CAJ | 학술논문

针对当前的农药喷雾飞行器需要人工时刻观察是否飞出农田边界这一现状，提出了一种基于高斯混合模型的农药喷雾飞行器越界自动检测算法。首先，采用混合高斯模型进行目标检测，然后通过计算形状描述符获取其形状特征并进行目标识别定位，最后通过计算目标质心位置结合目标的运动方向判断其是否出界。实验结果表明：该算法能够准确地判断飞行器的越界行为，准确率达到91．31％以上。
침대당전적농약분무비행기수요인공시각관찰시부비출농전변계저일현상，제출료일충기우고사혼합모형적농약분무비행기월계자동검측산법。수선，채용혼합고사모형진행목표검측，연후통과계산형상묘술부획취기형상특정병진행목표식별정위，최후통과계산목표질심위치결합목표적운동방향판단기시부출계。실험결과표명：해산법능구준학지판단비행기적월계행위，준학솔체도91．31％이상。
In view of the reality that people have to supervise whether pesticide spraying aircraft fly out of the boundary of farmland all the time , an algorithm of detecting aircraft cross-border is proposed to solve this problem above .Firstly , the Gaussian mixture model is used for target detection .And then characteristics of shape descriptor are obtained for identif-ying the targets .Finally , the target centroid and moving direction are combined to determine whether the aircraft was out of the boundary .Results show that this algorithm could make a right judgment for aircraft cross-border actions accurately and quickly , where the accuracy of detection could reach to 91 .31%.