Image analysis – Applications – Vehicle or traffic control
Reexamination Certificate
1997-10-17
2001-04-03
Au, Amelia (Department: 2623)
Image analysis
Applications
Vehicle or traffic control
C382S279000, C382S283000, C701S028000
Reexamination Certificate
active
06212287
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to a method for identifying marking stripes demarcating road lanes and, in particular, a method for identifying marking stripes in a system for aiding the guidance of a vehicle.
BACKGROUND OF THE INVENTION
Methods capable of determining marking stripes demarcating road lanes are commonly used in systems for aiding vehicle guidance. Such methods, as described, for example, in “Parallel and Local Feature Extraction: A Real-Time Approach to Road Boundary Detection,” A. Broggi—
IEEE Transactions On Image Processing,
Vol. 4, No. 2, February 1995, pp. 217-223, typically comprise a phase of low-level filtering which extracts the relevant characteristics from the road image, followed by a phase of high-level analysis which determines a representation of the marking stripes. Rather simple processing techniques, which exhibit a low computational cost, capable of identifying the discontinuities present in the road image are generally used in the low-level filtering phase. For example, the image is subjected to a mathematical convolution operation with a suitable mask (or kernel) matrix, and the result is compared with a specified threshold value. The mask always exhibits very small dimensions, as in the case of the Prewitt operator, consisting of a matrix of three rows and three columns of ternary elements (−1, 0, +1). Such techniques, however, also pick out the outlines associated with various objects present in the image, for example, other vehicles, road signs, shadows, and the like (noise). They therefore require the use of complex high-level analyses for eliminating noise, with heavy demands for computational power.
Other techniques, such as the use of geometrical models described in “A Morphological Model-Driven Approach to Real-Time Road Boundary Detection for Vision-Based Automotive Systems,” A. Broggi, S. Berte—
Proc. Second IEEE Workshop on Applications of Computer Vision
, pp. 73-90, Dec. 5-7, 1994—IEEE Computer Society, transform the image on the basis of a predefined model so as then to determine the representation of the marking stripes of the road lanes. Such transformations have, however, to be applied over extensive portions of the image and require the use of rather complex basic operations. They therefore need large processing resources and cannot be carried out efficiently by means of standard digital systems.
What is needed is a method for identifying marking stripes demarcating road lanes which obviates the aforesaid drawbacks.
SUMMARY OF THE INVENTION AND ADVANTAGES
Accordingly, the present invention provides, in a system for aiding the guidance of a vehicle, a method for identifying marking stripes demarcating at least one road lane in a road image represented by a matrix of elements each indicative of the luminosity level of an elementary portion of the image. The method comprises the steps of setting a mask matrix, determining a convolved image matrix via a convolution operation between the matrix of the road image and the mask matrix so as to identify luminosity discontinuities present in the image, determining a binary image matrix in which each element has a first (0) or a second (1) binary value depending on the result of a comparison of a corresponding element of the convolved image matrix with a threshold value, and determining from the binary image matrix a representation of the marking stripes. The step of setting the mask matrix includes determining a value L indicative of the number of elements of the image matrix corresponding to the width of the marking stripes wherein each row of the mask matrix comprising L contiguous elements having a first value and wherein the other elements of the row having a second value of opposite sign to the first value.
The present invention also provides a method, in a system for aiding the guidance of a vehicle, for identifying marking stripes demarcating at least one road lane in a road image represented by a matrix of elements each indicative of the luminosity level of an elementary portion of the image. The method comprises the steps of determining a first and a second convolved image matrix via a convolution operation between the matrix of the road image and, respectively, a first and a second mask matrix so as to identify luminosity discontinuities present in the image, the elements of each row of the second mask matrix being equal in reverse order to the elements of a corresponding row of the first mask matrix. The method also includes the step of determining a first and a second binary image matrix in which each element has a first (0) or a second (1) binary value depending on the result of a comparison of a corresponding element, respectively, of the first and of the second convolved image matrix with, respectively, a first and a second threshold value. The method further includes the step of determining a single binary image matrix via a logic AND operation between the corresponding elements of the first and of the second binary image matrix and determining from the single binary image matrix a representation of the marking stripes.
The present invention advantageously reduces the effects of noise, produces a good signal
oise ratio, and limits the amount of information which has to be processed in the high-level analysis phase, without requiring the use of complex operations or the introduction of complicated analysis models.
Another advantage of the present invention is that it requires only low-level arithmetic operations such as sums and multiplications, can therefore be carried out using standard digital devices, and allows a high degree of parallelism in the various basic operations. Moreover, these low-level operations are suitable for being carried out very efficiently using dedicated analog architectures with a high level of parallelism.
Another advantage of the present invention is that it offers good performance even with low levels of accuracy of the information used, i.e., with a limited number of bits assigned to each element of the image and of the mask. Hence, the present invention is suitable for being carried out on low-precision analog processing systems which, by virtue of their speed, are able to perform the operations required very efficiently.
Another advantage of the present invention is that it offers high reliability under different ambient conditions and good real-time speed of response, using low-cost compact systems, thereby allowing mass-consumer applications.
REFERENCES:
patent: 5398292 (1995-03-01), Aoyama
patent: 5555312 (1996-09-01), Shima et al.
patent: 5706355 (1998-01-01), Raboisson et al.
patent: 5790403 (1998-08-01), Nakayama
patent: 5835614 (1998-11-01), Aoyama et al.
patent: 2674354 (1992-09-01), None
Vision in Man and Machine, Martin D. Levine; McGraw Hill Book Company, ISBN 0-07-037446-5, pp. 176-180, 1985.*
“Neural Networks on a Chip”World Congress on Neurel Networks, vol. 2, pp. 453-454, Jul. 1995.
“Parallel and Local Feature Extraction: A Real-Time Approach to Road Boundary Detection,”IEEE Transactions on Image Processing, vol. 4, No. 2, pp. 217-223, Feb. 1995.
“A Discriminating Feature Tracker for Vision-Based Autonomous Driving,”IEEE Transactions on Robotics and Automation, vol. 10, No. 6, pp. 769-775, Dec. 1994.
“Lane Recognition System for Guiding of Autonomous Vehicle,”Suzuki, et al., pp. 196-201 (no date available.
“Road and Relative Ego-State Recognition,”Behringer, et al., pp. 385-390 (no date available).
Fabbrizio Vito
Guerrieri Roberto
Kramer Alan
Olivieri Massimiliano
Au Amelia
Dastouri Mehrdad
Galanthay Theodore E.
Jenkens & Gilchrist P.C.
SGS--Thomson Microelectronics S.r.l.
LandOfFree
Method for identifying marking stripes of road lanes does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for identifying marking stripes of road lanes, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for identifying marking stripes of road lanes will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2512458