基于统计模型和活动轮廓的运动目标检测与跟踪

提出了一种静止背景视频序列中运动目标的检测与跟踪方法.对连续两帧图像序列作差分计算,对差分图像的灰度分布建立混和高斯模型(GMM),采用期望最大化(EM)算法估计模型参数,并引入基于GMM模型的边界检测算子,进而构造运动边界图像.改进静态图像轮廓提取算法GVF-Snake,利用运动边界图像修改GVF-Snake的能量项,使其能够提取视频序列中运动目标的轮廓.采用一种根据目标区域自动初始化轮廓的方法解决Snake初始轮廓需要手工设定的问题,采用一阶差分预测算法加快轮廓收敛速度.利用改进的GVF-Snake算法对运动目标进行检测与跟踪,结果表明,该算法对刚性和非刚性两类目标都具有较好的检测与跟踪效果.

Motion detection and object tracking based on statistical model and active contour

An approach was proposed to detect and track moving objects in a static background video sequence.The difference of two successive frames was computed,and a Gaussian mixture model(GMM) of the gray-level distribution of the difference image was constructed,whose parameters were estimated by expectation maximization(EM) algorithm.Based on the model,a motion detection operator was introduced to generate a motion border image.Then,a static image based algorithm,GVFSnake,was improved.The energy entry was modified by using the motion border image so that it could be used in video sequences.A method was proposed to initialize the Snake automatically,and the 1st-order difference predictive algorithm was used to accelerate the convergence of Snake.Experimental results prove that the algorithm is effective for both rigid and non-rigid objects.