Detector and method for identifying a road lane boundary

• US 8,355,540 B2
• Filed: 10/18/2008
• Issued: 01/15/2013
• Est. Priority Date: 10/31/2007
• Status: Active Grant

• Current Assignee: ADC GmbH (TE Connectivity Ltd. (United Kingdom))
• Original Assignee: ADC GmbH (TE Connectivity Ltd. (United Kingdom))
• Inventors: Walter, Michael, Klein, Wladimir, Maeder, Dominik
• Primary Examiner(s): Tabatabai, Abolfazl
• Application Number:
  US12/740,889
• Publication Number:
  US 20110081049A1
• Time in Patent Office: 1,550 Days
• Field of Search: 382/103, 382/104, 382/106, 382/199, 382/256, 340/901, 340/903, 701/493, 701/65, 701/80, 701/300, 701/301, 348/118, 348/169
• US Class Current: 382/104
• CPC Class Codes: B60W 30/12 Lane keeping
• CPC Class Codes: G06V 10/50 by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
• CPC Class Codes: G06V 20/588 Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road

Abstract

A detector for identifying a road lane boundary using a digitalized optical image of the region in front of the vehicle. The detector includes a correlator configured to select the edges which are to be used for road lane estimation by searching for extreme values of convolution response, and to weight each convolution response by a weighting factor. A histogram analysis unit configured to group the extracted edges into pairs, and to determine a frequency distribution of the distances between two paired edges as a histogram, and to use said frequency distribution to determine the distances between two grouped edges forming a frequency peak or a frequency plateau in the histogram as nominal edge widths. A weighting factor determination unit configured to determine, for an edge, the weighting factor of the convolution response in the correlator in such a manner that the weight determined by the weighting factor for an edge of the paired edges is higher the smaller the deviation of the distance between the grouped edges from the nominal edge width.

Claims

1. A detector for identifying a road lane boundary using a digitalized optical image of the region in front of the vehicle, the detector comprising:

a correlator configured to select the edges which are to be used for road lane estimation by searching for extreme values of convolution responses, and to weight each convolution response by a weighting factor before searching for extreme values;

a histogram analysis unit configured to group the extracted edges into pairs, and to determine a frequency distribution of the distances between two paired edges as a histogram, and to use said frequency distribution to determine the distances between two grouped edges forming a frequency peak or a frequency plateau in the histogram as nominal edge widths; and

a weighting factor determination unit configured to determine, for an edge, the weighting factor of the convolution response in the correlator in such a manner that the weight determined by the weighting factor for an edge of the paired edges is higher the smaller the deviation of the distance between the grouped edges from the nominal edge width.

2. A detector according to claim 1, wherein the weighting factor determination unit is configured to determine the weighting factor G(k) of an edge k on the basis of the target function

3. A detector according to claim 1, wherein the histogram analysis unit is configured to take into account a frequency distribution in a previous iteration step when determining the frequency distribution of the distances between two paired edges.

4. A detector according to claim 3, wherein the histogram analysis unit is configured to determine the nominal edge width in such a manner that, in case the value of a nominal edge width which has been determined exceeds a nominal edge width which has been determined in a previous iteration step, this higher value is used as nominal edge width.

5. A detector according to claim 3, wherein the histogram analysis unit is configured to determine the nominal edge width in such a manner that, in case the value of a nominal edge width which has been determined is lower than a nominal edge width which has been determined in a previous iteration step, these two values determined for the nominal edge width are subjected to low-pass filtering.

6. A detector according to claim 1, wherein the histogram analysis unit is configured to calculate a plurality of histograms each of which is determined assuming different pitch angles while the optical image is recorded, wherein the histogram analysis unit is further configured to select that pitch angle as valid on which the histogram having the most pronounced frequency peaks is based, and wherein the detector further comprises a conversion unit which is configured to use the optical image and the selected pitch angle to determine the real dimension of the road lane boundary in front of the vehicle.

7. A detector according to claim 1, wherein the histogram analysis unit is configured to identify the edges of a multiple marking by comparing the distances and frequencies of paired edges at least two different locations where frequency peaks or frequency plateaus occur in the histogram.

8. A detector according to claim 7, wherein the histogram analysis unit is configured to identify a double marking if a frequency peak or a frequency plateau is detected in the histogram whose width is two to three times and whose number of frequencies is half that of another frequency peak or another frequency plateau.

9. A method for identifying a road lane boundary using a digitalized optical image of the region in front of the vehicle, in which edges that are present are marked, wherein the method comprises the following steps:

selecting edges which are to be used for road lane estimation by searching for extreme values of convolution responses, wherein each convolution response is weighted by a weighting factor before searching for extreme values;

creating and analysing a histogram in order to group the extracted edges into pairs, and to obtain a frequency distribution of the distances between two paired edges as a histogram, and to use said frequency distribution to determine the distances between two grouped edges forming a frequency peak or a frequency plateau in the histogram as nominal edge widths; and

determining the weighting factor in order to determine, for an edge, the weighting factor of the determination of the convolution response in the correlation step in such a manner that the weight determined by the weighting factor for an edge of the paired edges is higher the smaller the deviation of the distance between the grouped edges from the nominal edge width.

10. A non-transitory computer readable medium having a computer program stored thereon for carrying out a method according to claim 9 if the computer program is executed on a computer.