摘要
针对由高光谱图像的大数据量和噪声等因素造成的计算量大、处理时间长以及应用效果差等问题,本文提出了一种高光谱图像快速分类方法,充分利用高光谱图像的空间信息,在实现整个原始图像空间及光谱信息特征提取的基础上,达到降低计算量、抑制图像噪声从而改善分类器性能的目的。通过仿真实验表明:该方法显著抑制了"麻点现象",在改善分类器分类效果的同时,运算时间也明显减少。
A fast classification method of hyperspectral image is presented to resolve these problems caused by large processing data and noise influence. First, space information is used to extract Spatial Region Feature Spectral. Next, the non-linear method of feature extraction is used to extract the feature of SRFS. The simulation results show that the method could significantly improve the classification results of classifiers and reduce computing time.
出处
《微计算机信息》
2010年第21期233-234,230,共3页
Control & Automation
