Online classification of frequent behaviours based on multidimensional trajectories

Mining regular behaviors based on spatiotemporal trajectory multi‐dimensional features

“…[28] applied a similar idea to the spatiotemporal trajectories, but it only works well for sub‐trajectory clustering, which is not suitable to some application scenarios which should be clustering the whole trajectories. In recent years, some trajectory clustering algorithms based on multi‐feature fusion are proposed [29, 30]. Pan et al.…”

“…Ansariet et al [28] applied a similar idea to the spatiotemporal trajectories, but it only works well for sub-trajectory clustering, which is not suitable to some application scenarios which should be clustering the whole trajectories. In recent years, some trajectory clustering algorithms based on multi-feature fusion are proposed [29,30]. Pan et al [31] proposed a multi-dimensional trajectory clustering algorithm (MTCA) to complete unsupervised clustering of multi-dimensional trajectories.…”

Random forest‐based track initiation method

“…Track classification also has some applications in radar data processing. For example, Pan et al put forward an online classification algorithm based on multi‐dimensional trajectories to analyse the situation of targets in the radar early warning surveillance system [12]. Different from the above research on user behaviour pattern mining, in this paper, we are concerned with the movement characteristics of targets at different flight stages in the ATC radar data, which are important criteria for distinguishing real tracks produced by targets from false tracks.…”

Exploration on the Application of Computer Music in the Teaching of Public Music

“…To solve this problem, this paper refers to the repeated data cleaning technology [1] in data pre-processing technology, and regards each target track data as a record. In this way, when processing the multi-dimensional track data, the two tracks with similar behaviors (similar positions, speed, course and the same attributes, type) can be regarded as similar records, which transforms the complex problem of target track behavior law mining [2][3] into a relatively simple problem of similar duplicate record detection. In addition, in the traditional similar duplicate record detection [4][5], multi-source data is usually string data containing identifiers.…”