Master Thesis MSTR-2012-07

BibliographySingh, Saurabh: Selection, implementation and evaluation of a real-time, high level programmable image processing system and a content enhancement algorithm for side view mirror displays.
University of Stuttgart, Faculty of Computer Science, Master Thesis No. 7 (2012).
94 pages, english.

Reduction in aerodynamic drag on the vehicle is important to improve the acceleration, handling and fuel mileage of the vehicle. One of the ways to do that is to remove the side view mirrors from a vehicle and use a display inside the vehicle for showing the realtime traffic situation around the driver. For this a real-time, high level programmable image processing system is required inside the vehicle. The thesis selects, implements and evaluates such a system for the research vehicle. After comparing the hardware devices available in the market and keeping in mind the suitability for a vehicle, a FPGA device is selected and an approach of going from MATLAB to FPGA is proposed, implemented and evaluated. Also, there is a need of content enhancement algorithm as there is a reduction in perceptibility of content viewed on the displays in ambient light conditions. Observations are conducted and a target function is prepared for selecting the algorithm. Based on the requirements listed in the target function, an approach LDIE (local dynamic image enhancement) is selected, implemented and tested. LDIE is then optimized further to increase the perceptibility under ambient light conditions and subjective tests are conducted with human observers to see the improvement in perceptibility caused by using optimized LDIE. Feedback given by the subjects is evaluated and it is clearly seen that optimized LDIE enhanced images are better perceived than the original images in ambient light conditions.

Department(s)University of Stuttgart, Institute of Parallel and Distributed High-Performance Systems, Parallel Systems
Superviser(s)Simon, Prof. Sven; Wroblewski, Dr. Marek; Bauer, Jan
Entry dateMarch 26, 2020
   Publ. Computer Science