| 摘要 |
The invention relates to a method for traffic sign recognition that analyzes and classifies the image data of a sensor (1) in an information processing unit, wherein at least one image detail (10) that sufficiently probably contains an object that is a traffic sign belonging to a particular class of traffic signs is determined on the basis of the results of an analysis in a first procedure step, and this image detail (10) is submitted, in a second procedure step, to a classificator (5) that recognizes a traffic sign belonging to the particular class on the basis of the selected image detail (10) by means of a learning-based algorithm; according to the invention a class-specific feature (11) is identified in the image detail (10) in the first procedure step, a modified image detail (12) with the class-specific feature (11) in the image center thereof is generated, wherein the image regions (13, 14, 15) created by shifting the class-specific feature (11) to the image center are padded with suitable pixels, and the modified image detail (12) is submitted to the classificator (5). |
