基于轮廓纹理分解的三维谱象数据压缩  

Data Compression for 3D Spectrum-image Data Based on Contour-Texture Decomposition

作　　者：钱神恩[1] 李桂菊[1] 刘伟奇[1] 

机构地区：[1]中国科学院长春光学精密机械研究所应用光学国家重点实验室

出　　处：《环境遥感》1996年第1期14-19,共6页

基　　金：国家自然科学基金

摘　　要：本文介绍了一种基于轮廓纹理分解的三维谱象数据压缩方法。根据成象光谱仪三维谱象数据的特点，首先用轮廓纹理方法将其中二维单色图象分解成一系列轮廓及其包围的纹理。由于成象光谱仪三维谱象数据是同一地面景物沿波长方向展开的一叠单色图象，同一地面景物反映在不同波长的单色图的轮廓相同，但纹理不同。只需对其中某一单色图的轮廓编码即可。而各单色图因波长变化而引起的纹理变化只需对相应的多项式系数编码即可。采用本文的方法对用我国自行研制的６４波段航空机载成象光谱仪在澳大利亚达尔文地区试飞获得的三维谱象数据实验，当纹理系数用零阶多项式表示时可获得３７９：１的压缩比。A method of data compression for 3D spectrum-image data based on contour-texlure decomposition is described. According to the features of 3D spectrum-imagedata is produced by an imaging spectrometer. One of the 2D monochromatic imageswhich constitute the 3D spectrumimage data is decomposed into a set of contours andthe texture surounded by their contours with contour-texture decomposition. It is thefact that 3D spectrum-image data are a pile of monochromatic forages which extendalong spectral direction, one scene on ground yields the same contour in eachmonochromatic image With the exception of texture. It is enough to encodethe contours of one of the monochromatic images. For the texture variety, we encodethe polynomial coefficients corresponding each monochromatic image. The experimentwith the original 3D spectrum-image obtained by the 64 Bands Airborne ImagingSpectrometer developed by our country in the region Darwin, Australia shows that acompression ratio 379:1 can be achieved when the texture coefficients are expressed bya zero order polynomial.

关 键 词：数据压缩 轮廓纹理分解 成象光谱 图象处理 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]