基于无人机图像的实时人工目标检测  

Real Time Unnatural Object Detection for an In-flight UAV

作　　者：朱剑佑[1] 

机构地区：[1]中国人民解放军63961部队

出　　处：《无线电工程》2007年第10期15-16,20,共3页Radio Engineering

摘　　要：从无人机飞行时获得的图像中实时地判断出自然目标和人工目标是许多想定任务的基本需求,例如从海洋中检测出船只、从树丛中检测出车辆以及从自然环境中检测出建筑物和其他人造目标等。然而在实时处理中,图像处理过程总是受到计算时间及计算能力的限制,这就需要研究那些特性并能更好地用来区分人工目标和自然目标。探讨了基于色彩之外的其他特性来区分人工目标的方法,并给出一种所需计算量较小的简单算法。　The judgement of natural and unnatural objects in real time from Unmanned Aerial Vehicle（UAV） is the basis of many mission scenarios including detection of ships at sea,vehicles in wood,buildings and other man made objects in a natural environment.In real-time processing,the problem with many image processing operations is the time and computing capability required.With these limitations in mind,it was decided to investigate how well natural and unnatural objects could be distinguished using fundamental concepts and features.This investigation explores unnatural object detection based on features other than color,presents a simple techniques requiring minimal computing.

关 键 词：人工目标检测 实时图像处理 无人机 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]