Ideas how to make picture undetectable by by PCR?

[EliteFollow]

Hello,
I'm currently seeking a solution for adding text to an image in a way that optical character recognition (OCR) cannot read the text. If you have any suggestions or recommendations for achieving this, I would greatly appreciate your input.

 

[ZenaRip]

Design a custom font that their system cannot identify.

 

[Talkst]

Design a custom font that their system cannot identify.

Many thanks! I'll give it a shot.

 

[WeTNT]

Simply manipulate the text in a manner resembling a captcha.

 

[Bugsbu]

Simply manipulate the text in a manner resembling a captcha.

View attachment 28703

Is this functioning correctly?

 

[RomanceLinkin]

I have a strong belief that there are certain methods I've encountered, but I can't recall them precisely. It seems that the OCR is unable to interpret them, even though they appear normal to the human eye.

 

[Prelordtr]

Though distortion may be effective, it won't withstand the capabilities of advanced AI algorithms such as Google Cloud Vision and Rossum. Therefore, you should carefully assess the potential adversaries attempting to analyze it and then determine the appropriate level of "safeguard" required.

 

[Supremsten]

Is this functioning correctly?

I believe the primary purpose of captchas is to remain indistinguishable by bots, ensuring that at least some users find it challenging to recognize them.

 

[DarthBradley]

Can the alphabet be typed in a manner similar to captcha, where the letters are deliberately distorted, just as in captcha?

 

[Kinector]

Can the alphabet be typed in a manner similar to captcha, where the letters are deliberately distorted, just as in captcha?

If you are referring to fonts, I believe it is possible for you to create your own, but they would remain fixed or unchangeable.

 

[TearStories]

Design a custom font that their system cannot identify.

Generally, OCR doesn't identify text by recognizing 'fonts.' Instead, it analyzes the features and shapes of characters. It looks for similarities in open areas, shapes of different texts, and letters in the scanned file for conversion. That's why it can also recognize handwritten documents that don't use specific fonts. This method of identifying text through its features is known as Intelligent Character Recognition.