Theme 6 - It is all in the quality

For this week's theme, I choose yet another paper related to digital video processing. This one is called "Quality Assessment of D10 and DV25 Video Codecs for Broadcasting Purposes", and while quite dated by now, it is still a very interesting read. Since this paper was also a case study, I will be using it for both parts of this theme.

Qualitative research can be many things...

Question 1
This particular paper is somewhat unusual, for its specific field, in the sense that it is qualitative-oriented rather than quantitative. It revolves around a psychovisual experiment, using the MDS-method, to assert the difference in perceived quality between two different video codecs.

The MDS-method, Multi-dimensional Scaling, is used to visualise how similar (level of similarity) two or more instances of a given case really are. This is particularly useful, as the above notion suggests, when you are interested in the relation between these specific instances, along a chosen metric, and especially powerful when you want to explore an unknown difference.

Question 2
I am not sure I learned anything new about qualitative studies as a whole from this paper, however I did learn about the particular method mentioned above. This is of course somewhat unfair since half of our previous seminar (for various reason) was used to cover this specific question.

Question 3
In some sense the main methodological strength of the paper is also its greatest problem. Because the qualitative evaluation is psychological in nature, one has to account for a whole science of contextual parameters in order to come up with any meaningful result. There is definitely an advantage to such a method, in this case; studying the perceived quality (i.e. asking people what they think) instead of using image processing to mathematically measure the difference. The most prominent example being that since most information contained within a digital image cannot even be registered by the human eye, a "high quality" image, as in higher PSNR, might still lack the specific parts "we look for".

... in this case, it is a study

Question 1
A case study is an examination of a particular subject, its details and any related contextual conditions. In other words, you study a something existing in a specific time and place, such as an event or a group of people. All such studies, much like any other form of research, make usage of one or more methods when analysing the chosen phenomena.


Question 2

- DISCLAIMER - As I mentioned above I used the same paper for both question sets

To start of the research question itself is quite simple, which one of the two mentioned codecs produce the best perceived quality? At the time of its publishing, there were no similar studies, and since the codecs are built upon the same mathematical methods a qualitative study yields a far better result than a pure quantitative one.