Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Suz Hinton on Accessible images (AIMS)

Suz Hinton on Accessible images (AIMS)

In my latest experimentations in 'Guerilla Accessibility,' I stumbled upon this paper a mere two weeks after publication and was excited about the explorations in it. The ideas built upon an unlikely hero of the anti-surveillance field delighted me. There are wide implementation gaps that remain unaddressed in this paper, and I took liberty on these when implementing the algorithms. I'll be sharing both the triumphs and pitfalls of this proposed technique of improving the accessibility of online web content.

Papers_We_Love

February 15, 2018
Tweet

More Decks by Papers_We_Love

Other Decks in Technology

Transcript

  1. View Slide

  2. a model to build self‐describing
    images for assisting screen
    reader users

    View Slide

  3. Topics covered
    Web Accessibility
    Steganography techniques
    Browser Extensions

    View Slide

  4. Web Accessibility

    View Slide

  5. (A11y)

    View Slide

  6. Screen Readers

    View Slide

  7. Screen Readers
    JAWS
    NVDA
    VoiceOver
    Speakup
    TalkBack

    View Slide

  8. Demo

    View Slide


  9. View Slide


  10. View Slide

  11. Excerpt from Accessible images (AIMS): a model to build self‐describing images for assisting screen reader users

    View Slide

  12. Pros
    + Ease of implementation
    + Works with any image MIME type
    + Supported in browsers since
    HTML 2

    View Slide

  13. View Slide

  14. Cons
    – Requires coding knowledge
    – Decoupled from the actual image
    – Plenty* of devs have little to no a11y skills
    *presenter’s observation of peers from over 12 years of experience in front-end development field

    View Slide

  15. View Slide

  16. View Slide

  17. May 16, 2006 newscientist.com/article/dn9177-gamers-help-the-blind-get-the-picture

    View Slide

  18. View Slide

  19. Steganography

    View Slide

  20. Steganography
    the practice of concealing messages
    or information within other
    nonsecret text or data.

    View Slide

  21. View Slide

  22. View Slide

  23. View Slide

  24. View Slide

  25. there are many steganography
    techniques and approaches

    View Slide

  26. + concealing text in images
    + using LSB method

    View Slide

  27. 1 convert text to bits
    2 store bits in LSBs of pixel data

    View Slide

  28. close up photo of a lion facing to the left
    99
    1100011
    ASCII character code

    View Slide

  29. 20, 255, 102
    0010100
    LSB
    c
    99
    1100011
    ASCII character code
    save!
    1

    View Slide

  30. View Slide

  31. View Slide

  32. View Slide

  33. Assumptions
    ? knowing length of alt text
    ? method to compute bit position in image

    View Slide

  34. Knowing the length of alt text
    ? store length in consistent pixel location
    ? use special character to indicate end (EOF?)
    ? NLP (lol no)

    View Slide

  35. Computing bit position in image
    ? use a PRNG with seed
    ? decoder needs to know the seed

    View Slide

  36. Browser extension

    View Slide

  37. Browser extension
    third party script that can run on
    pages loaded by a user

    View Slide

  38. Browser extension
    a terrifying way to give third party
    code a same-origin policy waiver

    View Slide

  39. View Slide

  40. View Slide

  41. Implementation
    noopkat.github.io/a11y-alt-stego

    View Slide

  42. Pros
    + Description is portable with image
    + Less code
    + Less effort
    + Rad use of steganography

    View Slide

  43. Cons
    – Single language restriction
    – Lossless image format required
    – Compression can destroy / corrupt data
    – Most significant human labor not removed
    – Opt-in, standard adoption challenges

    View Slide

  44. thanks

    View Slide