提出了新的运动目标检测和跟踪的方法.首先采用差异累积的方法自适应地更新背景模型,用背景差法进行提取目标.把经过二值化后的视频帧进行分块,设定方块内前景点个数的阈值,接着对方块矩阵进行连通区域合并,从而确定前景点位置.接着提出通过协方差描述算子来跟踪行人的算法.用图像特征的协方差矩阵来表示目标窗口,就可以找到空间特性和统计特性,并在同一个表达式内描述它们之间的关系.通过搜索整个图像找到和当前目标模型的距离最小的区域,得到的最佳匹配区域就是当前帧目标的位置,从而达到跟踪的目的.协方差矩阵将不同形式的特征有效地融合到一起,并且它的维数很小.实验结果表明,算法具有良好的检测效果和实时性能.%The paper sets up a new method about moving object detection and tracking. Firstly, background was updated adaptively using difference accumulation information was detected by background subtraction. Based on the binary result, the image was divided into blocks. Then the threshold was set for the number of foreground pixels. Adjacent regions were combined and foreground regions were located. We propose a new algorithm to track pedestrians using a covariance descriptor. We represent an object window as the covariance matrix of features; therefore we manage to capture the spatial and statistical properties as well as their correlation within the same representation. We search the whole image to find the region which has the smallest distance from the current object model. The best matching region determines the location of the object in the current frame to achieve the goal of tracking. The covariance matrix enables efficiently fusion of different types of features and modalities, and its dimensionality is small. Experiment results showed that the proposed algorithm was effective and efficient.
展开▼
