cuihaoleo / UnseenCodeDesktop

UnseenCodeDesktop

UnseenCode embedding & extraction demo code.

Note

Code in this repo was used to conduct experiments explained in our paper.

This is not an out-of-box demo. You will find that some parts missing to implement the whole embedding-extraction pipeline, and some parameters are hard-encoded in our code.

All programs are originally tested in Arch Linux and MSYS2.

UnseenCodeCamera2 is a full extractor demo. It should be used with the embedding code in this repo.

Compile

Dependency:

• OpenCV 4.x
• SDL2

For C/C++ code, make them:

$ make s1_gen_msg
$ make wtmk_disp
$ make t1_center_detect
$ make t2_parse

For Python scripts, please use python 3.x.

Pipeline

Some example input/output files can be found in the res/ folder.

Embedding:

• Step 1 (Optional): s1_gen_msg.c

  • Encode string into binary with BCH error-correction and CRC error-detection.
  • ./s1_gen_msg STRING_TO_ENCODE

• Step 2: s2_gen_mask.py

  • Generate the barcode mask representing the binary string given by user.
  • Run ./s2_gen_mask.py and paste the binary obtained from s1_gen_msg.
  • Output: bitmap_orig.png and mask.png.

• Step 3: s3_color4.py

  • Generate a pair of barcode images from the given cover image and barcode mask.
  • ./s3_color4.py lena.png mask.png
  • Output: out1.png, out2.png by default.

• Step 4: wtmk_disp.c

  • A simple player to display the barcode images.
  • ./wtmk_disp out1.png out2.png

Extraction:

• Step 1a: crop_img.py
  • Helper script to crop screen area by hand.
  • ./crop_img.py cap.jpg
  • Output: cap_correct.png
• Step 1b: t1_center_detect.cpp
  • Illustrates how auto screen crop works. Full code is integrated into Android demo. This program just output an image outlining the screen area.
  • ./t1_center_detect cap.jpg
  • Output is disp.png.
• Step 2: t2_parse4pp.py
  • Extracting UnseenCode.
  • ./t2_parse4pp.py --ref bitmap_orig.png cap_correct.png
  • Or: ./t2_parse4pp.py --block 17 cap_correct.png
• Step 3 (Optional): BCH & CRC decode.