Image Processing Method for a Driver Assistance System of a Motor Vehicle for Detecting and Classifying at Least one Portion of at Least one Predefined Image Element

• US 20120162429A1
• Filed: 06/21/2010
• Published: 06/28/2012
• Est. Priority Date: 06/29/2009
• Status: Active Grant

• Current Assignee: Robert Bosch GmbH
• Original Assignee: Robert Bosch GmbH
• Inventors: Wuerz-Wessel, Alexander
• Granted Patent: US 9,030,558 B2
• Time in Patent Office: Days
• US Class Current: 348/148
• CPC Class Codes: G06F 18/211 Selection of the most significant subset of features
• CPC Class Codes: G06V 10/771 Feature selection, e.g. selecting representative features from a multi-dimensional feature space
• CPC Class Codes: G06V 20/582 of traffic signs

Abstract

In an image processing method for a driver assistance system for detecting and classifying a portion of a predefined image element having a road sign in a digital image captured by an image sensor of the driver assistance system, first scale-invariant image features and their relative geometric arrangement with respect to one another are computed based on at least one image region of the digital image to be searched, after which a classifier compares the first scale-invariant image features and their relative geometric arrangement with respect to one another to stored and/or learned second scale-invariant image features and their relative geometric arrangement with respect to one another which are computed based on the at least one predefined image element.

Claims

1-13. -13. (canceled)

14. An image processing method for a driver assistance system of a motor vehicle for detecting and classifying at least one portion of a predefined image element having a road sign in a selected digital image captured by an image sensor of the driver assistance system, comprising:

computing first scale-invariant image features and relative geometric arrangement of the first scale-invariant image features with respect to one another based on at least one image region of the digital image;

comparing, by a classifier unit of the driver assistance system, the first scale-invariant image features and the relative geometric arrangement of the first scale-invariant image features with respect to one another to second scale-invariant image features and relative geometric arrangement of the second scale-invariant image features with respect to one another which are computed based on the predefined image element; and

determining, by the classifier, the at least one portion of the predefined image element of the digital image to be detected and classified if the comparing produces a match.

15. The image processing method as recited in claim 14, wherein the second scale-invariant image features and the relative geometric arrangement of the second scale-invariant image features with respect to one another are (i) computed in advance in a training step using different designs of the predefined image element, and (ii) at least one of stored and learned by the classifier.

16. The image processing method as recited in claim 14, wherein only characteristic second scale-invariant image features and relative geometric arrangement of the characteristic second scale-invariant image features with respect to one another which are identified in the training step are taken into account.

17. The image processing method as recited in claim 14, wherein the first scale-invariant image features and the second scale-invariant image features are computed with the aid of one of scale-invariant feature transform (SIFT) method or speeded-up robust features (SURF) method.

18. The image processing method as recited in claim 14, wherein the at least one image region is identified in advance by searching for selected geometric shapes in the digital image.

19. The image processing method as recited in claim 14, wherein a measure for the correctness of the detection and classification is determined.

20. The image processing method as recited in claim 19, wherein the comparison result is discarded when the measure for the correctness of the detection and classification is below a predefined threshold value.

21. The image processing method as recited in claim 19, wherein the number of pixels of the second scale-invariant image features associated with the pixels of the first scale-invariant image features is used as the measure for the correctness of the detection and classification.

22. The image processing method as recited in claim 14, wherein an intrinsic motion which is ascertained based on a sequence of digital images including the selected digital image is also taken into account in detecting and classifying the at least one portion of the predefined image element.

23. The image processing method as recited in claim 15, wherein the predefined image element includes at least one text element of a road sign.

24. A non-transitory computer-readable data storage medium storing a computer program having program codes which, when executed on a processor of an image processing device of a driver assistance system of a motor vehicle, performs an image processing method for detecting and classifying at least one portion of a predefined image element having a road sign in a selected digital image captured by an image sensor of the driver assistance system, the method comprising:

computing first scale-invariant image features and relative geometric arrangement of the first scale-invariant image features with respect to one another based on at least one image region of the digital image;

comparing the first scale-invariant image features and the relative geometric arrangement of the first scale-invariant image features with respect to one another to second scale-invariant image features and relative geometric arrangement of the second scale-invariant image features with respect to one another which are computed based on the predefined image element; and

determining the at least one portion of the predefined image element of the digital image to be detected and classified if the comparing produces a match.

25. A driver assistance system of a motor vehicle, comprising:

at least one image sensor capturing a digital image; and

an image processing device connected to the image sensor, wherein the image processing device is configured to detect and classify at least one portion of a predefined image element having a road sign in the digital image captured by the image sensor, the image processing device including;

means for computing first scale-invariant image features and relative geometric arrangement of the first scale-invariant image features with respect to one another based on at least one image region of the digital image; and

means for comparing, by a classifier unit of the driver assistance system, the first scale-invariant image features and the relative geometric arrangement of the first scale-invariant image features with respect to one another to second scale-invariant image features and relative geometric arrangement of the second scale-invariant image features with respect to one another which are computed based on the predefined image element, wherein the at least one portion of the predefined image element of the digital image is determined to be detected and classified if the comparing produces a match.