基于全局最小化活动轮廓的多目标检测跟踪

Multiple objects detecting and tracking based on global minimization of active contour model

为了在噪声干扰以及目标和背景颜色相近情况下实现多目标跟踪,提出一种基于快速全局最小化的活动轮廓模型的目标检测跟踪算法。该算法结合了基于边缘的活动轮廓模型和基于区域的活动轮廓模型,对能量泛函进行全局最小化来检测目标活动轮廓,用卡尔曼滤波预测目标下一帧的特征信息,然后用改进的最近邻法进行多目标跟踪。对图像序列的实验结果表明该算法能有效地对运动背景下多目标进行跟踪。

This paper proposeded a multiple objects detecting and tracking algorithm, which was based on the fast global minimization of active contour model. The algorithm combined edge-based active contours model and region-based active contours into the framework of global minimization the energy functional to detect the objects contours. Predicted the features＇ information of the objects using the Kalman filter. Then tracked the multiple objects by using the improved nearest neighbor method. Experimental results demonstrate the efficiency of the proposed approach.