正在加载图片...
第8卷第5期 智能系统学报 Vol.8 No.5 2013年10月 CAAI Transactions on Intelligent Systems 0ct.2013 D0:10.3969/j.issn.1673-4785.201304035 网络出版地址:http://www.enki..net/kcms/detail/23.1538.TP.20130929.1120.008.html 一种频域特征的飞机目标提取方法 尹辉,谷峰 (中国人民解放军空军航空大学航空航天情报系,吉林长春130022) 摘要:针对军事遥感影像智能判读的目标定位与提取问题,将先验知识引入到目标提取过程中,提出一种基于频 域特征的飞机目标提取方法该方法通过绘制目标圆周剖面线将二维图像信号简化为一维信号,运用频谱能量的谐 波叠置原理对飞机形状进行频谱分析,确定目标轮廓的固有频率:然后设计多尺度的Gor滤波器组提取图像的纹 理特征图,并进行特征图融合形成目标显著图:最后利用数学形态学的方法对目标显著图进行处理,得到最终的目 标提取结果.实验结果表明,与基于无先验知识的视觉显著性日标检测方法相比,该方法抗干扰能力强,能够清楚、准 确、完整地提取目标区域,是一种有效的飞机目标提取方法. 关键词:频域特征;飞机;目标提取:频谱分析:Gaor滤波器组;显著图 中图分类号:TP753文献标志码:A文章编号:1673-4785(2013)05-0459-06 中文引用格式:尹辉,谷峰.一种频域特征的飞机目标提取方法[J].智能系统学报,2013,8(5):459464. 英文引用格式:YIN Hui,GU Feng.A method of extracting aircraft target based on the frequency domain feature[J】.CAAI Trans- actions on Intelligent Systems,2013,8(5):459-464. A method of extracting aircraft target based on the frequency domain feature YIN Hui,GU Feng Department of Aviation and Spaceflight Intelligence,Aviation University of Air Force,Changchun 130022,China) Abstract:Focusing on the positioning and extraction of the intelligent interpretation in the military remote sensing image,prior knowledge is introduced into the target extraction process,so as to propose an aircraft target extraction method based on the frequency domain feature.In this method,by drafting the target circumference profile,the two-dimensional image signal is simplified into the one-dimensional signal,the harmonic superposition principle on the energy of frequency spectrum is applied to frequency spectrum analysis of aircraft shape,so as to determine the natural frequency of the target outline;then,by the design of multi-scale Gabor filter banks,the texture feature di- agram on the image is extracted,and by combination of feature diagrams,a target salient image is formed;finally, the method of mathematical morphology is used to process the target salient image,obtaining the final target extrac- tion result.The experimental results show that,in comparison with the visual saliency target detection method with- out prior knowledge,the method has powerful anti-disturbance ability,it may extract the target domain clearly,ac- curately and completely,and it is a kind of effective aircraft target extraction method. Keywords:frequency domain feature;aircraft;target extraction;frequency spectrum analysis;Gabor filter banks; salient image 在军事遥感影像智能解译领域中,飞机作为一 观的方法[基于特征的方法尝试寻找区分性的图 类重要的目标受到广泛的关注,并一直都被研究和 像特征,然后在合理的几何布局上验证这些特征是 探索,而这方面成型的资料较少,外军的研制情况均 否存在,这类方法简单,容易实现,但特征的选取需 是保密资料.公开的资料中目标检测方法可以分为 要加入一定的先验知识.基于模板的方法能够处理 基于特征的方法[]、基于模板的方法)和基于表 形状和视角较大范围的变化,但对模板要求较高,计 算复杂不适合实时检测.基于表观的方法通过扫描 收稿日期:2013-04-15.网络出版日期:2013-09-29 图像中有重叠的小矩形区域来寻找候选区域,然后 通信作者:尹辉.E-mail:image_.recognition@163.com 用一组具有选择性的检测算法来标定目标,这类方第 8 卷第 5 期 智 能 系 统 学 报 Vol.8 №.5 2013 年 10 月 CAAI Transactions on Intelligent Systems Oct. 2013 DOI:10.3969 / j.issn.1673 ̄4785.201304035 网络出版地址:http: / / www.cnki.net / kcms/ detail / 23.1538.TP.20130929.1120.008.html 一种频域特征的飞机目标提取方法 尹辉ꎬ谷峰 (中国人民解放军空军航空大学 航空航天情报系ꎬ吉林 长春 130022) 摘 要:针对军事遥感影像智能判读的目标定位与提取问题ꎬ将先验知识引入到目标提取过程中ꎬ提出一种基于频 域特征的飞机目标提取方法.该方法通过绘制目标圆周剖面线将二维图像信号简化为一维信号ꎬ运用频谱能量的谐 波叠置原理对飞机形状进行频谱分析ꎬ确定目标轮廓的固有频率ꎻ然后设计多尺度的 Gabor 滤波器组提取图像的纹 理特征图ꎬ并进行特征图融合形成目标显著图ꎻ最后利用数学形态学的方法对目标显著图进行处理ꎬ得到最终的目 标提取结果.实验结果表明ꎬ与基于无先验知识的视觉显著性目标检测方法相比ꎬ该方法抗干扰能力强ꎬ能够清楚、准 确、完整地提取目标区域ꎬ是一种有效的飞机目标提取方法. 关键词:频域特征ꎻ飞机ꎻ目标提取ꎻ频谱分析ꎻGabor 滤波器组ꎻ显著图 中图分类号:TP753 文献标志码:A 文章编号:1673 ̄4785(2013)05 ̄0459 ̄06 中文引用格式:尹辉ꎬ谷峰.一种频域特征的飞机目标提取方法[J]. 智能系统学报ꎬ 2013ꎬ 8(5): 459 ̄464. 英文引用格式:YIN Huiꎬ GU Feng. A method of extracting aircraft target based on the frequency domain feature[J]. CAAI Trans ̄ actions on Intelligent Systemsꎬ 2013ꎬ 8(5): 459 ̄464. A method of extracting aircraft target based on the frequency domain feature YIN Huiꎬ GU Feng (Department of Aviation and Spaceflight Intelligenceꎬ Aviation University of Air Forceꎬ Changchun 130022ꎬ China) Abstract:Focusing on the positioning and extraction of the intelligent interpretation in the military remote sensing imageꎬ prior knowledge is introduced into the target extraction processꎬ so as to propose an aircraft target extraction method based on the frequency domain feature. In this methodꎬ by drafting the target circumference profileꎬ the two ̄dimensional image signal is simplified into the one ̄dimensional signalꎬ the harmonic superposition principle on the energy of frequency spectrum is applied to frequency spectrum analysis of aircraft shapeꎬ so as to determine the natural frequency of the target outlineꎻ thenꎬ by the design of multi ̄scale Gabor filter banksꎬ the texture feature di ̄ agram on the image is extractedꎬ and by combination of feature diagramsꎬ a target salient image is formedꎻ finallyꎬ the method of mathematical morphology is used to process the target salient imageꎬ obtaining the final target extrac ̄ tion result. The experimental results show thatꎬ in comparison with the visual saliency target detection method with ̄ out prior knowledgeꎬ the method has powerful anti ̄disturbance abilityꎬ it may extract the target domain clearlyꎬ ac ̄ curately and completelyꎬ and it is a kind of effective aircraft target extraction method. Keywords:frequency domain featureꎻ aircraftꎻ target extractionꎻ frequency spectrum analysisꎻ Gabor filter banksꎻ salient image 收稿日期:2013 ̄04 ̄15. 网络出版日期:2013 ̄09 ̄29. 通信作者:尹辉. E ̄mail:image_recognition@ 163.com. 在军事遥感影像智能解译领域中ꎬ飞机作为一 类重要的目标受到广泛的关注ꎬ并一直都被研究和 探索ꎬ而这方面成型的资料较少ꎬ外军的研制情况均 是保密资料.公开的资料中目标检测方法可以分为 基于特征的方法[1 ̄2] 、基于模板的方法[3] 和基于表 观的方法[4] .基于特征的方法尝试寻找区分性的图 像特征ꎬ然后在合理的几何布局上验证这些特征是 否存在ꎬ这类方法简单ꎬ容易实现ꎬ但特征的选取需 要加入一定的先验知识.基于模板的方法能够处理 形状和视角较大范围的变化ꎬ但对模板要求较高ꎬ计 算复杂不适合实时检测.基于表观的方法通过扫描 图像中有重叠的小矩形区域来寻找候选区域ꎬ然后 用一组具有选择性的检测算法来标定目标ꎬ这类方
向下翻页>>
©2008-现在 cucdc.com 高等教育资讯网 版权所有