- 104 - http://www.sjie.org Scientific Journal of Information Engineering June 2014, Volume 4, Issue 3, PP.104-110 The Duplicate Variance Lacunarity Feature of Vehicle Targets in SAR Imagery and the Discriminating Kai Huang # , Yu Li, Yi Su School of Electronic Science and Engineering, National University of Defense Technology, Changsha Hunan 410073, China # Email: [email protected]Abstract A duplicate variance lacunarity feature was proposed by analyzing the issue of lacunarity and multi-scale lacunarity features. The box mass is defined as the intensity variance of pixels locating in the box when obtaining this feature by DBC (Differential Box Counting). Then we add the ratio of the boxes mass variance to their square mean with 1 and get the duplicate variance lacunarity. The experiments on this discriminating feature using a large number of SAR imagery ROIs (Region of Interest) and target discrimination of large SAR scene are carried out to demonstrate the outstanding capability of the duplicate variance lacunarity in discriminating the vehicle targets in SAR imagery. Keywords: SAR; Targets Discriminating; Fractal; Duplicate Variance Lacunarity SAR 图像车辆目标二重方差间隙度特征及其鉴别 * 黄凯,李禹,粟毅 国防科学技术大学 电子科学与工程学院,湖南 长沙 410073 摘 要:在对间隙度特征及多尺度间隙度特征存在问题分析的基础之上,提出二重方差间隙度特征。利用差分盒维法提取 该特征时,将盒子质量定义为盒子内像素幅度值的方差,再用盒子质量的方差与其均值平方之比加上 1 得到二重方差间 隙度。运用大量 SAR 图像 ROI 切片鉴别特征统计分析实验,及对整幅场景 SAR 图像目标鉴别结果对比分析实验,证明 本文所提二重方差间隙度特征对 SAR 图像车辆目标的鉴别性能明显提高。 关键词:SAR;目标鉴别;分形;二重方差间隙度 引言 合成孔径雷达(SAR)图像车辆目标解译处理过程一般包括目标检测、目标鉴别、目标分类三个步骤。目 标检测阶段常采用恒虚警检测(CFAR),但这种方法无可避免地会出现大量虚警。同时,由于目标分类难度 较大,计算代价昂贵,这就要求目标鉴别阶段在防止漏警的情况下,尽可能多地剔除虚警。对于目标检测 阶段得到的感兴趣区域(ROI)切片,目标鉴别阶段常采用特征提取的方法,选取一种在目标与背景这两类切 片当中类内散度低,类间距离大的特征,以将两类切片区分开来。其中分形特征是比较常用的一类特征。 Mandelbrot 提出的分形理论长于描述自然界中不规则的地物形态。基于此,Kreithen [1] 和 Novak [2] 设定车 辆目标的 Hausdorff 分形维数小于自然地物,将其作为一种鉴别特征用于 SAR 图像目标鉴别。不过,分形维 数特征虽然指出了 SAR 图像纹理的不规则程度,却没有描述其幅度值起伏的快慢,最终鉴别效果并不理想。 李禹等在此基础上提出了另一种分形特征,即间隙度特征 [3][4] 。该特征可以定量描述 SAR 图像当中像素起伏 * 基金资助:受国家自然科学基金项目支持资助(基金号:61171135)。
7
Embed
The duplicate variance lacunarity feature of vehicle targets in sar imagery and the discriminating
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
- 104 -
http://www.sjie.org
Scientific Journal of Information Engineering June 2014, Volume 4, Issue 3, PP.104-110
The Duplicate Variance Lacunarity Feature of
Vehicle Targets in SAR Imagery and the
Discriminating Kai Huang
#, Yu Li, Yi Su
School of Electronic Science and Engineering, National University of Defense Technology, Changsha Hunan 410073, China