摘要
现有遥感图像的许多分类方法大都忽略了混合像元存在的事实,通过理解遥感影像像元点目标的空间分布特性,提出基于Hopfield神经网络的遥感图像超分辨率目标识别算法。在Hopfield神经网络模型下,利用模糊分类技术进行模糊分类,然后用分类结果约束Hopfield神经网络的方法获取超高分辨率的遥感图像,能够提高遥感图像的目标分辨率,使其目标特征信息更清晰。
A remote sensing image super resolution object recognition algorithm based on Hopfield Neural Networks was proposed. Fuzzy classification technology is used for classification, Then the result is used to restrict Hopfield Neural Networks. When there are only few learning samples, Hopfield Nerve Net can also output object information with higher resolution. Therefore, this remote sensing image processing approach can enhance the object resolution of remote sensing image and make the object character characteristic more in focus.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2007年第14期3223-3225,共3页
Journal of System Simulation
基金
863计划(2006AA06Z419)
国家自然科学基金资助(40671141)
