ACM mmsec 09, the hidden (and fun) parts

By Tsatsiki
Miss cucumber is right, there were interesting papers in the ACM workshop on Multimedia and Security. And the participants had some fun as well. This is mainly thanks to one guy: Scoot Craver, the co-chair of the conference.

Scott is born data-hider, let me tell you why.
Firstly, when the participants arrived to register, they were offered a nice usb key like this one:


- these were the two first hidden messages. (look at it upside down)







Secondly, Scoot launched the workshop saying that there was a hidden (and cyphered) message on the building of the conference (the Friend Center in Princeton). This was true indeed since during the next coffee break the participants were able to see this:

- this was the second hidden message! Did Scoot add the bricks just for the conference? François Cayre, knowing ASCII code by heart, succeeded to decipher the message, while some others asked Google for the solutions.


Of course the talks of the workshop were also filled with hidden information (it was voluntary or not) and two of them were rather fun:
- the magic display cable (Atakli et al.) which is able to recognize, descramble and display encrypted images of a computer desktop on the fly,
- during the rump session, a teaser of the first contest on steganography was presented by T. Pevny. According to him it will be launch in 2010 and called BOSS (for Break Our Steganographic Scheme). I think that this contest is going to be rather popular.

Let's hope that WIFS 2009, the next conference on data-hiding, will be as fun as this one.

ACM mmsec 2009: the rise of the dragon?

ACM mmsec took place under the black and orange flags of the amazing campus of Princeton University, 7-8 september.

The general chair was Ed Felten, Prof. of Computer Science and Public Affairs (might be the only one to have this title), and the director of the brand new Center for Information Technology Policy.

He made a very good introduction speech calling for more discussions between computer scientists and law makers, and then, he almost disappeared from the conference (I could understand, after all, Monday was Labor day)! Never mind, Scott Craver organized a very good social event between nowhere and goodbye in a house near the lake and the lunch in Prospect House was also terrific.

Helas, there were only 40 to 50 attendees! Scott Craver defended the idea that, despite the decline of submissions, their overall quality was good. Therefore, good papers don't fear the financial crisis (if this is the real reason for the submission number decrease).

Like in IH'09, there were indeed quite few papers of lower interest so that the idea of S. Craver holds.

So, what is hot? MP3 quality assessment! The scenario is as follows: a low bitrate MP3 file is transcoded at a higher bitrate and sold as a pretendedly high quality tune. How to detect this? By revealing the signs of a double compression. Actually, H. Farid also presented a talk about double compression to detect video edition. Three speeches, three different approaches: from the "I don't know what happens, but my SVM will tell" heuristic to a more convincing statistical modeling.

The asian dragon? The era of old asian professors reading with difficulty their slides full of approximative english long sentences is over. We are witnessing, in multimedia security (it is probably the case for a long time in other fields), a new generation of young, convincing communicator and quite good researchers. Among the speakers but also in the audience. Some aggressive but very relevant questions were asked.

We also met the future leaders in steganalysis: Jan Kodovsky and Jan Judas replacing the two Tomas evil twins leaving Binghamton (Tomas has already left for Grenoble and soon Praha, while Tomas is finishing his ph.D). The Czech eagle vs. the asian dragon... An interesting battle is coming.

Talks I liked:

- "Two Key Estimation for the Borken Arrows watermarking Scheme", P. Bas and A. Westfeld. Another evil twin couple is born. How to break Broken Arrows (the watermarking technique used for BOWS-2) again and again. I guess, since P. Bas is one of the inventor, it is easier to analyze one's own scheme.

- "Additive Spread-spectrum watermarking detection in demosaicked images", P. Meerwald and A. Uhl. Hack your camera! The Austrian team modified the firmware of a camera so that watermark embedding is done in the device just after the CFA capture. Cool hack! This rises the challenge of watermark detection after demosaicking.

- "Exposing digital forgeries in video by detecting double quantization", W. Wang and H. Farid.

The best paper of all the "double compression" works in this conference. Nice but oversimple talk (as always), which plays the role of a teaser to read the paper for further information. The paper is indeed good: statistical model, identification of the parameters with an 'E-M' like algorithm, hypothesis test.

- "Calibration revisited" J. Kodovsky and J. Fridrich. My understanding about calibration is that natural images follow no statistical model or a so flat pdf p(X) that no hypothesis test will work with so few discrimination. Calibration is about estimating a set of parameters P specializing the model, switching from the flat marginal p(X) to the much more contrasted conditional p(X|P). Calibration is about statistical modeling. Whereas this paper uses calibration with SVM, which is, IMHO, the contrary of statistical modeling. SVM means "I don't know how to model things but, once trained, the SVM will find its way". Therefore, I was a little confused by this approach. But the results are there.

- "Square Root Law" A. Ker. During IH'09, I was not convinced by the talks about the square root law: It is was not enough formalized, the assumptions were not clearly delimited, so that if the law indeed holds, we could not tell the real reasons. This is exactly what Andrew Ker did in between the two conferences, and now, this became serious math with theorem (conditions and results) and proofs.

- "Detection of seam carving and localization of seam insertions an digital images" A. Sarkar, L. Natarj, and B. Manjunath. Seam carving is this image processing tool. It a priori seems a 'mission impossible' to detect seam carving, but a heavy tool (a 324 Markov features classifier) starts producing good results. Work in progress. A maybe too dense presentation.

Miss Cucumber

IEEE DSP 2009

What would you expect from a conference located in a gorgeous place (Santorini, Greece), during summer and with an acceptance rate of 76%?

It was very good (Sea, Sirtaki & Sun), and very bad (security with no secret key, security with no threat analysis,...). So bad? No, so nice invited talk by M. Unser, nice compressed sensing presentations (link with traitor tracing becomes more and more obvious). And what about watermarking? Although there were 5 sessions dealing with multimedia security and forensics, I was quite disappointed in general. Some good presentations:

• Content-Adaptive Semi-Fragile Image Authentication Based on JPEG2000 Compression, M. Schlauweg. The idea of J. Eggers but pushed to a more practical level.
• Blind DT-CWT Domain Additive Spread-Spectrum Watermark Detection, P. Meerwald. The idea of Loo & Kingsbury, but improved to a more practical level. At last, someone no longer uses PSNR for quality assessment. 
  
• Analysis of Denoising Filters for Photo Response Non Uniformity Noise Extraction in Source Camera Identification, I. Amerini. At last, somebody noticed that since PRNU is multiplicative noise, then, linear estimation is not the optimal. Too bad, the correlation is still here.

Miss Cucumber from the tatziki country.

Something's rotten in the realm of watermarking?

2009 is a bad year for watermarking conferences. All of them (IH, MMSEC, IWDW, SPIE, ...) are facing a low submission number. There are certainly too many yearly conferences, but no PC wants to kill its child. Too bad! Clearly, publishing papers is no longer a problem.

Some think this is due to the brand new "yet another conference on content security", ie. WIFS. THE conference which would wipe out all the others. But no! There were 120 submissions. Not enough for the organizers who decided to restrain their ambition, as I have been told. WIFS might be single threaded without any poster session (enter the rumormill!). Maybe, they were expecting too much for a first edition.

Some think this is due to the financial crisis. Researchers have no money for traveling. The Earth thanks them for the saved CO2. But no! Conferences of our cousins the crypto broke submission records in 2009.

So? what is rotten in the realm of watermarking? Time to move on to another realm? Which one?

But there is a glimpse of hope at the end of the 2009 black tunnel. According to the latest Journal Citation Report, the TIFS journal's impact factor has jumped from 1.089 to 2.23. It is currently in the first third of the JCR ranking for electrical engineering journals, which is really good.

Conclusion: Stop calling your travel agency, and write long deep and damn good journal articles!

Miss Cucumber switching the gossip radio off. 

Who judges? Who does science?

No need for links here. One day or another, we all stumbled upon one of these papers in which the hidden payload is a binary bitmap logo. Maybe it is deeply rooted in one of the founding papers of the field.

Anyway. The point is: Are we doing science or not?

If we are really doing some science, we are preferably to talk about quantities like capacity. In real life with sound experiments, it boils down to compute/simulate the Bit Error Rate (BER).

But then people come around and say that a judge will better recognize a logo than the fact that a probability is quite low. We agree. A judge does not do science.

But it is always easier to make a covert channel with a given BER carry a given logo, than to compare two given data-hiding schemes only based on the visual appearance of the recovered logos.

We all aim at doing science. So please: No more logos. Please.

On image steganography (as people do it)

For sure, steganography is becoming more and more interesting.

But hey! What about all those papers dealing with bitmap images?

Do any of you Estimated Readers ever share any images in BMP or PGM format? Do you all trust these awesome steganographic rates?

Images in GIF format are encumbered with patents and JPEG2000, apart from Digital Cinema, is mostly unused on the WWW. Old school JPEG still rules.

Guess what? Due to the DCT energy compaction properties, it is pretty sure that the actual steganographic rates are to be found much lower than what claimed for bitmap images.

And it is because of the popularity of JPEG, not because of distortion reasons.

The real issue is not how many bits are to be hidden into lena.pgm, rather in lena.jpg. Even better: how many bits can be hidden into a bunch of contents?

Distortion-free 3D steganography!

Most of us think as data-hiding as a communication problem with side-information at the encoder. This widely accepted view has led to dramatic improvements of data-hiding techniques over the years. As a direct consequence, any watermark can be seen as a noise that is added to the host content.

These days, a new trend is emerging in 3D data-hiding. Such data is twofold: there is geometry and there is connectivity. Roughly speaking, geometry is a bunch of 3D Cartesian samples, and connectivity is the way the vertices connect to one another. Some people, including several of the most respected in the mesh processing area (see this paper by Bogomjakov, Gotsman and Isenburg) [1], want to hide information in the way the connectivity is described: since geometry is not affected, it leads to distortion-free data-hiding!

To us, it reads like "communications at no power" and other "infinite capacity channel" odd stories. Frightening. Sounds like a definitive breakthrough in information theory.

But wait!

There is a problem. The real question is whether connectivity should be considered useful to describe 3D data. The answer is not easy. There is already a host of works dealing with 3D point clouds data and how to synthesize connectivity from scratch. But there is more: Isenburg has done extremely interesting works on some sort of a "ghost geometry" that is already present only in the connectivity information (see his connectivity shapes).

Turn again to data-hiding as we know it: with regularly sampled host contents. Those ones that we love and that do not need description, except the dimensions of the image or the duration of the song.

Now let's make the following crazy assumption: each and every pixel of an image is to be numbered and transmitted separately. The decoder now has to know the neighborhood of every pixel. Yes. And we do not transmit triangles anymore (like for 3D meshes), but quads. That's images with explicit connectivity.

And finally we can do the same thing: distortion-free data-hiding for images. Sounds weird uh?

Interestingly enough, the paper by Bogomjakov et al. states that one needs to have a reference ordering of the vertices, so the decoder can catch the difference with the transmitted connectivity and compute the hidden message. Similar ideas are shared with so-called permutation watermarking. Therefore, there is actually some sort of distortion in their scheme. BTW, their scheme also has a capacity!

Why people transmitting images do not send quads with YUV/RGB values being the attribute data? That's because everyone assumes this reference ordering of the pixels inside an image. Everyone is happy with YUV/RGB values only.

And since there is this implicit public reference ordering of the vertices and the decoder can catch the difference, an adversary should be able to detect the hidden information quite easily. Then why is it called steganography?

So let's make it clear once for all: these guys do data-hiding on graphs. They don't do distortion-free 3D steganography. For sure.



[1] Although Gotsman already did some sort of Karhunen-Loeve Transform based compression for 3D meshes... Either it's not as real-time as claimed (one needs the decoder to compute the basis vectors -- cost: several O(N^3) diagonalizations of ~500x500 Laplacian matrices), or it is not compression (one needs to transmit the basis vectors). The improvement presented here suffers from a problem apparented to the Gibbs phenomenon.