PAPRICA system has been used to process images acquired live from a moving vehicle (and tested on MOB-LAB vehicle). In the first stage of the project, the aim of the processing is the generation of warnings to the driver through a control-panel installed near the driver.
The algorithms developed in our lab are related to the determination of the road and lane boundaries.
A few different apporaches were followed: starting from massively parallel feature extraction [IEEE Trans on Image Processing, Feb 1995], up to the removal of the perspective effect [ Proc. IEEE ICIP-95].
In the following you can find a few MPEG movies showing how the algorithm works on real image sequences.
Technical data of the MPEG movies:
- Size: 128 x 128
- 256 Gray levels
- Should be played back at a rate of about 5-6 frames/s
- The first MPEG contains 200 frames, while the second contains 1000 frames
The original sequence (183k MPEG) (749k MPEG)
The reorganized sequence (where the perspective effect has been
removed) (153k MPEG) (649k MPEG)
The binarized version of the previous sequence (252k MPEG) (
1085k MPEG)
The resulting sequence after the introduction of the
perspective effect (128k MPEG) (570k MPEG)
The final result superimposed onto the original sequence (221k
MPEG) (918k MPEG)
A sequence showing them in parallel
(1500k MPEG)
Please do not hesitate to contact the Page Manager (Alberto Broggi) for any additional information on PAPRICA project, and on the algorithms developed on PAPRICA system.