基于EMD图像融合的高光谱异常目标检测
首发时间:2009-03-02
摘要:高光谱图像具有很高的光谱分辨率,较低的空间分辨率,由于高维特性,在通常要经过融合等降维处理,但是融合图像空间细节表达能力比较差。本文针对这一问题结合高光谱图像非线性的特点,应用希尔伯特黄变换理论(HHT),提出了一种基于经验模态分解(EMD)的高光谱数据融合算法,分解并提取了突出异常和轮廓的小尺度细节信息,并根据波段划分自适应子空间进行融合,用融合后的图像检测异常目标。实验表明,这个算法改善了视觉效果, 提高融合图像的空间表达能力,突出了异常点,可以很好的提高目标检测概率及降低虚警概率。
For information in English, please click here
Anomaly Detection for Hyperspectral Imagery Based On EMD Image Fusion
Abstract:Hyperspectral remote sensing images have high spectral resolution and low spatial resolution. Because of the high dimension, we often use fusion or other method to reduce dimension, but the fused spectral images cannot express enough spatial information. So aimed at the nonlinear of Hyperspectral data, we use HHT and apply the method of EMD to image fusion .Through it, we extract spatial information and texture information. Then use ASD (adaptive subspace decomposition) to fusion the image and do the anomaly target detection .Experiments shows that this algorithm can enhance the space and spatial information and greatly improved the performance of the proposed algorithm.
Keywords: image fusion EMD Hyper spectral imagery anomaly detection
论文图表:
引用
No.2981412890123598****
同行评议
勘误表
基于EMD图像融合的高光谱异常目标检测
评论
全部评论0/1000