CAJ | 학술논문

针对SAR图像飞机目标识别过程中的目标识别问题，提出一种基于峰值匹配的目标识别方法。该方法首先使用基于8邻域像素检测局部极大值的峰值提取方法提取目标和模板的峰值特征点；然后对目标进行方位角计算，设定置信区间从而缩小需要匹配的模板库；最后计算目标图像峰值点集与模板图像峰值点集的匹配代价函数，当匹配代价函数取最小值时表明目标与相应模板图像相匹配。实验结果表明该算法有效，且分类性能和分类效率较现有的一些算法有所提升。
침대SAR도상비궤목표식별과정중적목표식별문제，제출일충기우봉치필배적목표식별방법。해방법수선사용기우8린역상소검측국부겁대치적봉치제취방법제취목표화모판적봉치특정점；연후대목표진행방위각계산，설정치신구간종이축소수요필배적모판고；최후계산목표도상봉치점집여모판도상봉치점집적필배대개함수，당필배대개함수취최소치시표명목표여상응모판도상상필배。실험결과표명해산법유효，차분류성능화분류효솔교현유적일사산법유소제승。
A target recognition method based on peak?value matching is proposed for the recognition problems in recogni?tion process of aircraft target in synthetic aperture radar (SAR) image. This method adopts the peak?value extraction method based on local maximum value detected by 8?neighbourhood pixel to extract the peak?value feature points of the target and tem?plate. The azimuth angle of the target is calculated,and the confidence interval is set to shrink the template library which is needed to match. The marched cost functions of the peak point sets in target image and template image are calculated. The target is marched with the template image when the minimum value is get from the marched cost function. The experimental results show that the proposed method is effective. The classification performance and classification effectiveness were improved by this method in comparison with the existing algorithms.