具有距离保持特性的高光谱彩色可视化模型

提出一种将高维的高光谱图像非线性优化到三维彩色空间的可视化方法. 以距离保持特性作为主要设计标准，同时保证生成的图像具有尽可能大的类间可分性，使生成图像的各像元间的距离差与高光谱数据各光谱间距离差高度相关. 该方法由以下四部分组成：1)将高光谱图像中各端元光谱降维到二维空间作为色品坐标；2)由第3维亮度值的优化使相关性达到最优，进而确定各端元的颜色标签；3)根据像素所含各类别的丰度值进行颜色的线性混合；4)利用局部优化方法对彩色图像进行优化校正，最终实现整幅图像的彩色可视化. 实验表明，该方法合理可行，生成的图像具有较好的视觉效果，并能获得较好的距离保持特性及可分性，适用于高光谱数据的可视化显示.

Distance-Preserving Color Visualization Model for Hyperspectral Imagery

A color display model of hyperspectral imagery is proposed, which optimizes high dimensional hyperspectral images nonlinearly into a three-dimensional color space. The property of spectral distance preserving in the model is considered as a main design criterion, and good separability is guaranteed in the
final visualized color images so that the range difference between pixels and the range difference between the spectra are highly correlated. The method consists of four parts: 1) End-members are reduced to a two dimensional space as chromaticity coordinates. 2) Optimization of brightness values in the third dimension maximizes the relevance, thus color labels of each end-member are determined. 3) Linear mixing of colors are implemented in accordance with the various abundance value of each pixel. 4) A local optimization method is used to optimize and correct the color image to achieve color display of the image. Experiments show that the proposed model is feasible as the final color images have good visual quality with spectral distance well preserved and separable. It is suitable for HSI visual display.