电光与控制, 2017, 24 (6): 8, 网络出版: 2017-07-10
基于序列图像的运动目标检测新算法
A New Algorithm for Moving Target Detection Based on Image Sequences
运动检测 碰撞检测 多帧关联 SURF配准 序列图像 motion detection collision detection multi-frame association SURF registration image sequence
摘要
基于SURF帧间配准和运动历史图像构建, 提出了一种基于序列图像的运动目标检测新方法。针对运动检测容易出现虚警的情况, 结合多帧关联思想和碰撞检测理论提出多帧碰撞检测算法, 以剔除虚假运动目标。在标准测试集上测试序列仿真验证, 结果表明, 该运动目标检测算法具备较高的查全率以及极高的检测精度和极低的虚警率。
Abstract
On the basis of SURF registration and history motion image building,a new algorithm for moving target detection based on image sequence was proposed.To deal with the problem of false alarm in motion detection,a multi-frame collision detection algorithm was put forward to eliminate the false moving target, which combining multi-frame association with collision detection theory.Simulation was made by using standard test sequences.The results show that the moving target detection algorithm has high recall ratio, high detection precision and very low false alarm probability.
闫可平, 陈洪亮, 文海龙, 李建勋. 基于序列图像的运动目标检测新算法[J]. 电光与控制, 2017, 24(6): 8. YAN Ke-ping, CHEN Hong-liang, WEN Hai-long, LI Jian-xun. A New Algorithm for Moving Target Detection Based on Image Sequences[J]. Electronics Optics & Control, 2017, 24(6): 8.