![Page 1: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/1.jpg)
Blind Quality Assessment System
for Multimedia Communications Using Tracing Watermarking
P. Campisi, M. Carli, G. Giunta and A. Neri
IEEE Transactions on Signal Processing, VOL. 51, NO. 4, April 2003
![Page 2: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/2.jpg)
Outline
• Introduction• Tracing watermarking procedure• QoS system embedding• QoS evaluation• Experimental results
![Page 3: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/3.jpg)
Motivation
• 3G mobile communication systems are expected to offer multimedia services with negotiation of end-to-end quality of service (QoS)– Effective and simple billing systems
related to the QoS provided– The quality assessment systems that
will not increase the bit rate of transmission
![Page 4: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/4.jpg)
Digital Watermarking
Embedding Procedure
Data
Watermark
Key
Watermarked Data
DetectionProcedure
SuspectData
Key
DetectedWatermark
• Conventional usages – Copyright protection– Fingerprinting– Broadcast monitoring– Multimedia indexing– Content-based retrieval– Authentication
• Altered or not• Localization information
• Requirements – Perceptual invisibility– Robustness– Capacity– Fragility
![Page 5: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/5.jpg)
MPEG-2/4 Video Standards
• MPEG-2– Layered syntax
• Video sequences -> GOPs -> Frames -> Slices -> Macroblocks -> Luma/chroma blocks
– DCT + Quantization + VLC– Motion compensation
• MPEG-4– Object-based representation– Scenes are composed of video objects, each
described by motion, texture and shape• Frame types in the MPEG-2/4 standards
– I frame, P frame and B frame
![Page 6: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/6.jpg)
Watermarking for Evaluating QoS?
• Fragile watermarking is used to evaluate the QoS in multimedia mobile communication– Why?
• The mark undergoes the same alterations suffered by the data over a communication Channel
– How?• The watermark is estimated and compared
with the original at the receiving side• The watermark degradation is used to trace
the overall alterations endured by the data
![Page 7: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/7.jpg)
Tracing Watermarking
• For each frame, the watermark is the same, while the PN matrices are different for each frame to improve the perceptual fidelity.
• Embedding is performed in the DCT domain – the middle-band frequencies of the whole image.
Video Watermark Embedding
Mod-2 Adder
TracingWatermark
PN-Sequence
Coder
Channel
DecoderMatched Filter
PN-Sequence
+QoS Assessment
![Page 8: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/8.jpg)
QoS System Embedding (1)
• Video sequence: – {fi[n1,n2], i=1,2,…,M}
• DCT transformed video frame: – {Fi[k1,k2], i=1,2,…,M}
• Middle-high frequency: – S of Fi[k1,k2]
• Watermark: – {w[k1, k2]}
• PN matrices: – {pi[k1, k2], i=1,2,…,M}
• Spread version of watermark: – wi
(s)[k1,k2]=w[k1, k2]*pi[k1,k2],i=1,2,…,M
![Page 9: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/9.jpg)
QoS System Embedding (2)
• Embedding:– if (k1,k2) belongs to S,
• Fi(w)[k1,k2]=Fi[k1,k2]+ awi
(s)[k1,k2], – else
• Fi(w)[k1,k2]=Fi[k1,k2]
– where • a is a scaling factor and is chosen as 0.04
• Obtaining watermarked frame:– fi
(w)[n1,n2]=IDCT{Fi(w)[k1,k2]}
![Page 10: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/10.jpg)
QoS Evaluation • Received video sequence:
– fi‘(w)[n1,n2], i=1,2,…,M
• DCT transformed receiving frame– Fi
‘(w)[k1,k2]=DCT{fi‘(w)[n1,n2]}
• Watermark Estimation– wi
‘(s)[k1,k2]= Fi‘(w)[k1,k2]*w[k1,k2]
• Disperading– wi‘[k1,k2]= wi
‘(s)[k1,k2]*pi[k1,k2]• Averaging
– w’[k1,k2]=Sum(wi‘[k1,k2])/M• MSE calculation
![Page 11: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/11.jpg)
Experiments• QCIF video sequences are used to
match the limited dimensions of mobile terminal’s display
• Marked video is transmitted over a noisy channel simulated by a Poission error generator
• MPEG-2/4 videos are tested at different bit-rates– The MSE of the extracted watermark
increases when the BER increases and the bit rate decreases
![Page 12: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/12.jpg)
Experimental Results (1/6)
Akiyo
![Page 13: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/13.jpg)
Experimental Results (2/6)
Suzie
![Page 14: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/14.jpg)
Experimental Results (3/6)
Akiyo at 600 kb/s
![Page 15: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/15.jpg)
Experimental Results (4/6)
Akiyo at 200 kb/s
![Page 16: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/16.jpg)
Experimental Results (5/6)
“Mother and Daughter” object in the “Mother and Daughter” sequence
![Page 17: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/17.jpg)
Experimental Results (6/6)
“Background” object in the “Mother and Daughter” sequence
![Page 18: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/18.jpg)
Discussion (1/2)• The estimated quality of the received
video can be used by the service provider as a feedback information for billing purposes– False declaration about the QoS
• If the mobile station declares a received quality lower than the provided one, it implies that the channel is not suitable for current bit rate and the base station will lower the emitted bit rate in a few seconds.
• If the mobile station declares a null quality, the operator interrupt the call.
• Frequent declaration of poor or null quality is a valid reason to stop further access from the user.
![Page 19: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/19.jpg)
Discussion (2/2)
• The mobile station must perform real-time processing to evaluate the QoS.– As compared to the complexity of the
MPEG-2/4 codec, the overheads of the QoS evaluation scheme is negligible.
![Page 20: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/20.jpg)
Conclusions• An unconventional method of tracing
watermarking as a hidden technique suited for estimating the QoS in multimedia mobile communications has been presented.
• The error affecting the watermark is very sensitive to the channel bit error rate and to the compression rate.
• One can blindly estimate the QoS of the video communication
![Page 21: Blind Quality Assessment System for Multimedia Communications Using Tracing Watermarking P. Campisi, M. Carli, G. Giunta and A. Neri IEEE Transactions](https://reader036.vdocuments.site/reader036/viewer/2022062317/5a4d1b567f8b9ab0599a9577/html5/thumbnails/21.jpg)
Application Scenarios
• Control feedback to the sending user of the effective quality of the link
• Detailed information to the operator for billing purposes
• Diagnostic information to the operator on the effective status of the link