Traffic Sign Recognition Combining Speed Limit and Conditional Data

The patent segments the recognition process into distinct phases: capturing multiple images at different positions relative to the traffic sign (first image at first position, second image at second position), recognizing traffic signs in each image separately, and then combining the recognition results. This segmentation allows the system to distinguish between main traffic signs and conditional traffic signs by analyzing their spatial relationships and persistence across multiple images, thereby preventing erroneous recognition of one sign as multiple signs.

The patent performs preliminary actions by capturing a first image before the vehicle reaches the traffic sign position, then capturing a second image after passing the sign. This preliminary multi-stage image capture enables the system to pre-process and compare traffic sign information across different positions, allowing it to identify and retain only valid traffic signs while filtering out false recognitions before final output.

The patent implements feedback by comparing traffic sign recognition results across multiple images captured at different positions. The control device determines whether a recognized traffic sign is valid by checking if it appears consistently across images and matches expected patterns. This feedback mechanism allows the system to correct erroneous recognitions and distinguish between main traffic signs and conditional traffic signs, ensuring accurate speed limit information output.

The patent adds a spatial dimension to traffic sign recognition by capturing and analyzing images at multiple positions (first position before the sign, second position after the sign). This dimensional approach allows the system to differentiate between actual traffic signs and false recognitions based on their spatial persistence and positional relationships, thereby improving reliability without sacrificing information capture.