Document Image Layer Segmentation for Highlighted Text Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques often misclassify highlighted text sections as images, leading to poor image quality due to high compression, resulting in illegible or blurred text in digital documents.
Innovation Solution
A method that detects highlighted portions in documents, segments them into separate text and image layers, and assigns a higher resolution to the highlighted text layer, using JBIG2 compression for text layers and JPEG compression for image layers, integrated into a Mixed Raster Content (MRC) file format to enhance image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If highlighted text sections are processed as images with high compression ratio, then file size is reduced, but image quality deteriorates causing illegible or blurred text
Solution Approach 1:
The patent segments the document into multiple layers: text layers and image layers. Highlighted text sections are specifically extracted into separate text layers, allowing them to be processed with higher resolution and lossless compression (JBIG2), while non-highlighted portions use standard compression. This segmentation enables differentiated quality control for different content types.
Solution Approach 2:
The patent applies local quality enhancement by assigning different resolution values to different layers. Text layers containing highlighted content are assigned higher resolution values than image layers, ensuring that only the critical highlighted text sections receive enhanced quality treatment while the rest of the document maintains standard compression ratios.
2Manufacturing precision
If higher resolution is assigned to highlighted text layers, then image quality is improved, but compression ratio decreases increasing file size
Solution Approach 1:
By segmenting highlighted text into separate text layers, the patent enables selective application of high resolution only where needed. This prevents the entire document from being processed at high resolution, thus minimizing the file size increase while ensuring quality improvement for critical sections.
Solution Approach 2:
The patent implements local quality enhancement by assigning higher resolution values specifically to text layers containing highlighted content, while maintaining lower resolution for image layers and non-highlighted text. This localized approach optimizes the balance between quality improvement and file size control.
3Productivity
If highlighted text sections are compressed using high compression algorithms, then compression ratio is improved, but text legibility deteriorates
Solution Approach 1:
The patent segments highlighted text into separate text layers that can be processed independently with appropriate compression algorithms. This allows text layers to use lossless JBIG2 compression which maintains legibility, while image layers use JPEG compression for better compression ratios on graphical content.
Solution Approach 2:
The patent changes the compression parameter by applying different compression algorithms to different layers. Text layers use JBIG2 compression which provides lossless or near-lossless compression suitable for text, while image layers use JPEG compression. This parameter differentiation ensures both legibility and compression efficiency.
Data Source
AI summary
Examples of systems and methods for enhancing image quality for documents with highlighted portions are described. A highlighted portion from a received input is detected. The highlighted portion is automatically segmented as a text layer and the remaining portion is further segmented as a separate text layer and an image layer. A resolution is assigned to the highlighted portion, wherein the resolution assigned to the highlighted portion is greater than the respective resolution for the text layer and the image layer, to improve quality of the highlighted portion. The text layer corresponding to the highlighted portion and the separate text layer and the image layer are integrated together to generate a scanned document in a Mixed Raster Content (MRC) file format.