Method and device for traffic sign recognition

• US 8,643,721 B2
• Filed: 07/17/2008
• Issued: 02/04/2014
• Est. Priority Date: 07/24/2007
• Status: Active Grant

• Current Assignee: CARIAD SE (Volkswagen AG)
• Original Assignee: Hella KGaA Hueck & Company
• Inventors: Becker, Lars-Peter
• Primary Examiner(s): Avellino, Joseph
• Assistant Examiner(s): CONAWAY, JAMES E
• Application Number:
  US12/670,285
• Publication Number:
  US 20100283855A1
• Time in Patent Office: 2,028 Days
• Field of Search: 348/148, 382/181, 382/224
• US Class Current: 348/148
• CPC Class Codes: G06V 20/582 of traffic signs
• CPC Class Codes: G06V 30/242 Division of the character sequences into groups prior to recognition; Selection of dictionaries

Abstract

In a method and a device for traffic sign recognition, at least one significant feature for a traffic sign is determined which is standardized for a region. The region that corresponds to the determined feature is determined. At least one classification feature and/or at least one classification method is defined depending on the determined region for the recognition of the traffic sign and/or at least one further traffic sign. The recognition of the traffic sign and/or the at least one further traffic sign is performed by using the defined classification feature and/or the defined classification method.

Claims

1. A method for traffic sign recognition, comprising:

on the basis of available image data of at least one reproduction of a traffic sign, performing a first traffic sign recognition with at least one of a first classification feature specified for a first geographical region or a first classification method specified for the first geographical region to determine a first recognition quality of the first traffic sign recognition, wherein the first recognition quality is a probability indicating to what extent at least one possible traffic sign of the first geographical region is the traffic sign;

on the basis of the available image data of the at least one reproduction of the traffic sign, performing a second traffic sign recognition with at least one of a second classification feature specified for a second geographical region or a second classification method specified for the second geographical region to determine a second recognition quality of the second traffic sign recognition, wherein the second recognition quality is a probability indicating to what extent at least one possible traffic sign of the second geographical region is the traffic sign;

using the at least one of the first classification feature specified for the first geographical region or the first classification method specified for the first geographical region to identify the traffic sign as a specific traffic sign when the first recognition quality is greater than the second recognition quality; and

using the at least one of the second classification feature specified for the second geographical region or the second classification method specified for the second geographical region to identify the traffic sign as a specific traffic sign when the first recognition quality is less than the second recognition quality.

2. The method according to claim 1, characterized in that the second traffic sign recognition is performed responsive to the recognition quality determined with the aid of the first traffic sign recognition falling below a preset recognition quality limit value.

3. The method according to claim 1, characterized in that, on the basis of available image data of at least one image with a reproduction of at least a part of an area surrounding a vehicle, at least one traffic sign present in this area is determined.

4. The method according to claim 3, characterized in that during a processing of the available image data the traffic sign is detected as an object and is pre-classified as a traffic sign with the aid of a pre-classification method, wherein the object is tracked over several images.

5. The method according to claim 1, characterized in that a geographical region comprises at least one country, a federal state, a canton and another administrative unit having at least one traffic sign specifically specified for the administrative unit.

6. The method according to claim 1, characterized in that each time region-allocated data with the at least one classification feature specified for the respective geographical region or each time region-allocated program data for providing at least one classification method specified for the respective geographical region are held available by a data processing unit for the recognition of the traffic sign, wherein at least two different geographical regions are provided.

7. The method according to claim 6, characterized in that at least one of the geographical region-allocated data and the region-allocated program data are stored in a non-volatile storage area of the data processing unit.

8. The method according to claim 6, characterized in that a geographical region is preset, and in that, on the basis of the preset geographical region, at least one of a classification feature specified for the preset geographical region or a classification method specified for the preset geographical region are used for the recognition of the traffic sign, wherein for each of several possible traffic signs a probability is determined that the possible traffic sign is the specific traffic sign.

9. The method according to claim 8, characterized in that for each of several images of an image sequence with sequentially recorded images the probabilities for several possible traffic signs is determined.

10. The method according to claim 8, characterized in that the method steps for determining the geographical region are only implemented when the determined probabilities fall below a preset limit value or in periods in which no traffic signs are classified.

11. The method according to claim 1, characterized in that for at least two different geographical regions, data allocated to the respective geographical region are stored, and in that only those data are loaded for classification of the traffic sign which are allocated to the determined geographical region, wherein preferably for each geographical region a data record is stored which is used in the selection or presetting of this geographical region for the classification of traffic signs.

12. The method according to claim 1, characterized in that for each of at least two geographical regions the probability is determined whether, on the basis of the features determined, this geographical region is concerned, wherein preferably the geographical region with the highest determined probability is preset as the current geographical region.

13. The method according to claim 1, characterized in that a second traffic sign already classified with the first geographical region is classified again with the second geographical region, wherein it is checked whether given the classification with the second geographical region the second traffic sign is classified with a higher probability than given the classification with the first geographical region.

14. The method according to claim 13, characterized in that this repeated classification of the second traffic sign of the second geographical region takes place with the same available image data as the classification with the first geographical region, wherein the repeated classification of the second traffic sign is performed in a period in which no further traffic signs are to be classified, in particular when no traffic signs are detected in the recorded images.

15. The method according to claim 1, characterized in that given the recognition of a traffic sign specified for a geographical region this geographical region is preset for the classification of further traffic signs.

16. The method according to claim 15, characterized in that this traffic sign is an information board at border crossings, wherein the information boards for different geographical regions differ from one another in particular by a country code provided on the information board.

17. A device for traffic sign recognition, comprising:

an image recording system for generating image data of at least one image with a reproduction of at least a part of an area surrounding a vehicle; and

a processing unit

that processes the image data and determines a traffic sign that is present in the area covered by the reproduction,

that, on the basis of the image data of the traffic sign, performs a first traffic sign recognition with at least one of a first classification feature specified for a first geographical region or a first classification method specified for the first geographical region, wherein the processing unit determines a first recognition quality of the first traffic sign recognition, the first recognition quality being a probability indicating to what extent at least one possible traffic sign of the first geographical region is the traffic sign,

that, on the basis of the image data of the traffic sign, performs a second traffic sign recognition with at least one of a second classification feature specified for a second geographical region or a second classification method specified for the second geographical region, wherein the processing unit determines a second recognition quality of the second traffic sign recognition, wherein the second recognition quality is a probability indicating to what extent at least one possible traffic sign of the second geographical region is the traffic sign,

that uses the at least one of the first classification feature specified for the first geographical region or the first classification method specified for the first geographical region to identify the traffic sign as a specific traffic sign when the first recognition quality is greater than the second recognition quality, and

that uses the at least one of the second classification feature specified for the second geographical region or the second classification method specified for the second geographical region to identify the traffic sign as a specific traffic sign when the first recognition quality is less than the second recognition quality.