基于高光谱特征的迷彩伪装评价  

Camouflage assessment based on hyperspectral characteristics

作　　者：华文深[1] 杨佳[1] 刘恂[1] 马左红[2] 

机构地区：[1]军械工程学院 [2]中国人民解放军73101部队

出　　处：《应用光学》2013年第6期964-967,共4页Journal of Applied Optics

摘　　要：伪装效果分析是伪装设计和识别的重要环节,为了对高光谱探测下的目标伪装效果进行有效评价,结合高光谱图像光谱分辨率较高的特点,提出了一种基于高光谱特征的迷彩伪装评价方法。在光谱维,分析并提取伪装目标的光谱特征。在空间维,对图像第一主成分进行多层小波分解,提取高频分量和低频分量的比值、水平分量和垂直分量的比值以及低频图像的对比度、相关性、能量、同质性等6个特征作为纹理特征。利用欧氏距离分别计算光谱特征距离与纹理特征距离,最后将二者进行综合,距离越小说明伪装效果越好。实验数据表明:本文的方法可以量化高光谱成像下的迷彩伪装效果,评价客观。得到迷彩雨衣与草坪背景之间的综合距离为1.3225,两者融合效果最好。Camouflage assessment is an essential part in camouflage design and identification. In order to make an effective camouflage evaluation under the hyperspectral detection, we presented a new method based on hyperspectral characteristics considering the characteristics of high spectral resolution of hyperspectral images. Firstly, the spectral information was extracted and analyzed in spectral dimension. Then, the multi-wavelet decomposition was used in spatial di- mension of the first principal components to extract texture information such as the ratio of high frequency component and low frequency component, ratio of horizontal component and vertical component, contrast, correlation, energy and homogeneity of low frequency image. Finally, Euclidean distance was calculated to reflect the similarity of spectrum and texture. Experimental data demonstrates the objectivity of the approach. The comprehensive distance between camouflage raincoat and lawn background is 1. 3225 that proves they fuse best.

关 键 词：迷彩伪装 高光谱成像 光谱特征 纹理特征 

分 类 号：TN911.73[电子电信—通信与信息系统] E951.4[电子电信—信息与通信工程]