As an important branch of computer vision, abnormal behavior recognition and detection technology has been widely used in intelligent security, medical monitoring, traffic control and other fields. However, the definition and discrimination methods of abnormal behavior are closely related to the scene factors, and it is very important to appropriately choose the feature extraction as well as abnormal behavior recognition and detection methods according to the characteristics of different application scenarios, so as to improve the warning accuracy. Therefore, this paper reviews the video based human abnormal behavior recognition and detection methods. Firstly, the definition, characteristics and classification of human abnormal behavior are given. Secondly, the feature extraction methods are summarized. The selection of feature extraction methods and the quality of extracted features directly affect the subsequent discrimination results. Then, the paper analyzes and discusses the abnormal behavior discrimination methods from two aspects: abnormal behavior recognition and abnormal behavior detection. The common abnormal behavior detection data sets and related algorithms' performance are also given. Finally, the future research directions of this field are prospected.