Authors: Khan Mortuza Bin Asad*, Graduate Student, Yihong Yuan, Associate Professor
Topics: Transportation Geography, Urban Geography
Keywords: Human Mobility, Taxi Trajectory, Big Geodata, Urban Dynamics
Session Type: Virtual Paper
Start / End Time: 8:00 AM / 9:15 AM
Room: Virtual 9
Presentation File: No File Uploaded
Global positioning system (GPS)-enabled vehicles (e.g., taxis) provide an efficient way to acquire a large amount of human mobility data, but it is also challenging to process and mine information from such datasets due to the massive data volume and the complexity of human movement patterns. In this study, we apply texture analysis measures to taxi trajectory data to identify human mobility patterns in different urban areas. Texture analysis is widely used for landscape classification in remote sensing. It quantitatively describes the relationships of Digital Number (DN) values of neighboring pixels, which is used for per-pixel classification. In this study, we divided the study area (Nanjing, China) into small cells. Like DN values of image pixels, each cell of the study area can represent different indicators, such as the number of pick-ups, drop-offs, speed, during different time periods for both weekdays and weekend days. The texture analysis measures we use in this study include homogeneity, contrast, dissimilarity, Gray Level Co-occurrence Matrix (GLCM) mean, GLCM variance, entropy, angular second moment, and GLCM correlation, etc. The results demonstrate that texture analysis can be an effective way of mining human mobility information from taxi trajectory data.