Computer Engineering and Applications ›› 2013, Vol. 49 ›› Issue (3): 238-242.

Previous Articles     Next Articles

Fast UAV target recognition algorithm based on multiple features

ZHANG Qing1, LIU Huixia2, XI Qingbiao1,2   

  1. 1.College of Automation, Northwestern Polytechnical University, Xi’an 710072, China
    2.No.365 Research Institute, Northwestern Polytechnical University, Xi’an 710065, China
  • Online:2013-02-01 Published:2013-02-18

一种融合多特征的UAV快速目标识别

张  清1,刘慧霞2,席庆彪1,2   

  1. 1.西北工业大学 自动化学院,西安 710072
    2.西北工业大学 第365研究所,西安 710065

Abstract: To settle the problem of identifying the target in the UAV(Unmanned Aerial Vehicle) reconnaissance image in real time, a UAV fast target identify algorithm is proposed which is based on invariant moments and the SIFT(Scale Invariant Feature Transform) features of the image. A similarity function is designed based on invariant moments, which is taken as the fitness function of GA(Genetic Algorithm). The image is searched globally using GA to extract the Region Of Interest(ROI) quickly which may contain targets. The object is located accurately in the ROI region based on the SIFT transform. Simulation results indicate that the method is of strong robustness and can identify the targets effectively and also reduce the recognition time significantly.

Key words: target recognition, Unmanned Aerial Vehicle(UAV), Region Of Interest(ROI), Hu moments, Genetic Algorithm, Scale Invariant Feature Transform(SIFT) algorithm

摘要: 针对UAV(Unmanned Aerial Vehicle)侦察图像快速目标识别问题,提出一种融合多特征的UAV快速目标识别算法。该算法结合图像的不变矩特征和SIFT特征,用不变矩特征构造适应度函数并利用遗传算法的全局搜索能力,在侦察图像中进行搜索,快速提取出可能包含目标的感兴趣区域(Region Of Interest,ROI)。采用尺度不变特征变换算法(Scale Invariant Feature Transform,SIFT)在ROI区域中进行匹配识别,从而精确确定目标的位置。仿真结果表明:该算法的鲁棒性较强,能有效识别特定目标并显著减少识别时间。

关键词: 目标识别, 无人机, 感兴趣区域(ROI), 不变矩, 遗传算法, 尺度不变特征变换(SIFT)算法