| Περίληψη: | Intelligent Transportations Systems (ITS) have become an integral part of every modern car since they provide advanced assistance in driving conditions. According to a study [1], 90% of rear-end collisions can be avoided if the driver is notified one second earlier. For this reason, in this thesis we focus on the problem of tracking the front vehicle through an on-board camera module and estimating its distance. We include a literature review on technologies that are currently being employed for distance estimation and we point out advantages and disadvantages. Additionally, we design a novel algorithm for vehicle tracking and distance estimation based on the licence plates that are, according to the law, in a visible place at the rear of each vehicle. The algorithm detects the plates based on their rectangular shape and colour variation and calculates the distance based on the standardized dimensions enforced by each country. Moreover, we use a low power ARM processor and a low cost 640x480 webcam to test its performance. Tracking the front vehicle works for distances of up to 9.6 meters from the camera, has an error of approximately 5% of the estimated distance and the ARM processor can process at least 20 frames per second in a video sequence. The results make it ideal for real time applications but also allow the possibility of increasing the image resolution in order to achieve longer distances during tracking and still being able to perform in real time.
|
|---|
