Text Region Detection Using Oriented Gradient Summation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional text detection methods often erroneously recognize potential text regions without actual text, leading to inefficient resource consumption and unrecognizable results, especially in images with complex non-text backgrounds or patterns.
Innovation Solution
A method involving the generation of oriented gradient images, calculation of sums of oriented gradients, and classification using a series of classifiers to determine whether a candidate text region contains text, thereby reducing erroneous detections and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional text detection methods are used, then text recognition can be performed, but erroneous detections increase particularly when images include complex non-text backgrounds or patterns
Solution Approach 1:
The verification unit divides the candidate text region into multiple detection window images and processes each separately through oriented gradient analysis. This segmentation allows localized verification of text characteristics, reducing false positives from complex backgrounds while maintaining detection accuracy for genuine text regions.
Solution Approach 2:
The patent introduces oriented gradient images as an intermediary representation between the original image and text detection. By transforming the image into oriented gradient space and calculating sums of oriented gradients, the system creates a intermediate feature space that highlights text characteristics while suppressing complex background patterns, thereby improving detection reliability.
2Productivity
If text recognition methods are applied to falsely detected text regions, then comprehensive text recognition is attempted, but computing resources are consumed inefficiently
Solution Approach 1:
The verification unit performs preliminary verification using oriented gradient analysis before committing to full text recognition. By pre-screening candidate regions with the sum of oriented gradients calculation and threshold comparison, the system eliminates false positives early, preventing wasteful consumption of computing resources on non-text regions while maintaining high throughput for genuine text regions.
3Measurement precision
If multiple detection window images are analyzed with oriented gradient calculations, then detection accuracy improves, but processing complexity increases
Solution Approach 1:
The patent transforms the verification problem by changing the parameter space from raw pixel analysis to oriented gradient domain. By calculating sums of oriented gradients for different orientations and comparing against thresholds, the system achieves high verification accuracy through parameter transformation rather than complex structural analysis, simplifying the overall verification process while maintaining precision.
Data Source
AI summary
A method for detecting a text region in an image is disclosed. The method includes detecting a candidate text region from an input image. A set of oriented gradient images is generated from the candidate text region, and one or more detection window images of the candidate text region are captured. A sum of oriented gradients is then calculated for a region in one of the oriented gradient images. It is classified whether each detection window image contains text by comparing the associated sum of oriented gradients and a threshold. Based on the classifications of the detection window images, it is determined whether the candidate text region is a true text region.