with the measurements of thermal imaging turbulence MTF by Watkins et al .8It is assumed that atmospheric MTF is the product of turbulence and aerosol MTFs,and therefore aero-sol MTF is obtained when atmospheric MTF is di-vided by turbulence MTF.The difference in aerosol MTFs between the right and left edges stems from the assumption that aerosol and turbulence MTFs are independent of one another.However,this is not completely true.The refractive-index structure coef ficient C n 2generally increases with increased aerosol loading,which is attributed to slightly in-creased atmospheric absorption and heating by the aerosols.2,6As seen in Ref.1,cross winds cause dif-ferences in air temperature at the two edges,which causes differences in C n 2.However,temperature differences also affect aerosol concentrations 9be-cause of convection,as well as aerosol sizes because of temperature effects on evaporation of water vapor adsorbed to aerosols.9Hence it is not surprising to see different aerosol MTFs as well as turbulence 1MTFs sometimes at the right and left edges.How-ever,such differences are small.References1.K.Krapels,R.G.Driggers,R.H.Vollmerhausen,N.S.Kopeika,and C.E.Halford,“Atmospheric turbulence modulationtrans-Fig.22.C n 2comparison as a function of the time of day at APG and APHill.10January 2004͞Vol.43,No.2͞APPLIED OPTICS481fer function for infrared target acquisition modeling,”Opt.Eng.40,1906–1913͑2001͒.2.N.S.Kopeika,A System Engineering Approach to Imaging,Vol.38of the SPIE Press Monographs͑SPIE,Bellingham,Wash., 1998͒.3.J.W.Goodman,Statistical Optics͑Wiley,New York,1985͒.4.D.Sadot,G.Kitron,N.Kitron,and N.S.Kopeika,“Thermalimaging through the atmosphere:atmospheric MTF theory and validation,”Opt.Eng.33,880–887͑1994͒.5.D.Sadot,A.Dvir,I.Bergel,and N.S.Kopeika,“Restoration ofthermal images distorted by the atmosphere,based upon mea-sured and theoretical atmospheric modulation transfer func-tion,”Opt.Eng.33,44–53͑1994͒.6.D.Sadot and N.S.Kopeika,“Forecasting optical turbulencestrength on the basis of macro scale meteorology and aerosols: models and validation,”Opt.Eng.31,200–212͑1992͒.7.D.Sadot and N.S.Kopeika,“Effects of absorption on imagequality through a particulate medium,”Appl.Opt.33,7101–7111͑1994͒.8.W.R.Watkins,S.B.Crow,and F.T.Kantrowitz,“Character-izing atmospheric effects on target contrast,”Opt.Eng.30, 1563–1575͑1991͒.9.J.Gottlieb,B.Fogel,I.Dror,Z.Y.Ofer,and N.S.Kopeika,“Prediction of airborne particle statistics according to weather forecast:concentration and scattering area,”Opt.Eng.34, 1208–1218͑1995͒.482APPLIED OPTICS͞Vol.43,No.2͞10January2004。