Abstract:
This thesis discusses the Global Positioning System (GPS), the digital maps and map matching algorithms, the image based lane detection algorithm and the ways of integrating these three techniques together to solve the lane detection problem. GPS has been developed and utilized for more than thirty years. Several industries have been established based on the GPS, and it also helps existing industries to upgrade their services towards more cost effective, fast, and reliable solutions. Digital maps are built to translate the GPS raw data to a human readable format. They contain both the location information and its reference to the earth. The basic digital maps that are used for car navigation contain the road names and their reference to the earth. Since the GPS raw data contains errors, some of the GPS raw locations fall out of the road on digital maps. The map matching algorithms are used to help map the GPS raw data to the occupied road. By correctly locating the occupied route, the device could also provide drivers with some useful information, such as the road geometry in front of the vehicle, reaching traffic lights, etc. Driver assistance systems (DAS) are a new research area considered to be the next revolution of the industry. The lane detection application is a kind of advanced DAS to solve the lane keeping problem. The image based lane detection application use the camera as the sole information source. Several algorithms have been developed to help understand the local environment. Since large volumes of calculations are involved in these algorithms, other information sources can be imported to reduce the computation expense. This thesis presents a solution to utilize and combine the information that comes from two sensors - the GPS and the camera, to provide a light weight solution for the lane detection applications. This thesis introduces some background knowledge of the GPS together with the digital maps and the map matching algorithm. The image based pattern detection techniques are also discussed. The detailed approaches and experiments will be presented at the end