采用测地线活动轮廓模型检测与跟踪运动目标

水平集几何活动轮廓模型能较好地适应曲线的拓扑变化.为了跟踪和获取刚体和非刚体运动目标的轮廓信息,提出了一种基于改进测地线活动轮廓(GAC)模型和Kalman滤波相结合的算法以检测和跟踪运动目标.该算法首先采用高斯混合模型和背景差分获取目标的运动区域,在运动区域内采用引入距离规则化项的GAC模型进行曲线演化,使改进GAC模型在运动目标的真实轮廓处收敛;然后通过结合Kalman滤波预测目标下一帧的位置,实现对目标轮廓跟踪.实验结果表明,该方法适用于刚体和非刚体目标,在部分遮挡的情况下也能保持良好的检测和跟踪效果.

Moving object detection and tracking based on geodesic active contour model

The geometric-active contour model based on the level set can better handle the variations of the curve topology. In order to track a rigid or non-rigid moving object and extract its contour information, we propose a combination method of the improved geodesic active contour (GAC) model and Kalman filter. In this method, the moving regions of the object are determined by using Gaussian mixture model and the background difference method; the GAC model with a distance regularization term is used to perform the curve evolution in the moving region, making the evolving curve approaching to the true contours of the object. The tracking of the moving object is realized by using Kalman filter to predict the object position of the next frame. Experimental results show that the proposed method is applicable to both rigid and non-rigid objects, achieving good detection and tracking effect even in the case of partial occlusion.