论文部分内容阅读
高光谱影像具有“谱像合一”的特征,因而可以从谱和像两个角度对高光谱影像进行分形分析。本文把计算表面分形维的双层地毯法用于计算Hyperion高光谱曲线的分形维,根据每一像元的高光谱曲线计算出每一尺度的上分形信号和下分形信号,并对分形信号值及分形特征尺度进行分析。结果表明对于不同地物其分形信号在某些特定尺度(分形特征尺度)有较大的差异,而且对于Hyperion高光谱影像,下分形信号可更有效地突出地物特征。文章最后简要讨论了上分形和下分形、起始计算尺度以及原始采样点数目对分形特征尺度的影响。
Hyperspectral images have the feature of “Spectrum ” so that fractal analysis of hyperspectral images can be performed from both spectral and image perspectives. In this paper, the double-layer carpet method for calculating the surface fractal dimension was used to calculate the fractal dimension of Hyperion hyperspectral curve. The upper fractal signal and the lower fractal signal of each scale were calculated according to the hyperspectral curve of each pixel. And fractal dimension. The results show that the fractal signal has a great difference in some specific scales (fractal dimension) for different features, and the fractal signal can highlight the features more effectively for Hyperion hyperspectral images. Finally, the article briefly discusses the influence of the upper fractal and the lower fractal, the initial computational scale and the number of original sampling points on the fractal scale.