运动目标跟踪在社会安防、交通监控、医学检查、军事防御等方面都发挥着举足轻重的作用，是计算机视觉领域的核心内容，也是社会科技发展必不可少的推动力量，具有广泛的实用价值和广阔的应用前景。在视频图像跟踪过程中，目标的实时跟踪、跟踪算法的鲁棒性是视频图像跟踪过程中对运动目标准确跟踪的基本要求。据此，具体的研究内容为： 首先，介绍了能够有效处理非线性、非高斯问题的粒子滤波算法，对粒子滤波理论做了详细介绍，同时通过对比实验证明了算法的有效性，并通过变量改变，确定了算法的最佳跟踪条件。 其次，提出了基于主动轮廓模型的粒子滤波跟踪算法，按照主动轮廓模型获取各目标的外形特征，并以此作为目标模板在后续帧进行目标匹配，同时通过目标轮廓获取目标位置，以此为依据进行粒子滤波的样本来源，预测下帧图像中目标的位置，并在预测区域进行模板匹配，若匹配后目标与模板发生较大偏差，则利用主动轮廓模型及时修正、更新模板；否则，继续进行下一帧图像中目标的跟踪。 再次，根据图像SIFT特征与团块特征在图像梯度变化急缓区域的互补性，将二者有机的结合进行目标识别，提出了结合团块与SIFT特征的目标识别算法对再现目标进行识别。识别过程中，将提取出两种特征的目标做为模板，在图像帧中进行目标识别，提出了模板记忆与更新算法。 最后，结合所提出的基于主动轮廓的粒子滤波跟踪算法与具有团块与SIFT特征的目标再识别算法，提出了新的目标再次识别跟踪算法，并应用新算法进行仿真实验，新算法有效提高了跟踪过程的准确性与鲁棒性。

Moving target tracking have played a pivotal role in field of social security, traffic monitoring, medical examination, military defense. It is not only the core content of computer vision, but also the driving force for the development of social science and technology. It own a wide range of prospects at practical and application. In the image tracking process, real-time tracking and the robustness of tracking is a basic requirement for accurate tracking. Accordingly, the specific contents as follows: The article introduces the particle filter which could effectively handle non-linear and non-Gaussian system at first. The article made a detail introduction about the particle filter theory. Then through the comparative experiments demonstrate the effectiveness of the algorithm and change the variables found out the algorithm's best track conditions. Secondly, this paper proposes a particle filter tracking algorithm based on active contour models. According to the active contour model get the shape characteristics of the target as a template matching in subsequent frames. At the same time, get the target position and obtain some points as the source sample of particle filter then predict the target position of next frame and match template at predicted position. When there are large deviation between the obtained target and template correct the template using active contour model. Thirdly, according to the complementary of SIFT features and clumps proposed the target recognition algorithm based on clumps with SIFT features. At recognition process, take target which extracted the features of SIFT features and clumps as template to recognise in the other frames. Meanwhile a algorithm of the template update and memory is proposed. Finally, combined the algorithm of particle filter tracking based on active contour model and the target recognition algorithm based on clumps with SIFT features presented in this paper. By the simulation results verify the effectiveness of the algorithm.