中文摘要：

人体行为识别在视频监控和人机交互中具有重要应用，利用深度数据进行行为识别是近年兴起的技术，并取得了一定的进展，但还没有一个公认的、鲁棒性好的行为描述方法，且性能有待提高。针对以上问题，本文提出了3种鲁棒的、深度数据上的行为描述方法，并结合支持向量机（SVM）分类器在两个公开的且具有挑战性的深度数据集上对它们进行评估。实验结果表明，本文提出的行为描述方法具有较好的区分性和鲁棒性，其性能比一些先进的且具有代表性的算法性能更好。

英文摘要：

Human behavior recognition has important applications in video surveillance and human-computer interaction, and it is also an important and challenging issue. Behavior recognition using depth information becomes a hot and popular spot. So far, human action recognition based on depth information has made some achievements, but it still does not have robust descriptors, and their accuracies are still not satisfactory. To solve above problems, this paper proposes three robust action descriptors for depth information, and then SVM classifier is adopted. Experiments on two public and challenging behavior rec- ogmition datasets show that our descriptors have strong robustness, distinction and stability, whose performance is much better than that of the state of the art algorithms.