摘要
提出一类新型的嵌入式视频活动目标检测算法,该算法采用Surendra算法对背景进行更新以减低系统误报警的几率,对连续3帧图像分别采用差帧法,对2次帧差取交集实现对前景目标的模糊跟踪,而后对粗糙的运动区域图像进行阈值面积消去处理和数学形态学运算,最后实现目标定位跟踪。仿真结果表明,与传统的二阶帧差的方法相比,视频活动的目标检测算法具有高实效,高精度的特点。
This paper presents a novel algorithm for moving objects detection in video. Such an algorithm first employs Surendra algorithm to renew video background to improve accuracy of detection results. It takes the difference of sequential three frames,and gets their intersection part to realize coarse tracing for active objects. The moving objects are traced and located by applying mathematical morphology operation in coarse active areas. Simulation results show that, compared with traditional second order frame-difference method, our algorithm is more efficient and accurate.
出处
《现代电子技术》
2007年第18期24-26,共3页
Modern Electronics Technique
基金
福建省资助省属高校项目基金(2006F5056)