Ancient Book Recognition via Text Line Positioning and Reading Order
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image recognition methods struggle with accurately identifying ancient book images due to intricate typesetting and annotations, and current individual character detection techniques fail to consider positional relationships between characters, resulting in low recognition accuracy.
Innovation Solution
A method that aggregates individual character positions and content with text line positions and character reading direction by extracting backbone classification features, detecting character and text line positions, recognizing character content, and arranging information based on positional ratios to improve recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If individual character detection and recognition techniques are used, then the recognition process can be simplified, but the recognition accuracy deteriorates due to lack of consideration for positional relationships between characters
Solution Approach 1:
The patent combines individual character detection with text line position detection and reading order prediction into a unified recognition framework. By merging these previously separate processing stages, the system maintains relative simplicity while achieving high accuracy through integrated positional relationship analysis.
Solution Approach 2:
The patent introduces the dimension of spatial positioning and reading order by predicting text line positions and character sequences. This transforms the recognition problem from simple character identification to a multi-dimensional task that incorporates positional information, thereby resolving the accuracy-complexity contradiction.
2Productivity
If conventional image recognition methods are used, then the processing speed can be maintained, but the recognition accuracy deteriorates due to intricate typesetting and annotations in ancient books
Solution Approach 1:
The patent segments the ancient book image into individual character regions and text line positions, allowing the system to process complex layouts by breaking them down into manageable units. This segmentation approach maintains processing efficiency while improving accuracy for intricate typesetting.
Solution Approach 2:
The patent introduces text line position prediction and reading order prediction as intermediary steps between image input and final recognition output. These intermediaries help navigate the complexity of ancient book layouts without significantly increasing overall processing time.
3Device complexity
If character positions are detected without considering text line positions, then the detection process remains simple, but the final recognition result accuracy deteriorates
Solution Approach 1:
The patent performs preliminary detection of text line positions and reading orders before final character recognition. This preliminary action establishes the spatial framework needed for accurate recognition without adding significant complexity to the overall detection process.
Data Source
AI summary
Provided are a method and a device for recognizing an ancient book. The method comprises: extracting classification features of a target ancient book image based on a backbone network to obtain backbone classification features; detecting the backbone classification features and determining individual character positions and text line positions included in the target ancient book image; recognizing the individual character positions to obtain content information of individual characters, and predicting the text line positions to obtain a reading order of characters in the text line positions; and arranging, based on a ratio between the individual character positions and the text line positions, the content information of the individual characters following the reading order of the characters in the text line positions to obtain a recognition result of characters in the target ancient book image.