摘要
针对海战场图像信息的目标检测与识别,提出了一种适于海战场区域特征的遥感图像目标检测与识别方法.研究采用线性滤波器将图像划分为若干个空间尺度,并对不同空间尺度的图像,根据生物视觉生理特性的原理,提取图像中目标的视觉显著性特征,此特征包含目标不同于其周围区域的程度和空间分布状态.根据分析提取的目标空间特征信息,使用支持向量机对视觉显著性特征图像进行分类,实现目标信息提取,并通过Dempster-Shafer证据理论的分析方法判断目标的相关信息及其置信度,达到识别目标的目的.实验结果表明:此方法能以高可靠性和高精确度检测出海战场图像信息中的目标,获取目标相关信息.
To detect and recognize battlefield＇s targets on the sea reliably, a method for detecting and recognizing the targets from battlefields was presented. A modified saliency analytic method was used to exploit targets and their interactions across space. For picking up the target information, the sup- port vector machine was applied to classify the targets in the saliency map. Dernpster-Shafer evidence theory was used to judge the information and confidence level of the targets. Experiment show that this method can detect highly variable targets and recognize the targets information reliably and effec- tively.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2012年第10期9-12,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
'十二五'国防重点预研项目
