摘要
提出了一种高光谱数据彩色显示策略,可在宏观、局部和微观三个级别上对图像信息进行更加丰富准确的可视化表达。该策略通过以下三种方法实现:根据类别空间相关性来分配彩色标签的方法,能够生成具有更好的类间可分性的彩色图像;基于方向裂片纹理技术的高光谱数据显示方法,能够同时表示地物的空间信息和光谱信息,且生成的图像具有良好的特征独立性和视觉直观性;定序饼图显示方法,不仅能够准确显示混合像素成分,还具有更好的类别空间分布特性。实验表明,文中提出的方法可行、有效,且更加符合人类视觉特性和类别分布规律。
A color display scheme was presented for hyperspectral imagery.It can proceed more various and accurate visual expression of the overall spatial information,local material distribution,and subpixel details.The scheme was conducted with the following three methods： automatic color distribution method based on class spatial correlation,which can produce a colorful image with better and clear class distinction;hyperspectral imagery display method based on oriented sliver textures technique,which can simultaneously show the surface features＇ spatial information and spectral information with good independence features and perceptual intuition in the produced images;and sequence pie display method,which can not only show precise elements of the mixed pixels,but also has classified spatial distribution features.Experiments show that the proposed visualization methods of hyperspectral image are feasible,effective,and in line with human visual quality and class distribution pattern.
出处
《红外与激光工程》
EI
CSCD
北大核心
2012年第9期2527-2533,共7页
Infrared and Laser Engineering
基金
国家自然科学基金(60802059
61275010)
教育部博士点新教师基金(200802171003)
