摘要
提出一种新的遥感图像混合像元分解方法,通过最小化一种带约束条件的能量函数,可实现多通道遥感图像中混合像元更精确的分解.同时针对所提议的算法进行了模拟和实际数据的实验验证,并将结果与BP神经网络的分解结果进行比较,结果表明,本文所提出的带约束条件的能量函数最小化方法在分解准确性和抗噪声能力方面,明显优于基于BP神经网络的分解方法.
A new scheme for the detection and classification of subpixel spectral signatures in remote sensing images was presented. By minimizing the energy function with two special constraints, the mixed pixels in muhichannel remote sensing images can be decomposed more precisely. In this study, experiments on the proposed scheme and BP neural network with artificial and real-world data were performed. The experiments show that the proposed scheme can get more precise results and is obviously more robust than BP neural network.
出处
《红外与毫米波学报》
SCIE
EI
CAS
CSCD
北大核心
2005年第6期463-466,共4页
Journal of Infrared and Millimeter Waves
基金
国家重点基础研究项目(2001CB309400)
航天支撑技术基金(2004-1.3-03)
国家自然科学基金(30370392)
上海市自然科学基金(04ZR14018)资助项目
关键词
约束条件
能量函数
混合像元分解
BP神经网络
constraint conditions
energy function
decomposition of mixed pixels
BP neural network
