论文部分内容阅读
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。
This paper introduces a method of 3D spectral image data compression based on contour texture decomposition. According to the characteristics of 3D spectral image data of imaging spectrometer, first of all, the two-dimensional monochromatic image is decomposed into a series of contours and the surrounding texture by contour texture method. Because three-dimensional spectral data of imaging spectrometer is a stack of monochromatic images of the same ground scene along the wavelength direction, the monochromatic images of the same ground scene reflected at different wavelengths have the same contour but different textures. Simply outline the outline of one of these monochrome images. The monochromatic image due to changes in wavelength caused by texture changes only need to be encoded in the corresponding polynomial coefficients. In this paper, we experimentally test the 3D spectral data obtained from the self-developed 64-band airborne imagery imaging spectrometer in Darwin, Australia. When the texture coefficients are represented by the zero-order polynomial, the compression ratio of 379: 1 can be obtained.