CAJ | 학술논문

在航拍图像的目标检测中，由于目标外部形态以及复杂背景的影响，对一个有效的航拍目标检测仍然是目标识别领域的一个巨大挑战。提出了一种基于 Hu矩、径向矩（Radial moment）组成的 HRM （Hu Radial Moment）特征和支持向量机（SVM ）的目标检测方法。该方法为首先提取目标的 Hu矩特征，并结合其各向同性的径向矩特征共同构成新的特征向量HRM ，然后采用支持向量机来对特征进行分类、预测。实验结果表明，针对不同的目标外部形态和复杂的外部场景，所提出的方法具有较高的检测率，且是高效的。
재항박도상적목표검측중，유우목표외부형태이급복잡배경적영향，대일개유효적항박목표검측잉연시목표식별영역적일개거대도전。제출료일충기우 Hu구、경향구（Radial moment）조성적 HRM （Hu Radial Moment）특정화지지향량궤（SVM ）적목표검측방법。해방법위수선제취목표적 Hu구특정，병결합기각향동성적경향구특정공동구성신적특정향량HRM ，연후채용지지향량궤래대특정진행분류、예측。실험결과표명，침대불동적목표외부형태화복잡적외부장경，소제출적방법구유교고적검측솔，차시고효적。
In the target detection of the aerial images ,due to the effect from the external morphology of target and com-plex background ,an effective target detection of aerial image is still a great challenge in the field of object recognition . This paper presents the target detection method based on Hu moments ,radial moments (RM ) consisting of HRM (Hu Radial Moment) and support vector machine (SVM ) .At first ,Hu moment feature of target is extracted ,then a new feature vector ,HRM ,is constituted with its isotropic characteristics of radial moments together ,and finally ,the fea-tures are classified and predicted by support vector machines .Experimental results validate that ,for different external morphology goals and complex external scenario ,the proposed method has high detection probability and is efficient .