一种大孔径静态干涉高光谱成像数据压缩方法  

A Large Aperture Static Interference Hyperspectral Imaging Data Compression Method

作　　者：汪巍 冯向朋 张耿[1,3] 刘学斌 李思远[1,3] WANG Wei;FENG Xiangpeng;ZHANG Geng;LIU Xuebin;LI Siyuan(Key Laboratory of Spectral Imaging Technology,Xi′an Institute of Optics and Precision Mechanics of CAS,Xi′an 710119,China;University of Chinese Academy of Sciences,Beijing 100049,China;Shaanxi Provincial Key Laboratory of Optical Remote Sensing and Intelligent Information Processing,Xi′an 710119,China)

机构地区：[1]中国科学院西安光学精密机械研究所光谱成像技术重点实验室,西安710119 [2]中国科学院大学,北京100049 [3]陕西省光学遥感与智能信息处理重点实验室,西安710119

出　　处：《光子学报》2024年第6期226-239,共14页Acta Photonica Sinica

基　　金：基础加强计划重点项目(No.2022-JCJQ-ZD-215-03);自主部署项目(No.S22-037)。

摘　　要：大孔径静态干涉成像遥感数据的数据量较大,需要寻找一种合适的方法对其压缩。从大孔径静态干涉成像机理出发,分析了干涉数据的空间和干涉维冗余性,并基于现有成熟混合压缩编码方法,提出了基于相似干涉曲线与不同光程差之间冗余去除的算法,对冗余数据进行去除预处理。对干涉数据进行基于曲线表的干涉曲线编码表示,对不同光程差之间的图像进行相关性预测,减少了大孔径静态干涉成像遥感图像的量化深度并降低了图像的信息熵,再结合JPEG2000算法进行无损或有损压缩。实验结果表明,对于大孔径静态干涉成像数据,该算法可实现压缩比为3.1倍的无损压缩,有损压缩的率失真曲线也优于其他对比算法,其复原图像反演出的光谱曲线的光谱角和相对二次误差均优于其他对比算法处理的数据,有效保护了光谱信息。After spectral reconstruction of large aperture static interferometry remote sensing data,a spectral image data cube can be generated that contains both spatial information about the ground objects and interference information.Considering the large volume of large aperture static interferometry remote sensing data and the scarce bandwidth of space-to-earth links,it is necessary to find suitable compression methods to compress this data.Starting from the mechanism of large aperture static interferometry imaging,based on the principles of large aperture static interferometry spectral imaging and the redundant information in the data,a compression algorithm called Spectral-Interference-Optical Path Difference Redundancy Removal(SIORR)is proposed.This algorithm fully considers the similarities between the interference curves of similar ground points and the redundancy between multiple frames.The SIORR algorithm can be divided into three parts.First,it analyzes and processes the interference curves in the hyperspectral data.In large aperture static interferometry spectral imaging remote sensing images,due to the continuity of spatial distribution of adjacent ground objects,the differences between interference curves of the same category are small.By constructing a table of typical interference curves to encode representations of different categories of interference curves,indexes of matching items and necessary correction information are recorded.Each table item not only represents a specific interference curve but also serves as a reference for compressing that type of curve.During the actual compression process,each interference curve in the original data is matched with an item in the curve table,and data compression and recovery are achieved by recording the index of the matching item and necessary correction information.Subsequently,during the interferometric imaging process,there is a high similarity between different optical path difference images,specifically reflected in the texture features of the remote sen

关 键 词：大孔径静态干涉成像 图像压缩 信息冗余 干涉 光谱 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]