A “tourist guide” to Mathematical Morphology ?

After spending so many years using tools coming from Mathematical Morphology I endeavored to build a small set of lectures on this subject with the determination to stick to the basics and favor illustrations and intuitive understanding rather that mathematical formulas.

The whole set (slides, transcripts and references) is there: MorphoDiapo

Comparatively to other resources on the same subject, these lectures look more like a “holiday pictures sideshow”  but in order to keep it interesting I built it with the main objective to introduce and explain as clearly as possible measures that make it possible to analyze the shapes of objects present on an image.

Giving a glimpse backstage: these SVG slides could be built thanks to Inkscape et and its LaTex and JessyInk extensions.

Academic Genealogy

Since genealogy is trendy and the weather on Paris is a bit gloomy I played a little with the Mathematics Genealogy Project.

The aim of this project is to trace back the scientific lineage of “all” mathematicians throughout history. The main link considered is the “student – advisor” relationship, though sometimes looser relations are taken into account like “student – professor” (for mere lectures or mail correspondence, even when a PhD or a specific degree is not at stake) or even brotherhood (as for the Bernoulli brothers — Jacob and Johann -  who actually worked hand in hand during their studies)

Here, I tried to follow my scientific legacy starting with my PhD advisor Dominque Jeulin, his own advisor Jean Serra, himself having had Georges Matheron as advisor, and so on…

Academic Lineage

Mathematical lineage (click on image to get the full version)

Continue reading

publication (noun): The act of publishing printed or other matter

A quick post to celebrate the online availability of my PhD thesis !

Whatever its (lame) scientific significance, this will remain the very first book I’ve ever written, and, what’s more one that took me 3 years of hard work to gather its content.

After waiting 3 more years for it to come out the “confidential” closet in which it had been put, this is afterall a nice occasion to rejoice.

Continue reading

Correlation does not imply causation

I’m trying something new today: a small experiment falling in between programming and scientific litterature.

I’ve been interested in “literate programming” for quite some time now, without going any further though. But I recently found the tools that would make it actually practicable and decided to dive in.

So, I’ve implemented a very small Python module that illustrates the notion of autocovariance and autocorrelation and the way to compute them via Fourier transforms:

Literate MusingAutocorrelations

The subject is not new at all, but such measures are something that I had to re-implement (for image processing puproses) more than I’d like to in past years, and each possible implementation having its own statistical biases, it always takes time to remember all the tricky details. So, this new document should at least help (me) quickly find all the relevant info.

This was also a subject underpinning a big part of my PhD work… a PhD that I defended 2 years ago, is there some “causation” here ?

:)

For the technical side of things, the “tools” that helped me setup this “experiment” are essentially:

Culture on internet: highs…

Since Miro2.0 has been released a few weeks ago, I’ve had some time to use it and I thought I could take this occasion to share some of the podcasts I folow regularly on the internet.

But first of all a great thanx to Miro’s developer that did a great job in tackling the two major problems I had with the software: impossibility to navigate the list of feeds, while playing a video (or audio) and performance. The only remaining thing on my whishlist would be that Miro handles image feeds (ala CoolIris for instance).

Now for the list of podcasts I listen to, most of them are in french, so here is a quick selection among the english speaking ones:

For a more exhaustive list see my OPML subscription file: miro_subscriptions