局部分布信息增强的视觉单词描述与动作识别

An Improved Scheme of Visual Words Description and Action Recognition Using Local Enhanced Distribution Information

传统的单词包(Bag-Of-Words,BOW)算法由于缺少特征之间的分布信息容易造成动作混淆,并且单词包大小的选择对识别结果具有较大影响。为了体现兴趣点的分布信息,该文在时空邻域内计算兴趣点之间的位置关系作为其局部时空分布一致性特征,并提出了融合兴趣点表观特征的增强单词包算法,采用多类分类支持向量机(Support Vector Machine,SVM)实现分类识别。分别针对单人和多人动作识别,在KTH数据集和UT-interaction数据集上进行实验。与传统单词包算法相比,增强单词包算法不仅提高了识别效率,而且削弱了单词包大小变化对识别率的影响,实验结果验证了算法的有效性。

The traditional Bag-Of-Words（BOW） model easy causes confusion of different action classes due to the lack of distribution information among features. And the size of BOW has a large effect on recognition rate. In order to reflect the distribution information of interesting points, the position relationship of interesting points in local spatio-temporal region is calculated as the consistency of distribution features. And the appearance features are fused to build the enhanced BOW model. SVM is adopted for multi-classes recognition. The experiment is carried out on KTH dataset for single person action recognition and UT-interaction dataset for multi-person abnormal action recognition. Compared with traditional BOW model, the enhanced BOW algorithm not only has a great improvement in recognition rate, but also reduces the influence of BOW model＇s size on recognition rate. The experiment results of the proposed algorithm show the validity and good performance.