Document Type Identification via String Density and Variance

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image processing technologies face challenges in accurately identifying the type of a document, particularly when document sizes overlap or lack distinct monochromatic print sides, leading to incorrect classification.

Innovation Solution

An image processing apparatus and method that utilize character string density and variance calculations to identify document types, incorporating modules for input image acquisition, document size determination, print side identification, and text direction analysis to differentiate between receipts, business cards, and photographs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If document classification is based on document size, then classification can be performed quickly, but accuracy deteriorates when document sizes overlap

Engineering Contradiction:

Improveclassification speedVSAvoiddocument type identification accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent changes the classification parameters from document size to character string density and character string variance. By calculating the density of character strings and the variance of character string sizes in the input image, the system achieves accurate document type identification without relying on document size, thereby resolving the contradiction between fast classification and accurate identification when sizes overlap.

Inventive Principle:

Principle #35Parameter changes

2Reliability

If document classification relies on monochromatic print side detection, then certain document types can be identified, but accuracy deteriorates when documents lack distinct monochromatic print sides

Engineering Contradiction:

Improvedocument type identification reliabilityVSAvoiddocument type identification accuracy

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The patent replaces the monochromatic print side detection parameter with character string density and character string variance parameters. This allows reliable document type identification for documents that do not have distinct monochromatic print sides, such as color photographs or documents with mixed-color text, thereby improving both reliability and precision simultaneously.

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If multiple classification methods are used to improve accuracy, then document type identification becomes more precise, but device complexity increases

Engineering Contradiction:

Improvedocument type identification accuracyVSAvoidprocessing system complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges the calculation of character string density and character string variance into a unified processing flow within the image processing apparatus. By combining these two parameters and using them together for document type identification, the system achieves high accuracy without significantly increasing device complexity, as both calculations can be performed using the same input image data.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS9305236B2Image processing apparatus, image processing method, and computer-readable, non-transitory medium

Publication Date: 2016.04.05 PFU LTD

AI summary

The image processing apparatus includes an input image acquisition module for acquiring an input image generated by reading a document, a character string information calculator for calculating character string density or character string variance in the input image, and a document type identification module for identifying a type of the document based on the character string density or the character string variance.