Image analysis – Applications – Vehicle or traffic control
Reexamination Certificate
2000-08-25
2004-10-19
Au, Amelia M. (Department: 2623)
Image analysis
Applications
Vehicle or traffic control
C382S106000, C382S293000, C701S028000, C348S118000
Reexamination Certificate
active
06807287
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to a method of and apparatus for road profile prediction suitable for use in an autonomous cruise control system for a road vehicle.
BACKGROUND OF THE INVENTION
Previous proposals for road profile prediction have involved the digitisation and storage of a video image of the road ahead of a vehicle, and the processing of the video image data to identify lane markings in the road. One proposed method, searches the image for lines, and attempts to establish when a pair of the lines appear to represent the parallel lane markings on the road. The process proposed involves successive stages of finding straight lines in the image and testing these in pairs to ascertain whether the lines of a pair converge to a vanishing point. Lines that fit the desired pattern are then analysed to determine the position of the vehicle between the lane markings (“offset”), its orientation relative to the lane markings (“heading angle”), and the curvature of the lane markings (“lane curvature”). This method is unsatisfactory, however, as the position of the “horizon” in the image is not known and can be a source of considerable error in the calculations involved. Offset, heading and curvature calculations all contain errors as a result.
EP-A-0697641 discloses a lane image processing system for a vehicle to recognise the lane in which the vehicle is travelling. However, the system produces lane information in screen co-ordinates and does not operate using real co-ordinates. The system is also very computer intensive.
In accordance with the present invention, there is provided apparatus for road profile estimation ahead of a vehicle, comprising digitisation and storage means operable to digitise and store video image data containing feature points relating to the road ahead of the vehicle, and processor means operable to process the video image data to identify which feature points represent lane markings in the road, the processor means including estimator means operable to i) estimate the real co-ordinates of feature points identified as representing a first lane marking simultaneously with the real co-ordinates of feature points identified as representing a second, paired, lane marking, the paired lane markings constituting a lane marking model, and ii) find the best fit of the lane marking model to the identified feature points by estimating the solution to an expression representing the real geometrical constraints of the road, including left and right lane marking offsets, lane heading angle, lane curvature and horizon error, using a single recursive estimation process, said estimator means being operable to take into account when processing data relating to one of said feature points from a video frame image, the result of the estimation process for said one feature point in respect of at least one previous video frame image.
SUMMARY OF THE INVENTION
In accordance with a further aspect of the invention, there is provided a method for estimating the road profile ahead of a vehicle, comprising the steps of digitising and storing video image data containing feature points relating to the road ahead of the vehicle, and processing the video image data to identify which feature points represent lane markings in the road, the processing step including i) estimating the real co-ordinates of feature points identified as representing a first lane marking simultaneously with the real co-ordinates of feature points identified as representing a second, paired, lane marking, the paired lane markings constituting a lane marking model, and ii) finding the best fit of the lane marking model to the identified feature points by estimating the solution to an expression representing the real geometric constraints of the road, including left and right lane marking offsets, lane heading angle, lane curvature and horizon error, using a single recursive estimation process, taking into account when processing data relating to one of said feature points from a video frame image, the result of the estimation process for said one feature point in respect of at least one previous video frame image.
REFERENCES:
patent: 5359666 (1994-10-01), Nakayama et al.
patent: 5790403 (1998-08-01), Nakayama
patent: 5922036 (1999-07-01), Yasui et al.
patent: 5991427 (1999-11-01), Kakinami et al.
patent: 6091833 (2000-07-01), Yasui et al.
patent: 6285778 (2001-09-01), Nakajima et al.
patent: 0 527 665 (1993-02-01), None
patent: 0 697 641 (1996-06-01), None
Kan W Y et al. “Model-Based Vehicle Tracking From Image Sequences With An Application To Road Surveillance.”Optical Engineering. vol. 35, No. 6, Jun. 1, 1996, pp. 1732-1729.
Au Amelia M.
Kim Charles
Lucas Industries Limited
Ohlandt Greeley Ruggiero & Perle
LandOfFree
Road profile prediction does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Road profile prediction, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Road profile prediction will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3304362