Modern day cars are often equipped with a vision system that collects informa- tion about the car and its surroundings. Camera calibration is extremely impor- tant in order to maintain high accuracy in an automotive safety applications. The cameras are calibrated offline in the factory, however the mounting of the camera may change slowly over time. If the angles of the actual mounting of the cam- era are known compensation for the angles can be done in software. Therefore, online calibration is desirable. This master’s thesis describes how to dynamically calibrate the roll angle. Two different methods have been implemented and compared.The first detects verti- cal edges in the image, such as houses and lamp posts. The second one method detects license plates on other cars in front of the camera in order to calculate the roll angle. The two methods are evaluated and the results are discussed. The results of the methods are very varied, and the method that turned out to give the best results was the one that detects vertical edges.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:liu-94219 |
| Date | January 2013 |
| Creators | de Laval, Astrid |
| Publisher | Linköpings universitet, Datorseende, Linköpings universitet, Tekniska högskolan |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0019 seconds