针对交通模式识别的问题,本文提出一种基于交通轨迹速度与角度变化有序程度的排列熵特征,利用新增排列熵特征属性进行轨迹交通模式分类.通过对GeoLIfe数据的预处理与特征提取,在获得轨迹基本属性的基础上提取出轨迹速度与角度的排列熵,通过深度神经网络进行训练、分类.实验表明该方法分类准确率有一定提高,说明排列熵属性对识别交通轨迹类别是有效的,可提高轨迹模式识别的准确性.%Aiming at solving the problem in traffic pattern recognition, this paper puts forward a new feature based on the orderly degree of the change of the track traffic speed and angle, using the classification method of new attribute of permutation entropy to classify trajectory traffic pattern. Through the pre-processing and feature extraction of GeoLIfe data, we extract the permutation entropy of the trajectory speed and the angle based on the basic properties of the trajectories, and then the features extracted are put into deep training by neural network and later classified. The experiment results show that with this method the accuracy in classification is raised, which verifies that the permutation entropy attribute is effective for the identification of trajectory model. It can improve the recognition accuracy of trajectory model.
展开▼
