Selective Super-Resolution for Identification Medium Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing identification medium recognition systems face challenges in recognizing characters and numeric characters from identification media, such as license plates and ID cards, especially when imaged at long distances, due to geometric distortion and varying imaging conditions, leading to increased processing loads and erroneous recognition.
Innovation Solution
The system includes an identification medium recognition device with an image input, region detector, super-resolution processor, and selector that determines the need for super-resolution processing based on region size and distortion, allowing for efficient recognition of characters and numeric characters by selectively applying super-resolution processing and evaluating similarity between character strings and reference images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If super-resolution processing is performed on all identification medium images to recognize both long-distance and short-distance images, then recognition capability is improved, but processing load increases
Solution Approach 1:
The patent applies super-resolution processing selectively only to identification medium images that meet specific criteria (long-distance images with small region sizes), rather than processing all images. This local application of the processing technique improves recognition capability for problematic images while avoiding unnecessary processing load on images that don't require enhancement.
Solution Approach 2:
The system changes the processing approach based on image parameters such as region size and distance. By detecting whether an image is a long-distance image with a small region size, the system dynamically adjusts whether to apply super-resolution processing, thereby optimizing the balance between recognition capability and processing load.
2Measurement precision
If super-resolution processing is applied to long-distance images, then recognition accuracy is improved, but processing time increases
Solution Approach 1:
The system performs preliminary detection of image characteristics (region size, distance) before deciding whether to apply super-resolution processing. This preliminary action identifies which images require enhancement, allowing the system to skip unnecessary processing and reduce overall processing time while maintaining recognition accuracy for images that need it.
Solution Approach 2:
Instead of applying super-resolution processing to all images (excessive action), the system applies it only to the subset of images that meet the criteria for long-distance, small region size images (partial action). This partial application is sufficient to improve recognition accuracy where needed without incurring the full time cost of universal processing.
3Productivity
If simple similarity evaluation is used for character string recognition, then processing speed is maintained, but erroneous recognition increases
Solution Approach 1:
The patent segments the character string recognition process into multiple independent evaluations. Instead of relying on a single similarity evaluation, the system performs multiple similarity calculations between the input character string and reference character strings, then combines these evaluations to determine the final recognition result. This segmentation improves reliability while maintaining processing speed through efficient parallel evaluation.
Solution Approach 2:
The system incorporates feedback mechanisms in the recognition process by comparing the input character string against multiple reference patterns and using the accumulated similarity evaluations to refine the recognition decision. This feedback loop increases recognition accuracy by considering multiple evidence points rather than relying on a single evaluation.
Data Source
AI summary
Identification medium recognition device (3) includes image input (11) that acquires a captured image imaged by an imaging device, identification medium region detector (12) that detects a region of the identification medium from the captured image, identification medium recognizer (18) that recognizes the character and/or the numeric character included in the identification medium from the region of the identification medium, super-resolution processor (17) that selectively performs super-resolution processing of the region of the identification medium, region storage (14) that stores a preset region in the captured image, identification medium region determiner (13) that determines whether the region of the identification medium is positioned within the preset region, and super-resolution processing selector (16) that selects, in a case where it is determined that the region of the identification medium is positioned within the preset region, execution of super-resolution processing by the super-resolution processor.