高光谱大气红外遥感图像的通道选择及压缩方法研究
发布时间：2019-01-26 19:26
【摘要】：随着高光谱大气红外遥感探测技术的发展,对大气的探测越来越精细,探测周期越来越短,从而探测信息的数据量也随之越来越大,无论是在星上还是在星下,对于探测信息的存储和传输是在数据应用过程当中必然面对的问题。因此,为达到快速传输高光谱大气红外遥感图像数据,并使其占用的存储空间小,同时保证对数据同化和反演的准确度的目的,对其进行辐射抽稀(radiance thinning)是非常必要的,辐射抽稀分为两种方面,即对数据的无损压缩和光谱通道选择。本文主要针对辐射抽稀的两种情况展开研究。首先对大气探测及其遥感数据传输和存储进行分析,并以AIRS探测仪所探测的典型高光谱大气红外遥感图像为典型实验数据,对其空间相关性和光谱相关性的特性进行分析,定性说明对其进行无损压缩研究和光谱通道选择的可行性和必要性。其次,考虑高光谱大气红外遥感图像的光谱相关性极大,为实现有效的压缩效果,本文采用ICA变换去除谱间冗余,使图像在变换域的ICs成分实现相互独立;之后对所得ICs成分及变换系数进行量化与反量化,保留量化残差,对量化后数据以及量化残差进行预测处理,以减小待编码数据量;在编码部分本文选取区间编码并利用随机学习弱估计方法(SLWE)改进其中的概率估计模型,以提高编码效果。在编码之前,对待编码数据进行正值化处理,使其更适合区间编码过程。最后对典型的AIRS实验数据进行压缩,压缩比可达3.35以上,并与部分现有的经典压缩方法对比,本文所研究的压缩方法在压缩比上具有一定的优势。最后,根据AIRS探测资料中典型的亮温资料以及温、湿度反演廓线,通过辐射传输模式(RTTOV)得到其温、湿度Jacobi矩阵,为了从大量的AIRS探测资料中抽取出与应用相关的通道信息,实现减小数据量并适合应用的目的,分别对温、湿度Jacobi矩阵进行基于PC-AIC的通道选择,选出对温、湿度影响较大的波段通道。并根据所选通道对AIRS探测亮温资料进行温、湿度反演应用,将其反演结果与卫星资料数值天气预报应用研究组(NWPSAF)所给通道的反演结果对比,本文所研究方法对温、湿度反演所得廓线的误差更小。进一步给出基于信息容量迭代的经典通道选择方法,并与本文的PC-AIC算法对比,得出PC-AIC算法所选通道组合反演效果更好的结论,说明其在具体应用上,保证数据量减小的同时,可进行有效的通道选择。
[Abstract]:With the development of infrared remote sensing detection technology for hyperspectral atmosphere, the atmospheric detection becomes more and more precise, and the detection period is shorter and shorter, so the amount of data of the detecting information becomes more and more large, whether on or under the star. The storage and transmission of detection information is an inevitable problem in the process of data application. Therefore, in order to transmit hyperspectral infrared remote sensing image data quickly, and to make the storage space small, and to ensure the accuracy of data assimilation and inversion, it is very necessary to carry out radiation-pumped (radiance thinning). Radiation pumping can be divided into two aspects: lossless compression of data and spectral channel selection. In this paper, two kinds of radiation pumping conditions are studied. Firstly, the atmospheric detection and its remote sensing data transmission and storage are analyzed, and the spatial and spectral correlation characteristics of the typical hyperspectral infrared remote sensing images detected by the AIRS detector are analyzed. The feasibility and necessity of lossless compression study and spectral channel selection are explained qualitatively. Secondly, considering the spectral correlation of hyperspectral infrared remote sensing image, in order to achieve an effective compression effect, this paper uses ICA transform to remove the redundancy between spectra, so that the ICs components of the image in the transform domain can be independent of each other. Then the quantization and inverse quantization of the ICs components and transformation coefficients are carried out, the quantization residuals are retained, and the quantized data and quantized residuals are predicted and processed to reduce the amount of data to be encoded. In the coding part, interval coding is selected and the probabilistic estimation model is improved by using the random learning weak estimation method (SLWE) to improve the coding effect. Before coding, positive processing of coded data is carried out to make it more suitable for interval coding process. Finally, the compression ratio of the typical AIRS experimental data is over 3.35, and compared with some classical compression methods, the compression method studied in this paper has some advantages in compression ratio. Finally, according to the typical bright temperature data and the inversion profile of temperature and humidity in AIRS detection data, the temperature and humidity Jacobi matrix is obtained by radiative transfer mode (RTTOV). In order to extract the channel information related to application from a large amount of AIRS detection data, For the purpose of reducing the amount of data and being suitable for application, the Jacobi matrix of temperature and humidity is selected based on PC-AIC channel, and the band channel which has a great influence on temperature and humidity is selected. According to the selected channel, the inversion of temperature and humidity of AIRS detection bright temperature data is carried out. The inversion results are compared with the inversion results of the channel given by (NWPSAF), a research group of satellite data numerical weather forecast. The error of the profile obtained by humidity inversion is smaller. Furthermore, the classical channel selection method based on information capacity iteration is given, and compared with the PC-AIC algorithm in this paper, it is concluded that the combination of channels selected by the PC-AIC algorithm has better results. At the same time, effective channel selection can be carried out while the amount of data is reduced.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP751
本文编号：2415826
[Abstract]:With the development of infrared remote sensing detection technology for hyperspectral atmosphere, the atmospheric detection becomes more and more precise, and the detection period is shorter and shorter, so the amount of data of the detecting information becomes more and more large, whether on or under the star. The storage and transmission of detection information is an inevitable problem in the process of data application. Therefore, in order to transmit hyperspectral infrared remote sensing image data quickly, and to make the storage space small, and to ensure the accuracy of data assimilation and inversion, it is very necessary to carry out radiation-pumped (radiance thinning). Radiation pumping can be divided into two aspects: lossless compression of data and spectral channel selection. In this paper, two kinds of radiation pumping conditions are studied. Firstly, the atmospheric detection and its remote sensing data transmission and storage are analyzed, and the spatial and spectral correlation characteristics of the typical hyperspectral infrared remote sensing images detected by the AIRS detector are analyzed. The feasibility and necessity of lossless compression study and spectral channel selection are explained qualitatively. Secondly, considering the spectral correlation of hyperspectral infrared remote sensing image, in order to achieve an effective compression effect, this paper uses ICA transform to remove the redundancy between spectra, so that the ICs components of the image in the transform domain can be independent of each other. Then the quantization and inverse quantization of the ICs components and transformation coefficients are carried out, the quantization residuals are retained, and the quantized data and quantized residuals are predicted and processed to reduce the amount of data to be encoded. In the coding part, interval coding is selected and the probabilistic estimation model is improved by using the random learning weak estimation method (SLWE) to improve the coding effect. Before coding, positive processing of coded data is carried out to make it more suitable for interval coding process. Finally, the compression ratio of the typical AIRS experimental data is over 3.35, and compared with some classical compression methods, the compression method studied in this paper has some advantages in compression ratio. Finally, according to the typical bright temperature data and the inversion profile of temperature and humidity in AIRS detection data, the temperature and humidity Jacobi matrix is obtained by radiative transfer mode (RTTOV). In order to extract the channel information related to application from a large amount of AIRS detection data, For the purpose of reducing the amount of data and being suitable for application, the Jacobi matrix of temperature and humidity is selected based on PC-AIC channel, and the band channel which has a great influence on temperature and humidity is selected. According to the selected channel, the inversion of temperature and humidity of AIRS detection bright temperature data is carried out. The inversion results are compared with the inversion results of the channel given by (NWPSAF), a research group of satellite data numerical weather forecast. The error of the profile obtained by humidity inversion is smaller. Furthermore, the classical channel selection method based on information capacity iteration is given, and compared with the PC-AIC algorithm in this paper, it is concluded that the combination of channels selected by the PC-AIC algorithm has better results. At the same time, effective channel selection can be carried out while the amount of data is reduced.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP751
【参考文献】
相关期刊论文 前3条
1 董瑶海;;风云四号气象卫星及其应用展望[J];上海航天;2016年02期
2 张水平;;AIRS资料反演大气温度廓线的通道选择研究[J];气象科学;2009年04期
3 杜华栋;黄思训;石汉青;;高光谱分辨率遥感资料通道最优选择方法及试验[J];物理学报;2008年12期
相关博士学位论文 前1条
1 蒋德明;高光谱分辨率红外遥感大气温湿度廓线反演方法研究[D];南京信息工程大学;2007年
，本文编号：2415826
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2415826.html
