基于GMM和超像素马尔科夫随机场的运动目标检测
|
摘要
针对传统的基于高斯混合模型(GMM)的运动目标检测算法抗噪声性能差、易受动态背景干扰的缺陷,提出一种高斯混合建模与超像素马尔科夫随机场(MRF)相结合的运动目标检测方法。采用GMM对视频图像进行建模,初步标记出前景目标区域;对原始图像进行超像素分解,并根据GMM提取的前景图像得到概率超像素图像;采用MRF建模对概率超像素图像建模得到最终的运动目标前景图像。通过实验对比分析,表明提出的算法对噪声干扰、动态扰动背景等复杂场景均可以得到优于传统算法的结果。
Aiming at the defect that the traditional moving objects detection( MOD) algorithm based on Gaussian mixture model( GMM) has poor anti-noise performance,sensitive to dynamic background interference,a Gaussian mixture modeling combined with Markov random field( MRF) moving target detection method. Firstly,the video image is modeled by GMM,and foreground target area is initially marked. The original image is decomposed into superpixels. According to foreground image extracted by GMM,the probabilistic superpixel images is obtained. MRF modeling is used to model the probabilistic superpixel images and to obtain the final foreground image of moving target. The comparison analysis of the final experiments show that the proposed algorithm can get better results than traditional algorithms for complex scenes such as noise interference and dynamic disturbance background.
Aiming at the defect that the traditional moving objects detection( MOD) algorithm based on Gaussian mixture model( GMM) has poor anti-noise performance,sensitive to dynamic background interference,a Gaussian mixture modeling combined with Markov random field( MRF) moving target detection method. Firstly,the video image is modeled by GMM,and foreground target area is initially marked. The original image is decomposed into superpixels. According to foreground image extracted by GMM,the probabilistic superpixel images is obtained. MRF modeling is used to model the probabilistic superpixel images and to obtain the final foreground image of moving target. The comparison analysis of the final experiments show that the proposed algorithm can get better results than traditional algorithms for complex scenes such as noise interference and dynamic disturbance background.
引文
[1] CHEN S Y,XU T F,LI D Q,et al. Moving object detection using scanning camera on a high-precision intelligent holder[J].Sensors,2016,16:1758-1765.
[2] CHEN H Y,QIE L Z,YANG D D,et al. Visual background extraction algorithm based on superpixel information feedback[J].Acta Optica Sinica,2017,37(7):1502-1518.
[3] KUMAR P,SINGHAL A,MEHTA S,et al. Real-time moving object detection algorithm on high-resolution videos using GPUs[J].Journal of Real-Time Image Processing,2016,11:93-109.
[4] CHENG C L,YU L,HUAN P,et al. Precise object detection using iterative superpixels grouping method[J]. Journal of Electronic Science and Technology,2017,15(2):153-161.
[5] GU W,LZ H,HAO M. Change detection method for remote sensing images based on an improved Markov random field[J].Multimedia Tools Application,2017,76:17719-17734.
[6] WANG Y,YIPIERRE J,FATIH P,et al. CDnet 2014:An expanded change detection benchmark dataset[C]∥The 2nd IEEE Change Detection Workshop in conjunction with CVPR,2014:393-400.
[7] ZIVKOVIC Z. Improved adaptive Gaussian mixture model for background subtraction[C]∥International Conference on Pattern Recogition,2004:28-31.
[8] HEIKKILA M,PIETIKAINEN M. A texture-based method for modeling the background and detecting moving objects[J]. IEEE Transaction on Pattern Analysis and Machine Intelligence,2006,28(4):657-662.
[9] CHRIS S,GRIMSON W E L. Adaptive background mixture models for real-time tracking[C]∥IEEE International Conference on Computer Vision and Pattern Recognition,1999,6(2):246-252.
[10] BARNICH O,VAN D. Vi Be:A universal background subtraction algorithm for video sequences[J]. IEEE TIP,2011,20(6):1709-1724.
[11] DOMENICO D. B,ANDREA P,LUCA I. Background modeling in the maritime domain[J]. Machine Vision and Applications,2014,25(5):1258-1269.
[2] CHEN H Y,QIE L Z,YANG D D,et al. Visual background extraction algorithm based on superpixel information feedback[J].Acta Optica Sinica,2017,37(7):1502-1518.
[3] KUMAR P,SINGHAL A,MEHTA S,et al. Real-time moving object detection algorithm on high-resolution videos using GPUs[J].Journal of Real-Time Image Processing,2016,11:93-109.
[4] CHENG C L,YU L,HUAN P,et al. Precise object detection using iterative superpixels grouping method[J]. Journal of Electronic Science and Technology,2017,15(2):153-161.
[5] GU W,LZ H,HAO M. Change detection method for remote sensing images based on an improved Markov random field[J].Multimedia Tools Application,2017,76:17719-17734.
[6] WANG Y,YIPIERRE J,FATIH P,et al. CDnet 2014:An expanded change detection benchmark dataset[C]∥The 2nd IEEE Change Detection Workshop in conjunction with CVPR,2014:393-400.
[7] ZIVKOVIC Z. Improved adaptive Gaussian mixture model for background subtraction[C]∥International Conference on Pattern Recogition,2004:28-31.
[8] HEIKKILA M,PIETIKAINEN M. A texture-based method for modeling the background and detecting moving objects[J]. IEEE Transaction on Pattern Analysis and Machine Intelligence,2006,28(4):657-662.
[9] CHRIS S,GRIMSON W E L. Adaptive background mixture models for real-time tracking[C]∥IEEE International Conference on Computer Vision and Pattern Recognition,1999,6(2):246-252.
[10] BARNICH O,VAN D. Vi Be:A universal background subtraction algorithm for video sequences[J]. IEEE TIP,2011,20(6):1709-1724.
[11] DOMENICO D. B,ANDREA P,LUCA I. Background modeling in the maritime domain[J]. Machine Vision and Applications,2014,25(5):1258-1269.