高光谱遥感混合像元盲分离模型

The hyperspectral mixed pixel blind separation model

针对高光谱遥感影像中"混合像元"问题,在前人研究的基础上,本文提出了结合盲源信号分离技术的混合像元盲分离模型,该模型的核心算法是HFC算法和基于负熵的改进FastICA算法,优点是能够通过非常少的先验知识就可以进行混合像元分解,完成特征提取工作,且收敛速度快,精度较高.采用高光谱模拟数据和加噪数据,分别检验混合像元盲分离模型的有效性和稳定性,实验结果表明:模型受端元提取数目多少的影响以及高斯白噪声的影响较小,对高光谱遥感数据特征提取具有有效性和稳定性.

Hyperspectral remote sensing images for ＂mixed pixel＂ issue, on the basis of previous studies, we propose a combination of blind source separation techniques blind separation of mixed pixel model, The model＇s core algorithm is based on negative entropy HFC algorithm and improved FastlCA algorithm, the advantage is the ability to very little prior knowledge can be carried unmixing complete feature extraction work, and fast convergence and high precision. Simulated data using hyperspectral data processing noise were tested blind separation of mixed pixel model validity and stability, the experimental results showed that： Million by the end of the model to extract the influence of the number of Gaussian white noise and less impact on the feature extraction of hyperspectral remote sensing data validity and stability.