Katakana Character Conversion CAPTCHA System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing CAPTCHA techniques lower human legibility by screening characters, making it difficult for humans to visually recognize characters, which affects the effectiveness of the CAPTCHA in distinguishing between human and computer responses.
Innovation Solution
A response determination system that generates a character-converted question by switching, replacing, adding, or deleting characters in a word, allowing humans to easily read the CAPTCHA while making it difficult for computers to recognize, using a question/correct answer acquiring means, character-converted question generating means, response receiving means, and response determining means.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If characters are screened or distorted to make them computer-unreadable, then the CAPTCHA becomes harder for computers to recognize, but human legibility is lowered and characters become difficult to visually recognize
Solution Approach 1:
The patent changes the parameter of character representation from standard alphanumeric characters to katakana characters. This parameter change creates a fundamental difference in character structure and form, making it difficult for computer-based optical character recognition (OCR) systems trained on standard characters to recognize the katakana-based CAPTCHA, while maintaining full readability for human users who are familiar with katakana script.
2Object-affected harmful factors
If standard character distortion techniques are used, then computer recognition is reduced, but human readability deteriorates
Solution Approach 1:
Instead of taking the conventional approach of distorting standard characters to prevent computer recognition, the patent inverts the approach by using katakana characters as the base form. This inversion creates a CAPTCHA system where the default character set is already difficult for computer-based OCR to process, eliminating the need for additional distortion that would compromise human readability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention includes: acquiring a question including text information and a correct answer to the question; converting part of a character string or character in the question into a different character string or character, and generating a character-converted question (42) corresponding to the question(S21); outputting information of the character-converted question to an output destination(S22); receiving an input of a response (43) to the character-converted question(S27); and determining whether or not the received response matches with the correct answer (S28).