Range and Polarity-Based Binarization for Text Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing methods struggle to reliably extract information from digital images with complex backgrounds or overlapping foreground and background regions, particularly when elements of interest are represented according to different polarities or have similar intensity and color characteristics, leading to difficulties in distinguishing between foreground and background.
Innovation Solution
A computer program product and method that render digital images using multiple binarization thresholds to generate range-binarized images, identifying connected components and text regions within these images, allowing for the extraction of text information regardless of polarity or background complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional binarization techniques are used to simplify image processing, then processing speed and ease of operation are improved, but the ability to distinguish foreground from background in complex images deteriorates
Solution Approach 1:
The patent segments the image processing into multiple polarity-specific stages: first processing dark foreground on light background regions, then processing light foreground on dark background regions. This segmentation allows each stage to use optimized binarization parameters tailored to its specific polarity characteristics, thereby maintaining high reliability of text extraction while keeping each processing stage computationally efficient.
Solution Approach 2:
The patent applies local quality by using different binarization thresholds and processing parameters for different regions of the image based on their polarity characteristics. Dark foreground regions use one set of binarization parameters while light foreground regions use another set, allowing each local region to be processed with optimal parameters for its specific characteristics.
2Reliability
If multiple binarization thresholds are used to improve extraction accuracy, then the reliability of information extraction is improved, but the device complexity and processing time increase
Solution Approach 1:
The patent divides the complex multi-threshold problem into two simpler sequential tasks: first applying a binarization threshold optimized for dark text, then applying a second threshold optimized for light text. This segmentation reduces the overall complexity compared to attempting to process all polarities simultaneously with a single complex thresholding mechanism.
Solution Approach 2:
The patent performs preliminary binarization processing for each polarity type before final text extraction. By pre-processing the image with polarity-specific binarization thresholds, the system simplifies the subsequent text recognition task, thereby improving overall accuracy without requiring equally complex post-processing mechanisms.
3Ease of operation
If color channel information is converted to grayscale intensity to simplify processing, then ease of operation is improved, but the ability to distinguish foreground from background deteriorates
Solution Approach 1:
The patent applies local quality by preserving and utilizing color information specifically for polarity determination, then converting only to grayscale for the binarization stage. This selective use of color properties allows the system to maintain the ability to distinguish foreground from background while still benefiting from the simplicity of grayscale processing in the main extraction pipeline.
Solution Approach 2:
The patent uses color information as an intermediary to determine polarity characteristics before converting to grayscale for binarization. This intermediary step allows the system to leverage color properties for classification purposes while maintaining the computational efficiency of grayscale processing for the actual text extraction operations.
Data Source
AI summary
Computerized techniques for improved binarization and extraction of information from digital image data are disclosed in accordance with various embodiments. The inventive concepts include rendering a digital image using a plurality of binarization thresholds to generate a plurality of binarized digital images, wherein at least some of the binarized digital images are generated using one or more binarization thresholds that are determined based on a priori knowledge regarding an object depicted in the digital image; identifying one or more connected components within the plurality of binarized digital images; and identifying one or more text regions within the digital image based on some or all of the connected components. Systems and computer program products are also disclosed.


