This report presents work on the detection and tracking of people in digital images. The employed detection technique is based on image processing and classification techniques. The work uses an object detection process to detect object candidate locations and a classification method using a Self-Organising Map neural network to identify the pedestrian head positions in an image. The proposed tracking technique with the support of a novel prediction method is based on the association of Cellular Automata (CA) and a Backpropagation Neural Network (BPNN). The tracking employs the CA to capture the pedestrian's movement behaviour, which in turn is learned by the BPNN in order to the estimated location of the pedestrians movement without the need to use empirical data. The report outlines this method and describes how it detects and identifies the pedestrian head locations within an image. Details of how the proposed prediction technique is applied to support the tracking process are then provided. Assessments of each component of the system and on the system as a whole have been carried out. The results obtained have shown that the novel prediction technique described is able to provide an accurate forecast of the movement of a pedestrian through a video image sequence.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:434782 |
| Date | January 2006 |
| Creators | Suppitaksakul, Chatchai |
| Contributors | Sexton, Graham |
| Publisher | Northumbria University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://nrl.northumbria.ac.uk/488/ |
Page generated in 0.0021 seconds