Movement Detection and Tracking of Remotely Sensed Video Data Using Feature Point Matching

Authors: Fan Shi*, University of Texas At Dallas, Fang Qiu, University of Texas at Dallas
Topics: Remote Sensing
Keywords: Movement, Remotely Sensed Video data, Feature Points
Session Type: Paper
Day: 4/6/2019
Start / End Time: 1:10 PM / 2:50 PM
Room: Buchanan, Marriott, Mezzanine Level
Presentation File: No File Uploaded


An important application of space-born video data is to detect and track moving objects on the Earth surface. In our research, an innovative system aimed to detect and track moving objects is developed. This system involves two steps: first, image sequence differencing is utilized to detect locations where movement is taking place. Then, a feature point-based tracking method visits sampled frames to update locations of the moving objects. Compared to conventional methods, the proposed system demonstrates two advantages. The feature point-based method is capable of not only object tracking but also frame registration as its traditional purpose. Therefore, frames are registered before being tracked. As a result, the delineation of the moving object could be of high accuracy. Moreover, the developed system avoids the needs for enormous training data, so objects could be detected and tracked successfully in a relatively short period of time. We test the proposed system on a 5 seconds satellite video data, the results show high accuracy for both detection and tracking tasks.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login