Text Recognition via Inpainting Overlapping Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing text recognition methods fail to accurately separate and recognize overlapping text in images due to errors in printing coordinates, leading to inaccurate results.
Innovation Solution
A server-based method that separates a first text region from an image, extracts undamaged and damaged portions of the text, and reconstructs the text by inpainting the damaged areas using a text separation network, including color clustering and neural networks to handle overlapping text.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If general text recognition methods are used, then the recognition process is simple, but the text recognition result is inaccurate due to overlapping text
Solution Approach 1:
The patent segments the overlapping text region into multiple separate text instances by detecting text boundaries and separating overlapping character groups. This allows each text to be recognized independently, improving accuracy while managing complexity through systematic division of the recognition task.
Solution Approach 2:
The patent introduces an intermediary processing step that generates a mask or separation layer between overlapping texts. This intermediary structure helps distinguish overlapping characters and enables accurate recognition by preventing direct interference between overlapping text regions.
2Reliability
If text separation algorithms are used, then overlapping text can be recognized, but the processing complexity increases
Solution Approach 1:
The patent employs dynamic adjustment of separation parameters based on the detected overlap characteristics. The system adapts its processing strategy according to the degree and pattern of overlap, optimizing reliability for different text configurations while avoiding unnecessary complexity for simple cases.
Solution Approach 2:
The patent changes key parameters such as text boundary thresholds, separation distances, and recognition sensitivity based on the analyzed overlap situation. This parameter adaptation improves recognition reliability for overlapping texts while maintaining efficient processing through conditional parameter adjustment rather than fixed complex algorithms.
Data Source
AI summary
A method performed by a server, may include: obtaining an image including a first text and a second text overlapping the first text; separating a first text region corresponding to the first text from the image; extracting pixels corresponding to the first text from the first text region to obtain an undamaged portion and a damaged portion of the first text; and reconstructing the first text by inpainting the damaged portion of the first text in which the first text overlaps the second text in the image.