农药喷雾飞行器越界检测算法实现  

Research on Pesticide Spraying Craft Cross-border Detection Algorithm

作　　者：穆春阳[1] 王晓龙[1] 马行[1] 张盼盼[1] 陈雪涛[1] 

机构地区：[1]北方民族大学信息与通信技术研究所,银川750021

出　　处：《农机化研究》2016年第4期6-10,共5页Journal of Agricultural Mechanization Research

基　　金：国家自然科学基金项目(61162005;61163002);宁夏自然科学基金项目(NZ14107);机器人技术与系统国家重点实验室开放研究基金项目(SKLRS-2013-MS-05)

摘　　要：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。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 identifying 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%.

关 键 词：农药喷雾飞行器 越界检测 混合高斯模型 形状特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]