Image Processing and Neural Network Techniques in Traffic Sign Recognition
| Autor: | Yu-Jhan Lin, 林郁展 |
|---|---|
| Rok vydání: | 2011 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 99 This article employs techniques in image processing and neural network to detect and recognize traffic signs with specific colors and shapes from road images acquired in a moving vehicle. The distribution thresholds in hue and saturation of traffic signs were used in distinguishing these signs into different colors, followed by block extraction and corner detection for shape interpretation. The central blocks of candidate sign image was then extracted to establish binary samples for use in back propagation neural network, Hopfield neural network, and normalized cross-correlation matching for performance comparison in traffic sign pattern recognition. Empirical results from standard patterns as training samples and 80 road images acquired in various conditions demonstrated that the approaches developed in this research are successful in shape and content recognition of traffic signs. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|