一种基于帧间差的快速UAV运动目标检测方法  

A Method for UAV Moving Target Detection Based onInterframe Difference

作　　者：刘玉婷[1] 席庆彪[1,2] 刘慧霞[2] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072 [2]西北工业大学第365研究所,陕西西安710065

出　　处：《计算机仿真》2014年第7期36-40,共5页Computer Simulation

基　　金：国家自然科学基金(61074155)

摘　　要：针对中小型UAV(Unmanned Aerial Vehicle)平台运动目标检测系统由于受机载摄像条件及外界环境等因素影响,造成检测结果准确性差和实时性低的问题,提出一种帧间差和背景差相结合的UAV运动目标检测算法。通过分别计算帧间差和背景差的图像并进行逻辑或运算提取运动目标,采用三帧差分法计算差分图像,单高斯背景模型法提取背景图像,背景自适应更新法更新背景图像。仿真结果显示:算法满足UAV平台运动目标检测要求,实时性及准确性优于背景差分和帧间差分等算法。Focus on the problems of the airborne camera condition and external environment in UAV moving target detection system, which results in low accuracy and real-time performance, a novel target detection method in combi- nation with interframe difference and background difference was proposed, which calculates the interframe difference and background difference of images and extracts moving targets by the results of the two algorithms. The difference image was obtained by calculating three continuous frame images. The background image.was extracted by the single gaussian background model, The background image was updated according to the adaptive method. Simulation results show that the algorithm in the real-time and accuracy is superior to the background subtraction and frame difference algorithm, which can satisfy the UAV target detection requirements.

关 键 词：无人机 三帧差分法 单高斯背景模型 背景自适应更新 运动目标检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]