Unless the computer system let me down, feed-back on Coursework 2 has been returned to most of you, including all those who handed in on time. The last three scripts will be marked no later than Monday. If you did hand in on time, and has not received the email, please complain by email. Printed mark sheets will be available in the p/g office from Monday afternoon.
Answers regarding the six test images have been published on the Coursework page.
There was limited attendence on the lecture 24 April. No new material was covered and no slides will be published from the session.
The very last lecture will be Wednesday 2 May 10-11 in 03DK01. I will continue to clear up probable misunderstandings and difficult points in previous lectures. Again, I do not expect to publish any dedicated slides for the session, as I am not sure I will use any.
The marked coursework has been returned to the P/G office. I am still working on some general feed-back which will be published here.
This Wednesday will be a mainly practical session. Slides and exercises are now available.
Moves this week only to 24BB02
Hopefully, you all know about the Maths/Stats support sessions run by Ms Howell of the Maths department for Computing students; on Thursdays 1-2pm. Normally, the sessions is in 44BB02, but on the 22 March, it will be in 24BB02.
The support sessions allows you to ask for help on preliminary material from mathematics and statistics, which you need in the module. This includes hypothesis testing, probability distributions, and solving quadratic equations...
There appears to be a misprint in Jessica Fridrich's paper on pairs analysis in the formula estimating R'(0.5), and this was replicated in my slides (which have now been corrected). The given formula estimates the absolute frequency, wheras R'(0.5) is defined to be the relative frequency.
I have published sample m-files for the task on chi-square steganalysis from the Formative Coursework, see the chi2demo page.
Slides and exercise sheets for tomorrow have been made available. There are also additions to the reference (reading) list (Westfeld's paper on F5). Remember, though, that the essential reading is listed in the slides.
The Matlab JPEG Toolbox can be used for research and education free of charge. It is hard to find on the web, so I have made a local copy, in case you want to try it.
Username and password will be given in lecture (if someone reminds me). It is used because I do not control the copyright of the material.
I am sorry to say that I am unable to provide any help in installing the package on any non-unix-like system, so I hope the README file included contain sufficient instructions. I do not know if Gary can help.
New information about the courseworks is now available, including
Please ask me if anything is unclear.
Material for next week is ready.
I note that most of you have not picked up coursework scripts for assessing. You are supposed to assess the coursework by Monday! I will not return assessed coursework to anyone who has not assessed another script.
The material for next week is almost complete (see Programme). A preliminary version is available. A minor addition will be made tomorrow, but the syllabus should be clearly indicated already.
Revised version published. Plotting the p-value is more interesting than plotting the chi2 statistic. The revised version reflects this fact. The revised version does not give you extra work, but the work is broken up into more sensible steps, so it should be simpler to do.
I have made a demo, including examples of the derivation of the Z and Z' sequences. Please let me no whether this answers the questions asked in today's lecture or not.
A list of available demos and matlab hints is maintained on the Software page.
There have been questions about the extract ( img )
function (and similar) in the courseworks. Should it not be
extract ( img, msg_length )?
The answer is, probably.
There are a couple of ways to do it.
In general, you may choose the approach which you think works best or is most useful. You are allowed to deviate from the specified function syntax in order to enhance the functionality.
I was also asked whether you should do error checking (such as verifying that the message is not longer than the image capacity). The answer is, that I would not expect it for a 50% mark, but I would expect it in a 90% mark.
Slides and exercises for Chapter 6 have been made available. I hope they are complete and final.
The programme for Wednesday will be
Due to popular demand, I have made a series of sample plots of the chi2 statistic and the corresponding p-value (cumulative probability). I have not included the m-functions doing the calculations, as you will have to do those for your coursework. However, a demo is included to show how they are supposed to be used. New page
Assignments for the peer assessment and for coursework 1 have been made available; see «Coursework» in the Right Hand menu. The overlap between the two sets is intentional.
Those who attended this morning will know that I learnt how to do a chi-square test for steganalysis. In order to have full benefit from the session next week, you should do your best to complete Exercise Set 3 Items 1-2, as well as Exercise Set 2. The functions will be reused in conjunction with palette (GIF) images.
Functions to compute CDF, PDF, and inverse CDF in Matlab are available. These functions do not do all the error checking that the Statistics Toolbox would do, but they do the calculations.
Slides and exercises up to and including Chapter 4 have been completed. I will still correct any errors found, but they are intended to have reached their final form.
Chapter 5 onwards are still in draft form.
Next Wednest (31st) I plan to continue with Exercise Set 2, Part 2-3 for about 1h. I will then give a talk on the chi-square test for steganalysis; i.e. the second half of Chapter 3. The last hour will be spent on Exercise Set 3.
![[Filler]](/images/umiext.jpg)