基于多特征融合的机场FOD图像检测技术  

Research on Intelligent Image Recognition Technology for Foreign Objects in Airport Runway

作　　者：陈济达 汤新民[1] 刘金安 CHEN Ji-da;TANG Xin-min;LIU Jin-an(College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China)

机构地区：[1]南京航空航天大学民航学院,江苏南京211106

出　　处：《航空计算技术》2020年第1期42-45,共4页Aeronautical Computing Technique

基　　金：国家自然科学基金项目资助(61773202);空管国家重点实验室开放基金项目资助(SKLATM201706);四川省科技计划项目资助(2018JZ0030).

摘　　要：为了降低机场外来入侵物检测成本,提出一种基于图像检测的机场异物识别方法。为了提高检测的可靠性,对经典ITTI模型改进,用Canny算子提取边缘特征,用蒙特卡罗方法和模糊数学的知识算出各个特征检测出异物的基本概率。自定义一种方法来表示特征之间的支持度,用D S证据融合理论对特征按照支持度的大小进行融合,对图像进行显著性检测。仿真表明,方法的异物识别准确率达到了92%以上,比单一特征中最高的颜色特征检测提高了11%左右。In order to reduce the cost of detection of foreign object debris in airport flight area,a method that relies on image detection to identify foreign invasive objects at the airport is proposed.The classical ITTI model is improved to improve the reliability of the saliency detection.The Canny operator is used to extract the edge features.The Monte Carlo method and the knowledge of fuzzy mathematics are used to calculate the basic probability of each feature detecting foreign objects.A method is defined to express the support between features.Finally,the D S evidence fusion theory is used to fuse features according to the degree of support,and saliency detection is performed on the processed images.The experiment proves that the accuracy of the foreign object recognition in the image is over 92%,which is about 11%higher than the highest color feature detection in the single feature.

关 键 词：经典ITTI模型 蒙特卡罗法 D S证据融合 显著性检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP751[自动化与计算机技术—计算机科学与技术]