Text Region Distortion Compensation via Affine Transform
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Text recognition systems face deterioration due to geometric distortion in captured images, which affects the accuracy of extracting and recognizing text regions from images taken with image capturing devices.
Innovation Solution
A method that extracts the text region, estimates the forms of the upper, lower, left, and right sides of the text region, and applies an affine transform to transform the text region into a rectangular diagram, compensating for distortion and improving recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If text recognition is performed on geometrically distorted images, then text can be recognized from captured images, but recognition accuracy deteriorates due to distortion
Solution Approach 1:
The patent applies preliminary action by estimating the distortion form of the text region before recognition and transforming it into a corrected region. The system estimates the upper end form, lower end form, left side form, and right side form of the text region, then uses these estimates to construct a corrected text region that compensates for geometric distortion before actual recognition occurs.
Solution Approach 2:
The patent changes geometric parameters of the text region by transforming it from a distorted quadrilateral shape to a corrected shape with estimated parallel sides. The transformation adjusts the coordinates and shapes of the text region boundaries based on estimated distortion parameters, converting the distorted text region into a form that maintains recognition accuracy.
2Ease of manufacture
If the text region is extracted as-is from the distorted image, then extraction is simple, but the distorted text region affects recognition rate
Solution Approach 1:
The system performs preliminary estimation of the text region's distortion characteristics before final recognition. By estimating the forms of the upper end, lower end, left side, and right side of the text region in advance, the system can construct a corrected text region that maintains reliability without significantly complicating the extraction process.
Solution Approach 2:
The patent introduces an intermediary corrected text region that serves as a bridge between the distorted extracted text region and the final recognition process. This intermediary region is constructed by estimating distortion forms and transforming the text region, thereby mediating between simple extraction and accurate recognition.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This method effectively compensates for geometric distortion in text recognition, ensuring accurate extraction and recognition of text regions even from distorted images, thereby enhancing the overall text recognition rate.
Implementation Method 1
transforming the text region constituting the estimated diagram into a rectangular diagram using an affine transform
Data Source
AI summary
A method of compensating for distortion in text recognition is provided, which includes extracting a text region from an image; estimating the form of an upper end of the extracted text region; estimating the form of a lower end of the extracted text region; estimating the form of left and right sides of the extracted text region; estimating a diagram constituted in the form of the estimated upper end, lower end, left and right sides, and including a minimum area of the text region; and transforming the text region constituting the estimated diagram into a rectangular diagram using an affine transform.