中文摘要：

本文针对视频图像中的目标检测问题，基于在线学习方法，研究具有自主学习能力的目标检测系统。该系统由目标检测模块及检测结果验证模块组成，目标检测模块由具有在线学习能力的分类器构成，检测结果验证模块通过粒子滤波对系统检测到的目标估计其后验概率分布，从而验证该检测结果是否为真实目标，并从中获得在线学习样本，实现检测系统的无需人工干预的自适应学习。为了减少验证错误对系统在线学习的影响，提出基于多信息融合的粒子滤波验证方法提高系统的鲁棒性。本文对PETS2006视频序列以及公交车内视频序列进行了目标检测实验，证明了其较强的自适应能力和较好的检测效果。

英文摘要：

The main research content is designing intelligent object detection system which can self learning and improve its detection performance based on online learning method. The system is composed by object detection module and valida- tion module. Online learning classifier was used in the object detection module. Samples, which was used to train the clas- sifier online, are acquired and labeled automatically from validation module. Instead of using another detection algorithm to label the new sample like other online learning framework, we ensure the correct label from particle filter tracking. The likelihood distribution of particle sets are used to verify the object detection results. This can greatly reduce the effort by la- beler. Furthermore, in order to reduce the impact of validation error, the Muhi-information fusion particle filter method is used to improve the robustness of the online learning object detection system. Experimental results on PETS2006 dataset and bus video dataset are provided to show the adaptive capability and high detection rate.