CrackTheCoke #8

Encoding data in an image.
Decoding the image with encoded data.
Decoding an image with no encoded data.

Inspiration

Driven by the fascination of covert communication and the intricate details of digital forensics, we sought to explore novel methods of data concealment.

What it does

CrackTheImage is a tool-set that allows users to embed a second black and white image within an existing image, rendering it virtually invisible to the naked eye. The companion decoding tool then extracts the hidden information, revealing the concealed image/message. This innovative approach leverages the subtle nuances of pixel values. No metadata is used to hide the information.

How we built it

We capitalized on the human eye's limited ability to perceive minor pixel variations. By strategically manipulating these subtle differences, we encoded data directly within the image's pixel values.

Challenges we encountered

One of the primary challenges we faced was effectively compressing the hidden data to maximize capacity without compromising image quality or detectability. To overcome this, we developed a custom black-and-white generative AI auto-encoder, specifically tailored to efficiently encode and decode the hidden image.

Accomplishments we're proud of

We are immensely proud of successfully developing a functional and robust tool set that effectively conceals and reveals images within seemingly innocuous files.

What we learned

This project underscored the significant computational and financial resources required to train and implement advanced machine learning models like image auto-encoders.

What's next for CrackTheImage

Our future endeavors will focus on enhancing the resolution of the hidden image by exploring more sophisticated data compression techniques and refining our auto-encoder model.