摘要
针对传统混合高斯模型(GMM)在检测运动目标时存在噪声、计算量大、效果不佳等问题,提出了一种混合四帧差分算法的改进混合高斯目标检测方法。通过选定不同规则,分别更新前后帧图像的学习速率来消除"鬼影";提出一种删除多余、过期背景模型的方法来减少计算量;最后通过形态学处理解决"空洞"问题。实验证明所提算法相对传统混合高斯模型算法在消除噪音、提取运动目标完整轮廓、解决光照变化等问题上具有较好效果,并且能很好的解决遮挡物问题。
Aiming at theproblems of noise,large amount of calculation and the poor effect in detecting moving object with the traditional Gaussian Mixture Model(GMM),this paper proposes a hybrid Four-frame difference to improve the GMM object detection method.The"ghost shadow"is eliminated by selecting different rules and updating the learning rate of before and after frame images respectively. A method to remove redundant and outdated background models is proposed to reduce the computational load. Finally,the problem of the"hole"is solved by morphological processing.The experimental results show that compared with the traditional Gaussian Mixture Model,the proposed algorithm has a good effect on eliminating noise,extracting the complete contour of moving object and solving the illumination change,and can solve the problem of obstructions effectively.
引文
[1] YAN Li Luo,HAN Lin Wan,LI Xia Xue,et al.Adaptive moving vehicle detection algorithm based on hybrid background subtraction and frame difference[J].Applied Mechanics and Materials,2014,3468(644-650):930-933.
[2] Ramya P,Rajeswari R. A modified frame difference method using correlation coefficient for background subtraction[J]. Procedia Computer Science,2016(93):478-485.
[3] Lu Guiliang,Zhou Yu,Yu Yao,et al. A novel approach for foreign substances detection in injection using clustering and frame difference[J]. Sensors,2011,11(10):9121-9135.
[4] Wu JR,Du XL,Liu HT. An improved adaptive background modeling algorithm based on Kalman filter[J]. Transducer&Micro System Technologies,2012,31(1):52-54,58.
[5] ThierryBouwmans,CarolineSilva,CristinaMarghes,et al. On the role and the importance offeatures for background modeling and foreground detection[J].Computer Science Review,2018,28:26-91.
[6]李广,冯燕.基于SIFT特征匹配与K-均值聚类的运动目标检测[J].计算机应用,2012,32(10):2824-2826.
[7] Pengcheng Han,Junping Du,Jun Zhou,et al. An object detection method using wavelet optical flow and hybrid linear-nonlinear classifier[J]. Mathematical Problems in Engineering,2013(2013-9-4),2013(1):831-842.
[8]杜鹃,吴芬芬.高斯混合模型的运动目标检测与跟踪算法[J].南京理工大学学报,2017,41(1):41-46.
[9]刘静,王玲.混合高斯模型背景法的一种改进算法[J].计算机工程与应用,2010,46(13):168-170.
[10]张燕平,白云球,赵勇,等.应用改进混合高斯模型的运动目标检测[J].计算机工程与应用,2010,46(34):155-157,223.
[11]孙毅刚,李鸿,张红颖.基于改进混合高斯模型的运动目标检测算法[J].计算机工程,2012,38(18):166-170.
[12]罗向荣,李其申.一种基于改进混合高斯模型的前景检测[J].计算机应用与软件,2017,34(4):238-242.
[13]汪成亮,周佳,黄晟.基于高斯混合模型与PCAHOG的快速运动人体检测[J].计算机应用研究,2012,29(6):2156-2160.
[14]亢洁,李晓静.基于均值背景与三帧差分的运动目标检测[J].陕西科技大学学报,2018,36(1):148-153.
[15]李战明,谭向阳.基于混合高斯模型与五帧差分的运动目标检测算法[J].计算机与数字工程,2018,46(2):284-288.
[16]张文,李榕,朱建武.基于混合高斯模型与三帧差分的目标检测算法[J].现代电子技术,2012,35(8):57-60.