基于TopHat背景建模的运动小目标检测算法分析  

Analysis of Moving Small Target Detection Algorithm Using TopHat Background Modeling

作　　者：贺遵亮 黄营磊 HE Zunliang;HUANG Yinglei(Hunan Allin Creatview Information Technology Co.,Ltd.,Hunan 412007,China)

机构地区：[1]湖南傲英创视信息科技有限公司,湖南412007

出　　处：《电子技术（上海）》2022年第8期64-66,共3页Electronic Technology

摘　　要：针对复杂背景下的运动小目标实时检测问题,提出一种对TopHat滤波结果进行建模的检测算法,并设计出快速计算方法。利用灰度形态学TopHat滤波抑制背景增强目标,在滤波图像序列上执行最大值操作,构建当前帧的背景图像,然后使用当前帧TopHat滤波图像减去背景图像得到前景图,最后采用适当阈值从前景图中提取目标信息。实验结果表明,与MoG2、ViBe+两种经典算法相比,提出的算法针对定量范围内的运动小目标具有1.4倍、3.39倍的性能提升和1.5倍、2.06倍的速度提升。Aiming at the real-time detection of moving small targets in complex background, a detection algorithm based on the modeling of TopHat filtering results is proposed, and a fast calculation method is designed. Firstly, the gray morphological TopHat filter is used to suppress the background enhancement target, then the maximum operation is performed on the filtered image sequence to construct the background image of the current frame, then the background image is subtracted from the TopHat filtered image of the current frame to obtain the foreground image, and finally the target information is extracted from the foreground image by using an appropriate threshold. The experimental results show that compared with MoG2 and ViBe+, the proposed algorithm has 1.4 times and 3.39 times performance improvement and 1.5 times and 2.06 times speed improvement for small moving targets in the quantitative range.

关 键 词：背景建模 小目标检测 运动目标检测 TopHat 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]