An Automated Image-Based Approach for Tracking Pedestrian Movements from Top-View Video
| Autor: | Abdullah Zawawi Talib, Halimatul Saadiah Md. Yatim, Fazilah Haron |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
business.industry
Computer science Movement (music) ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION 020206 networking & telecommunications Tracking system 02 engineering and technology Pedestrian Tracking (particle physics) Image (mathematics) Data extraction 020204 information systems Video tracking 0202 electrical engineering electronic engineering information engineering Computer vision Artificial intelligence business Focus (optics) |
| Zdroj: | Advances in Visual Informatics ISBN: 9783319700090 IVIC |
| DOI: | 10.1007/978-3-319-70010-6_26 |
| Popis: | In order to gain better and more understanding of pedestrian safety video, better tracking of pedestrian movements is necessary. However, existing works on video tracking of pedestrian movements focus in some specific places or situations, extracted limited data from the video and in some cases, a lot of human interventions are required in handling the data extraction. This paper presents an automated image-based approach for tracking pedestrian movements that takes advantage of the top-view video. The proposed approach consists of several steps namely detection, tracking, image calibration and extracting characteristics of a pedestrian from a video. The methods used in these steps are adapted or enhanced from some of the existing work in this area. These steps also allow automated video monitoring and require less human efforts. Besides, it is also used to estimate the speed of a pedestrian. The results of the experiment for the proposed approach using five videos with different scenario are presented. The pedestrian movement was plotted accurately and the numbers of pedestrians detected in the video were recorded correctly whereas the speed of the pedestrians from the framework was very close to the actual speed. The proposed approach can be used to monitor pedestrians in a sparse environment such as at the entrance of a hall or building or along a corridor. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|