Regarding the database schema and their relationship, please see below sections
The official release of USA traffic signs can be found at MUTCD. They offer sign resources in HTMLs and PDFs so it requires efforts to extract the sign images from their documents. We manually extracted 673 MUTCD code signs and provides those in the form of database with search interfaces.
Traffic sign recognition is a very challenging problem considering the various shapes, colors and texts used in the sign. Therefore, in order to recognize the traffic sign, saying specifically recognizing its MUTCD code, the approach toward recognition should be the composition of various sign-asscociated features. Extracted features and their interrelationship should be carefully balanced to compute the best traffic sign candidates. This section will investigate the proper sign categorization method by first analyzing the traffic sign statistics and later the order of computation and their relationships.
Our MUTCD code database above explained has a number of features of each traffic sign including its boundary shape, colors used in the legend and background, and texts with its number of lines if available. By using these features, the traffic signs can be categorized in detail to reduce the search space. This will also help us realize the necessary features to extract from the traffic images.
Image:mutcd_code_category_shape_back_legend_textline_text.png|Categorization using the sign shape, back color, legend color, text line and text Image:mutcd_code_category_shape_back_legend_textline.png|Categorization using the sign shape, back color, legend color, and text line Image:mutcd_code_category_shape_back_legend.png|Categorization using the sign shape, back color and legend color Image:mutcd_code_category_shape_back.png|Categorization using the sign shape and back color Image:mutcd_code_category_shape.png|Categorization using the sign shape
There is other resources on the Web, a persoanl website maintained by Richard C. Moeur, who is a voting member of the National Committee on Uniform Traffic Control Devices (NCUTCD), provides the complete list of images that he has been personally updating and maintaing for several years. We collected data from his web site and provide in multiple format (zipped files, SQL database, various formats, etc.) at Richard C. Moeur MUTCD database. However, we found that the MUTCD data set in his website is not complete with a lack of information for the research purposes and hence we developed our own databases as shown above.