Robust Watermarking of Video Streams

2,095 views

Published on

A proposed algorithm for robust data hiding in low resolution video streams

Published in: Technology, Art & Photos
0 Comments
7 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
2,095
On SlideShare
0
From Embeds
0
Number of Embeds
20
Actions
Shares
0
Downloads
0
Comments
0
Likes
7
Embeds 0
No embeds

No notes for slide

Robust Watermarking of Video Streams

  1. 1. Robust block selection for watermarking video streams Tamás Polyák, Gábor Fehér Budapest University of Technology and Economics Hungary
  2. 2. What is watermarking? <ul><li>Embed information into a digital signal </li></ul><ul><ul><li>Audio </li></ul></ul><ul><ul><li>Video </li></ul></ul><ul><ul><li>Text file </li></ul></ul><ul><ul><li>… </li></ul></ul><ul><li>Classification of watermarks </li></ul><ul><ul><li>Visible / invisible </li></ul></ul><ul><ul><li>Robust / fragile </li></ul></ul><ul><ul><li>… </li></ul></ul>
  3. 3. Use cases <ul><li>Multimedia security </li></ul><ul><ul><li>Fingerprinting </li></ul></ul><ul><ul><li>Copyright information </li></ul></ul><ul><ul><li>Copy protection </li></ul></ul><ul><ul><li>… </li></ul></ul><ul><li>Other areas </li></ul><ul><ul><li>Subtitles </li></ul></ul><ul><ul><li>Metadata </li></ul></ul><ul><ul><li>Secret communication </li></ul></ul><ul><ul><li>… </li></ul></ul>
  4. 4. Conflicting requirements Robustness Capacity Imperceptibility
  5. 5. Watermark life cycle Original ≈ Extracted
  6. 6. Our goals <ul><li>Watermarking low resolution video </li></ul><ul><ul><li>CIF (352 x 288) </li></ul></ul><ul><ul><li>QCIF (176 x 144) </li></ul></ul><ul><li>Requirements </li></ul><ul><ul><li>Robust against lossy compressions </li></ul></ul><ul><ul><li>Low quality loss </li></ul></ul><ul><ul><li>Limited complexity </li></ul></ul>
  7. 7. Base algorithms <ul><li>Ellinas (2007) </li></ul><ul><ul><li>Image watermarking in the wavelet domain </li></ul></ul><ul><ul><li>Embedding data around the edges </li></ul></ul><ul><ul><li>Too complex </li></ul></ul><ul><ul><li>Hides only 1 bit </li></ul></ul><ul><li>Cox (1998) </li></ul><ul><ul><li>Video watermarking in the spatial domain </li></ul></ul><ul><ul><li>Spread spectrum noise </li></ul></ul><ul><ul><li>Weak against lossy compression </li></ul></ul>
  8. 8. Proposed algorithm <ul><li>Data hiding in wavelet domain </li></ul><ul><ul><li>Fast 2-level DWT </li></ul></ul>
  9. 9. Proposed algorithm <ul><li>Edge detection in middle frequency components </li></ul><ul><ul><li>Sobel edge detector </li></ul></ul>
  10. 10. Steps of embedding process <ul><li>Apply DWT, calculate suitable area </li></ul><ul><li>Transform data into noise </li></ul><ul><ul><li>Extend data with a chip-rate (cr): </li></ul></ul><ul><ul><ul><li>1 0 1 0 </li></ul></ul></ul><ul><ul><ul><li>1 1 1 1 -1-1-1-1 1 1 1 1 -1-1-1-1 </li></ul></ul></ul><ul><ul><li>Multiplicate with key (N i ) </li></ul></ul><ul><ul><ul><li>1 1 1 1 -1-1-1-1 1 1 1 1 -1-1-1-1 Data </li></ul></ul></ul><ul><ul><ul><li>* 1 -1-1 1 -1 1 -1 1 1 -1 1 -1 -1 1 1 -1 Key </li></ul></ul></ul><ul><ul><ul><li>1 -1-1 1 1 -1 1 -1 1 -1 1 -1 1 -1-1 1 </li></ul></ul></ul><ul><ul><li>Apply strength (  ) </li></ul></ul><ul><ul><ul><li>5 -5-5 5 5 -5 5 -5 5 -5 5 -5 5 -5-5 5 </li></ul></ul></ul><ul><li>Add noise to the selected area </li></ul><ul><li>Apply IDWT </li></ul>
  11. 11. <ul><li>Apply DWT, calculate suitable area </li></ul><ul><li>Correlating key with values in area </li></ul><ul><li>Make decision: </li></ul><ul><ul><li>d j < -T : 0 </li></ul></ul><ul><ul><li>d j > T : 1 </li></ul></ul><ul><ul><li>else: x </li></ul></ul>Steps of detection process
  12. 12. Quality loss <ul><li>Original </li></ul><ul><li>Watermarked </li></ul>
  13. 13. Quality loss <ul><li>The watermarked frame and the watermark </li></ul>
  14. 14. Robustness test <ul><li>Hide 128 bits </li></ul><ul><li>Apply lossy compressions with various bitrates </li></ul><ul><ul><li>H.264/AVC </li></ul></ul><ul><ul><li>XviD </li></ul></ul><ul><li>Extract watermark, compare to original </li></ul><ul><ul><li>Correct bits: extracted bit = original bit </li></ul></ul><ul><ul><li>Erroneous bits: extracted bit ≠ original bit </li></ul></ul><ul><ul><li>Unrecognized bits: detector could not decide </li></ul></ul>
  15. 15. Video quality after H.264 /AVC <ul><li>Lower bitrate  lower quality </li></ul><ul><li>Lower bitrate  weaker watermark signal </li></ul>Uncompressed 1008 kbps 32,72 dB 720 kbps 31,31 dB 384 kbps 28,58 dB
  16. 16. Results of robustness test
  17. 17. Conclusion <ul><li>The algorithm is </li></ul><ul><ul><li>resistant against lossy compression  </li></ul></ul><ul><ul><li>imperceptible to the HVS  </li></ul></ul><ul><ul><li>not real-time  </li></ul></ul><ul><ul><ul><li>speed needs to be improved </li></ul></ul></ul>

×