Text Recognition Device Slope Correction Magnification Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional text recognition systems face challenges in recognizing characters with tilted angles and varying sizes in images, leading to reduced recognition rates and complex model development and maintenance due to the use of multiple models.
Innovation Solution
A text recognition device and method that includes a character position recognition part, a correction processing part, and a text recognition part, which recognizes individual characters, corrects the position and orientation of text regions, and calibrates magnification to perform accurate text recognition using a single model, regardless of character position, angle, or size.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a text detection model and a text recognition model are used continuously, then text recognition can be performed, but development and maintenance complexity increases
Solution Approach 1:
The patent combines text detection and text recognition functions into a single integrated model. The model simultaneously performs both detection (identifying text regions) and recognition (recognizing individual characters) operations, eliminating the need for separate sequential models and reducing overall system complexity while maintaining recognition capability
2Reliability
If a text recognition model is used in environments with specific conditions (illumination, brightness, horizontality, size), then recognition performance is guaranteed, but recognition rate decreases when characters are tilted or small
Solution Approach 1:
The patent modifies the model's processing parameters by introducing a slope correction module that adjusts for tilted characters. The system calculates the slope angle of text regions and applies rotational transformations to normalize character orientations, enabling the model to maintain high recognition rates for tilted and small characters while preserving performance under standard conditions
Data Source
AI summary
An embodiment text recognition device includes a character position recognizer configured to recognize individual characters in an image, and the character position recognizer is also configured to recognize a position of each of the individual characters, a correction processor configured to set a main region, the correction processor further being configured to perform one or both of correcting a slope of the main region and magnification calibration for at least one character recognized by the character position recognition part, and a text recognizer configured to perform text recognition in the main region corrected by the correction processor.