Abstract: In the traditional particle filter tracking，color histogram is usually used as the features vectors，there are some limits because of the loss of space distribution.To overcome this problem，an efficient tracking algorithm based on object feature and spatial information fusion within particle filter framework is proposed.The dissimilarity between the referenced target and the target candidate is expressed by not only color，but also space distribution.At the same time for particle filter can be characterized by parallelism，OpenMP shared memory parallel computing is used for the acceleration of particle filter tracking.Experiments show that the algorithm can improve accuracy and speed for particle filter object tracking in the objectives and background of complex applications.

Key words: information fusion, object tracking, particle filter, parallel computing

摘要： 传统的基于颜色直方图的粒子滤波跟踪算法不能很好地利用跟踪对象的空间结构信息，因此在邻域颜色相似或目标模型微小变化时，不能取得良好的跟踪效果。提出一种融合目标特征和目标空间位置信息的粒子滤波跟踪算法，该算法鉴于目标空间位置包含跟踪对象一定的结构信息，可以和目标特征互为补充，利用定义的融合目标特征和目标空间位置的度量函数来进行跟踪对象相似度度量，以提高跟踪算法的稳健性和精确性。同时针对粒子滤波计算粒子相似度时可并行的特点，运用OpenMP共享存储并行计算进行粒子滤波跟踪的加速。实验表明，基于融合目标特征和空间信息的粒子滤波跟踪算法能得到更鲁棒的跟踪效果，可以有效地提高目标跟踪的速度。

关键词: 信息融合, 目标跟踪, 粒子滤波, 并行计算