Document Image Compression via Symbol Segmentation and Template Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for compressing and authenticating document images containing text are inefficient and lack effective means to ensure document integrity during the scanning and printing process, particularly in self-authenticating documents.
Innovation Solution
A method that segments text regions into symbol images, classifies them using template images and feature comparison, and stores resized template images with class indices and bounding box information for efficient compression, enabling self-authentication by comparing scanned images with original authentication data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If conventional compression methods are used for document images, then compression is achieved, but compression efficiency is insufficient and document integrity cannot be ensured
Solution Approach 1:
The document image is segmented into multiple symbol images based on text regions, allowing each symbol to be independently classified and compressed. This segmentation enables more efficient compression by identifying repeated symbols across different parts of the document, while also preserving document integrity through systematic tracking of symbol locations and classifications.
Solution Approach 2:
The patent creates template images representing unique symbol classes and stores references to these templates rather than storing each symbol image individually. When a symbol matches an existing template, only a reference is stored; when it's a new symbol, a new template is created. This copying approach significantly reduces data storage requirements while maintaining the ability to reconstruct the original document image for integrity verification.
2Measurement precision
If detailed symbol classification is performed to ensure accurate document authentication, then authentication accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent performs preliminary classification of symbol images into distinct classes based on their visual characteristics before the authentication process. By pre-organizing symbols into classified templates with associated metadata (bounding boxes, positions), the system establishes a reference framework in advance that speeds up the subsequent authentication process while maintaining high accuracy through comprehensive symbol differentiation.
Solution Approach 2:
The patent transforms symbol images into classified representations by changing their parameter form from raw pixel data to structured template references with metadata. This parameter transformation includes extracting key features, assigning class labels, and storing positional information, which reduces computational complexity during authentication while preserving the detailed information needed for accurate verification.
3Measurement precision
If all symbol images are stored in original resolution for accurate comparison, then authentication accuracy is maintained, but data storage requirements increase significantly
Solution Approach 1:
The patent merges multiple identical or similar symbol images into a single template representation. By grouping symbols that belong to the same visual class into shared templates, the system dramatically reduces storage requirements. Each template stores the essential visual information once, and multiple references to this template are used throughout the document, achieving both space efficiency and comparison accuracy.
Solution Approach 2:
The patent applies different quality levels to different parts of the document representation. Template images store high-quality visual information for accurate comparison, while the majority of symbols are represented by lightweight references containing only positional and classification metadata. This local quality differentiation maintains authentication accuracy where needed while minimizing overall data storage volume.
Data Source
AI summary
A method for compressing a bi-level document image containing text is disclosed. The document image is segmented into symbol images each representing a letter, numeral, etc. in the document. The symbol images are classified into a plurality of classes, each class being associated with a template image and a class index. Classification is done by comparing each symbol to be classified with template of existing classes, using a number of image features including zoning profiles, side profiles, topology statistics, and low-order image moments. These image features are compared using a tolerance based method to determine whether the symbol matches the template. After classification, certain classes that have few symbols classified into them may be merged with other classes. In addition, the template images of the classes are down-sampled, where the final sizes of the template images are dependent on the likelihood of confusion of the template with other templates.