摘要
针对现有智能视频中运动目标检测算法存在的问题,提出一种新的对称差分及背景减除相融合的算法。该算法基于子块操作,首先利用高斯分布的概率特性,分离出运动变化区域和静止区域,对分割阈值的选取进行了改进;然后背景重构;最后通过背景和变化区域相差分得到精确运动目标分割。实验结果表明,该方法能够对监控场景中运动目标进行有效的分割,对光线变化、背景干扰不敏感,具有较好的鲁棒性和实用性。
A novel moving object fusion algorithm based on background subtraction and symmetrical differencing is proposed to overcome the problems of the current moving target detection algorithm in the intelligent video.This algorithm is based on sub-block operation： step one,according to gaussian probability distribution characteristics,isolate the motion area from the rest,and improve the selection of the segmentation threshold;step two,reconstruct the background;step three,achieve the accurate moving object segmentation by differencing the background and changing area.Experimental results demonstrate that this method can effectively isolate the motion object in monitoring scenes,and is immune to light changing and background interference.It is of great robustness and practicality.
出处
《现代电子技术》
2010年第10期96-98,113,共4页
Modern Electronics Technique
关键词
背景重构
对称差分
分块
高斯分布
background reconstruction
symmetrical differencing
blocking
Gaussian distribution
