BuschnicK

Now this my friends, is a truly beautiful toy. Aesthetic flash game if I ever saw one.

More fotos.

Thomas just blogged about BinDiff 3.0 (DifferDeluxe).

I have participated in one of the game prototyping challenges over at DanC's Lostgarden. The game (I'd rather call it a toy) is about peas thinking they are ninjas. They climb up blocks you place for them and jump from the highest points they can reach. If they survive they'll score points for every bounce - the higher the fall better. For detailed game rules and all the elite ninja moves the peas can do visit Dan's description of the game design.

Keys:
- ESC quits the game
- Space spawns peas
- Left mouse places a block
- Right mouse deletes a block
- Mousewheel or arrow keys cycle through available block types
That's about it.

Full source is included. Depends on a couple of external libraries. Do whatever you want with it - would be cool to send an email my way telling me whether it was useful.

There are a couple of known bugs, mainly that some impacts don't register properly, peas can get stuck when "overbuilt" and that peas will sometimes float through blocks until they are back on course. It's fully playable though and all basic game mechanics work as intended.

Download here.

cheers,

BuschnicK

An article about plastic pollution in our oceans.

Except for the small amount that?s been incinerated?and it?s a very small amount?every bit of plastic ever made still exists.

Seriously scary stuff. Why do we wrap candy in packaging that'll survive the product and it's consumer by hundreds of years?

SUV sales plummeting

Stephen King on video game violence

Could Massachusetts legislators find better ways to watch out for the kiddies? Man, I sure hope so, because there's a lot more to America's culture of violence than Resident Evil 4.

What really makes me insane is how eager politicians are to use the pop culture ? not just videogames but TV, movies, even Harry Potter ? as a whipping boy. It's easy for them, even sort of fun, because the pop-cult always hollers nice and loud. Also, it allows legislators to ignore the elephants in the living room. Elephant One is the ever-deepening divide between the haves and have-nots in this country, a situation guys like Fiddy and Snoop have been indirectly rapping about for years. Elephant Two is America's almost pathological love of guns. It was too easy for critics to claim ? falsely, it turned out ? that Cho Seung-Hui (the Virginia Tech killer) was a fan of Counter-Strike; I just wish to God that legislators were as eager to point out that this nutball had no problem obtaining a 9mm semiautomatic handgun.

Take a look at this advertisement photo for an "American Dream Home". You have to wonder whether this was even built for humans - I can't see an entrance except coming in through the garages.


I for one would certainly prefer to live in an area like this:



Fortunately, some people still have some common sense left, are not afraid to voice it and actually have some say in urban planning and city design. Here is an inspiring interview with Hermann Knoflacher.

Es lebe das Mittelklasse Gehzeug!

While reading the The Phaidon Atlas of Contemporary World Architecture. Comprehensive Edition (with a weight of 8kg the heaviest book I own ;-) ) I thought it would be a real shame if all our virtual architecture, i.e. computer game levels, especially 3D maps and buildings, will be lost forever once the corresponding games cease to work on new hardware. Alas, this being the internet, someone has thought of this before (for example SamSahrani). I for one would like to see a museum of computer game architecture. Think of the floating in space jump pad levels of Quake3, the organic "in an alien body" maps of System Shock 2 or the rotating maps of Unreal Tournament. Or the more realistic maps of the various World War 2 games or Halflife. Not to mention the online worlds in World of Warcraft, Second Life or the Sims. There are tons of very creative examples like this. Would be a shame to lose all that creativity. Architecture is architecture after all. The virtual kind may have different constraints and trade-offs than the real world one, but just like the real thing it's a mix of art and craft.

Either that or I'm just nostalgic for countless lost hours sank into self made levels for DooM 2, Duke Nuke'em and Quake 3.

I took the politcal compass test and scored:
Economic Left/Right: -6.38
Social Libertarian/Authoritarian: -5.95
which, according to the test, puts me in a group with Nelson Mandela, Ghandi and the Dalai Lama. Fine with me I guess, considering George W. Bush, Hitler and Stalin are also on the map...

Kleine Utopie

My brother is one of the puppets on the stairs and wrote and performed the song (second half of the video).

Buffet invests in railroads, railroads are cheaper, safer, make more ecologic sense, use less space and make less noise than cars.

Ah wtf. This was supposed to become a somewhat coherent entry about trains, investment opportunities and shit. I'll go sit on the balkony with my girl now instead of rambling.

Why don't you go and buy a "car-scratcher" from cycles for heroes in the meantime?

From a Bloomberg article:

Gasoline use is rising almost 5 percent above the five-year average.
[...]
Many Americans have no choice but to drive more, says Christopher Knittel, an economist who studies fuel consumption at the University of California in Davis.

Argh! There is lots of bad stuff to say about Sony, like it distributing rootkits for the good of the customer, but this time it's personal. I was dumb enough to buy a portable minidisc player (before all the mp3 players came out). The hardware is kinda ok, but the software must be the most hideously rotten piece of fucked up shit ever let lose on humankind. Of course it's your only option to use with the hardware. Of course it's windows only. Of course it has to scan all your drives in order to "register" your music. Naturally it only writes Sony native formats. Of course it doesn't read ogg vorbis. Of course it crashes about once a minute. For updating it you can of course be expected to install Internet Explorer (Sony's page doesn't like FireFox) and restart your computer about 500 times. Also, a simple download cannot be done without ActiveX and Flash activated. etc etc... It's just unbelievable that they should get away with it. Die Sony! Die, die, die!

Go Apple go! Drive em into the ground.

"If you took a nap in your office in a big company, it would seem unprofessional. But if you're starting a startup and you fall asleep in the middle of the day, your cofounders will just assume you were tired."

Paul Graham in his latest essay "Why to Not Not Start a Startup"

Although I've written large chunks of this software I'm not a typical user of it (i.e. I'm usually producing my own bugs instead of taking someone else's apart in binary code). All the more interesting to see what kind of (freaky I might add) stuff folks pull off with it. Ero published a nice article where he shows some debugging work, stringing together a PE executable with a hex editor of all things!

A couple of "human flowers" - highway interchanges viewed from above. The last one is from the movie "Alpha Dog". More in the "Museum of Ridiculous Highway Design".
Apparently abominations like these are but a small sacrifice for "personal freedom"...

Today he missed the rising sun by three hours. He met his day in the shower, washing his hair with shampoo that was guaranteed to have never been put in a bunny's eyes and from which ten percent of the profits went to save the whales. He lathered his face with shaving cream free of chlorofluorocarbons, thereby saving the ozone layer. He breakfasted on fertile eggs laid by sexually satisfied chickens that were allowed to range while listening to Brahms, and muffins made with pesticide-free grain, so no eagle-egg shells were weakened by his thoughtless consumption. He scrambled the eggs in margarine free of tropical oils, thus preserving the rain forest, and he added milk from a carton made of recycled paper and shipped from a small family farm. By the time he finished his second cup of coffee, which would presumably help to educate the children of a poor peasant farmer named Juan Valdez, Sam was on the verge of congratulating himself for single-handedly saving the planet just by getting up in the morning. He would have been surprised, however, if someone had told him that it had been two years since he had set foot on unpaved ground.

- Christopher Moore, Coyote Blue.

the freeways are a psychological
entanglement of
warped souls,
dying flowers in the dying hour
of the dying day....

what you see on the freeway is just what there is,
a funeral procession of the dead,
the greatest horror of our time in motion.

- Charles Bukowski

Pollution in China, already at disastrous levels, is growing faster than anyone could possibly control it. English Der Spiegel article on China

And what are we doing? Investing money into the "emerging markets", feeling disappointed when they don't reach our 30% p.a. growth expectations. Poisoning themselves to death is alright as long as we profit and the dirt stays on the other side of the globe. Now that the first clouds drift over Europe and Japan people at least start noticing...

China is expected to trump the biggest current polluter, the US, in the very near future. I bet you a thousand carbon credits that will be used as an excuse to pollute even more, not less. The argument you hear even now goes something like this: Why should I cut back on my emissions or my environmental footprint - look at them, they are even worse than me! If I change, that won't make a difference anyways. You'd think this argument is so obviously flawed noone would fall for it. Yet at the same time you hear it again and again. People driving their fucking cars everywhere because "what difference does my Prius make compared to his evil SUV?" I don't need to change my habits. Not me. Not me.
By this logic you can justify any crime because there'll always be one that is worse...

There is an analysis of this phenomenon in mathematical game theory: Cooperation would be for the net benefit of everyone, yet it's for your own short term benefit not to cooperate and thus it never works. Unfortunately nature suffers from exactly this phenomenon. For the global good of the planet nations should be able to agree on environmental policy yet it's in everyone's short term interest to let "the others do it" and boost the local economy instead.

Everyone will cry foul at China once we finally reach the tipping point. Yet they are only repeating our mistakes. If we were allowed to make them and if we have the right to our standards of living - why shouldn't they? It's gonna get real nasty once the roles reverse - imagine the reckless USA having to play economic catch up with China. Uh oh.

A new coal fired power plant per week - I bet you we'll be building nuclear power plants by the dozen in the name of ecology and green energy before I turn 35. Just so we stop the immediate threat.

Us fucking hypocrits.

I just finished reading Neal Stephenson's Cryptonomicon. It's a daring story with lots of characters and parallel plots all revolving around cryptography, gold treasure hunting and second world war action. This in itself is not really compelling. What is though, at least for me, is that this is a book by a geek for geeks and about geeks.
It's written by a geek because Stephenson has a way of playing language kungfu that seems to scream look at my wits/IQ and goes off on weird tangents all the time. It appeals to the inherent multi-tasking addiction in your average nerd in that you can ponder some off-hand remark and fancy idea thrown at you in a subordinate clause while continuing through the main thread.
It's written for geeks - who else would want to read detailed descriptions of crypto-systems as part of a novell? But that's actually besides the point. The real reason is that geekdom is characterized so well in this book that you could actually use it as a definition. The leading characters are all more or less socially inept almost to the point of autists or Asbergers and at the same time math geniouses living in their own alternate realities. I laughed out loud at a scene at a dinner table where Randy, chronically forgetting all the unimportant details about the people around him, like - say - names, blurts into the discussion after enduring silently for a while and turns it into a heated debate. The sarcastic remark is that "normal" people flock into "consent groups" soon-after, because a real argument is too strenuous.
And it's about geeks... well, just take the title for a hint.

I really enjoyed it.

I wasn't in the mood for "real" work this afternoon so I decided to bug my girlfriend instead. She has developed a serious Sudoku addiction lately and solves them by the dozen. I thought it was a beautiful problem for a genetic algorithm to solve so I gave it a shot. The whole thing took little over 3 hours to code, starting from scratch using only the STL and MFC. The result won't win any beauty prices and is most certainly not the fastest or most intelligent solver out there - but it works.

I've used the most basic of all GAs, implementing truncation selection for selecting suitable parents. Childs are created by using the first n blocks of one parent and the last 9 - n from the other. Mutation is archieved by simply swapping random fields in a block. The fitness function counts how often a digit appears in a row or a column, every deviation from one counts as an error.

Sudoku is a pretty hard problem for a GA because it has many solutions with the same error count, i.e. the same fitness. Such a multimodal error function needs a GA that is better capable of exploration vs exploitation than mine. Using truncation selection for inheritance is pretty bad, cause it will not usually try enough approaches and easily get stuck in local minima. I "solved" the problem by simply using a high chance for mutations and thus a high random factor.

You can tell that the GA isn't very good for exploring because it is highly dependant on the initial parameters and population. Usually, if you choose the right GA for the problem at hand, GAs are pretty robust and don't change much with respect to initialization.

As it is, the algorithm needs a couple of seconds and about five to ten thausand iterations to solve a moderately complex sudoku. For the fiendish and extra hard problems it'll take minutes and several hundred thausand iterations.

Anyways - that was fun ;-)

I have received a very well done eBay phishing spam mail that made it through both of my spam filters. I took the time to actually read the thing. It has a pretty long winded explanation of why I owe eBay some money and goes on to say where I should transfer it and how much. The line stating the amount due contained a pretty funny error:

Faelliger Gesamtbetrag:
||%RND_FIRST_DIGIT50,68

%RND_FIRST_DIGIT ?! LOL! I owe someone a random amount of money ;-) I can only assume that the original spam mail was generated in a different language (most likely english) and a different character set. I guess the dollar symbol was supposed to be replaced by a euro one and the parser choked on that. And, btw, we put our currency symbol at the end of the number, not in front anyways.


On another note I stumbled across some malware recently that infects your machine and goes on to encrypt random pieces of your data. It then pops up a dialog telling you your data is being held hostage and that you have to pay a certain amount to get the password to unlock it again. I found this a very creative idea and wonder if they actually do earn money with it? More original than just deleting and destroying stuff randomly in any case.

I guess with good open source encryption like: http://www.truecrypt.org/ it would actually be quite simple to write such a trojan that would work very well indeed.

I have been busy at work recently building neural network ensembles for solving quite a complex prediction/classification problem. There are two major issues to solve for my particular problem:

1) How do you reduce the dimensionality of your input data to something manageable while still maintaining the relevant information?

2) How do you find a set of neural networks that each produces good results individually while making it's prediction errors in disjoint areas of the input space, thus archieving maximal benefit from grouping.

In my research for 1) I've stumbled across two majorly cool algorithms: Isomap and Locally Linear Embedding. Both of which try to solve the same problem: Finding the inherent dimensionality of a dataset hidden in a high dimensional input space (finding an embedded manifold). As a simple example you can view the 2D still fotograph of a three dimensional object as a vector of pixels. This vector will have an insanely high number of components (i.e. dimensions), your average digital camera has millions. On the other hand the real world object they describe usually has far less degrees of freedom, thus needing far fewer components to be described. The aforementioned papers both have very cool graphical examples for this.
Unfortunately both algorithms suffer from the fact that you cannot train them on one dataset and subsequently apply them to another, unknown, one. Real world problems are often of this nature though... Still, I think they are among the coolest ideas in recent years and definately a worthwhile direction for more research.
I also investigated Self Organizing Maps, but they suffer from the same problem like LLE and IsoMap and additionally also include quantization as a necessary step, thus corrupting your data more than necessary.
In the end, I settled for boring old plain vanilla principal component analysis, which manages to simplify my data at least a little bit.

For the second problem ( 2) mentioned above) the primary issues are how to determine network topology in the first place and how to compare networks against each other. I implemented a genetic algorithm growing thausands of networks, thus solving the network topolgy problem in some pretty brute force way. Comparing network quality is more difficult than it sounds though, the standard mean squared error usually used in training backpropagation nets isn't very useful for this. I ended up creating a gini curve/coefficient for my alpha beta errors (it's a two class classification problem) and, more importantly, scatterplots of the output neurons' activation. The latter plots are very interesting, creating funky curves for certain training configurations and transfer functions (radial basis functions are most freaky). They also offer insight into the particular training samples a net makes errors on, thus allowing an optimal strategy for choosing a set of nets.

Fun ;-)

Late at night I'm lying awake reading. My girlfriend curled up close to me, sound asleep. Her slow breathing and the soft rustle of pages turning the only sounds reaching me in my little island of light. I'm through with the book, thoughts wandering in comfortable idleness. Everything seems slow, silenced, numbed in happy melancholy. The book is over, the story ended. I start wondering about the author and read the cover flap:

BuschnicK is.

leaving.
working.
having.
teaching.
winning.
publishing.
living.
writing.

The author's description flap reduced to the verbs. Reduction. Time is a glacier.

One of the companies I work for has a contract for a project which is clearly military in nature. I refused to work on that project for moral reasons, as did many of my fellow colleagues. Anyways, the ensuing discussions reminded me of an excellent movie:

From the movie Good Will Hunting. Will is a young maths genious and is being interviewed for a job at the NSA:

Why shouldn't I work for the N.S.A.? That's a tough one, but I'll give it a shot. Say I'm working at N.S.A. Somebody puts a code on my desk, something nobody else can break. So I take a shot at it and maybe I break it. And I'm real happy with myself, 'cause I did my job well. But maybe that code was the location of some rebel army in North Africa or the Middle East. Once they have that location, they bomb the village where the rebels were hiding and fifteen hundred people I never had a problem with get killed. Now the politicians are sayin', "Send in the marines to secure the area" 'cause they don't give a shit. It won't be their kid over there, gettin' shot. Just like it wasn't them when their number was called, 'cause they were pullin' a tour in the National Guard. It'll be some guy from Southie takin' shrapnel in the ass. And he comes home to find that the plant he used to work at got exported to the country he just got back from. And the guy who put the shrapnel in his ass got his old job, 'cause he'll work for fifteen cents a day and no bathroom breaks. Meanwhile my buddy from Southie realizes the only reason he was over there was so we could install a government that would sell us oil at a good price. And of course the oil companies used the skirmish to scare up oil prices so they could turn a quick buck. A cute little ancillary benefit for them but it ain't helping my buddy at two-fifty a gallon. And naturally they're takin' their sweet time bringin' the oil back, and maybe even took the liberty of hiring an alcoholic skipper who likes to drink martinis and play slalom with the icebergs, and it ain't too long 'til he hits one, spills the oil and kills all the sea life in the North Atlantic. So my buddy's out of work and he can't afford to drive, so he's got to walk to the job interviews, which sucks 'cause the shrapnel in his ass is givin' him chronic hemorrhoids. And meanwhile he's starvin' 'cause every time he tries to get a bite to eat the only blue plate special they're servin' is North Atlantic scrod with Quaker State. So what do I think? I'm holdin' out for somethin' better. Why not just shoot my buddy, take his job and give it to his sworn enemy, hike up gas prices, bomb a village, club a baby seal, hit the hash pipe and join the National Guard? I could be elected president.

I'm working on a side project of mine involving a webcam, a projector and shadows. Anyways, webcams only have very small lenses and thus a strong fisheye distortion effect towards the edges. I needed a way to get rid of that. I dug out my copy of the excellent open source OpenCV image library and found a nice little function to do that for me. You can see the result in this post. The first image is the original, distorted one, the second is the undistorted version and the third shows the calibration object pattern detection.

Next step is getting rid of the trapezoid distortion which should be far easier since it only involves a linear transformation as opposed to the radial lens distortion. Finding suitable reference points will be interesting though (or pins as such fix points are called in GIS applications).

Notes on the human visual system and perceptual image analysis and segmentation

This is intended to become a loose collection of links and notes to gather my thoughts, ideas and resources on the topic in one place. Usually I have an unorganized mess of scribbles on various post-its, hyperlinks sitting in emails, browser favourites and text files. Also, this collection is intended to provoke feedback and get your neurons going, maybe we could spark some new ideas.

Human perception, especially vision, is still an active research area and many things are yet unknown. On the other hand it is of tremendous interest to decipher nature's tricks. With today's computing power and camera systems lots of highly useful and practical applications could be built - if only we had the right algorithms to deal with the data. One of the most important and relatively low level vision tasks is segmenting an image into it's semantic parts and components. Human's do that tremendously efficient and fast, without even concious thought. Computers on the other hand are still relatively inept at this task and only perform well for certain small subsets of the general problem. So the first thing should be trying to understand how humans do image segmentation.

Theories about the human visual system (some proven and true, others not) and ideas of how to emulate them algorithmically:

- different light wavelengths provoke reactions of varying intensity on the retina. For example we can distinguish different shades of green far more easily and in greater fidelity than any other color. Evolutionary this makes perfect sense, considering that the great outdoors living habitat presents itself primarily in greens.
Conclusion: model the colorspace accordingly, i.e. vary the dynamic range (bitdepth) used for representing signal strength in certain wavelengths. In the simplest case this means having more bits available for representing green than for any other color.

- Humans perceive certain colors to be similar to each other while others appear dissimilar.
Conclusion: model the colorspace accordingly, i.e. find a color space such that some mathematical metric represents human perception in color differences. One such colorspace is the CieLAB colorspace which has been devised with several hundred test candidates such that the euclidian distance metric corresponds to perceived color similarity. It still has quite a lot shortcomings though (see appendix).
Problems: As opposed to this model human color perception is not absolute, but spatially and contextually varying. Our perception of color depends on the colors directly surrounding it and the number and type of colors visible simultaniously. It is easier to see color variation in a local color gradient than to count the number of different shades of a color randomly dotted all over an image. Also, we can only distinguish so many colors at once (a relatively small number).

- The human eye is not modelled on a regular grid. Rather the perceptors are irregularly spaced on a rounded surface (the inner surface of the eye). On the other hand all artificial sensors adhere to a strict, regular and flat lattice. Question: does the irregularity of signal sampling in the human eye have any advantages over a fixed grid? Is it more robust dealing with aliasing problems? Is it more robust against over saturation with a certain spatial frequency? What purpose does it serfe?

- Physical light intensity does not correspond 1 to 1 to perceived light intensity. Research indicates that perceived intensity approximately follows a logarithmic function of physical, measured, intensity. Computers already try to capture this effect for example with the nonlinear gamma curve for display devices. When modelling a color space or normalizing one this should be taken into account.

- Humans seem to be more sensitive to changes in brightness than in color. The human eye has far more intensity than color receptors. This indicates that a color space that models brightness or intensity seperately from color would be desirable. Color could be represented with fewer bits than brightness. When building gaussian or laplacian image pyramids brightness could be sampled with a smaller (higher frequency) filter kernel while color could be smoothed much more with a larger filter producing a "perceptual image pyramid".
BTW, why are all the usual color spaces 3 dimensional? There is nothing inherently 3 dimensional about color. Would it make sense to find another representation? (Update: there actually seems to be a theory saying color is at least 3 dimensional)

- Humans only focus on a very small part of the whole visible field of view at a time. The spatial and color resolution of the eye falls off towards the edges. Color can only be perceived in the center of the field of view. Both of these facts would imply a non-linear, non-kartesian, coordinate system to represent images. The log-polar coordinate system might be one such choice, giving high spatial resolution at it's center and falling off towards the edges. Thus one could model a sharp and crisp "center of attention" while suppressing noise and uninteresting detail in the surrounding area. This would only deal with the spatial issue though. A similar approach should be used for color, so that an image fades to gray towards the eges and color detail is reduced.

- The human eye and associated nervous system is organized higly hierarchical. Processing and understanding of the received data happen at each level of the hierarchy. Some simple metrics and information seem to be gathered very early in the process such as "it's dark/it's bright". Vertical and horizontal breakage lines are detected pretty early (probably because they are needed for navigating our three dimensional surrounding) as are foreground and background objects. Foreground segmentation is a particularly interesting subject since a lot of information seems to be used here. Examples would be: is the object in motion; a tendancy to place objects "on the ground", i.e. the lower part of an image where gravity would have them; a light vs. dark object segmentation; stereoscoping 3D segmentation, i.e. how close is the object. The last part of the system is the actual pattern and heuristics matching happening in the visual cortex. Pattern matching meaning matching the brain's database of actual known objects to the newly encountered ones. Heuristics matching means matching known physical truths to a probably unknown object (This last point is especially interesting since it implies that things that "cannot be true" won't easily be perceived. Supposedly this is anectotedly true for the first sightings of Columbus' ships by the natives - they didn't know such constructs and thus had problems "seeing" them - although they were huge).
The hierarchical layout implies that an artificial perception system should probably be modelled the same way. Basic receptors, followed by ever more complex and complicated feature detectors followed by logic combining those features into semantic objects. Maybe some form of neural net with different types of neurons and different layers serving the individual detection tasks could be constructed for this.

- Humans have two eyes. Stereo vision is useful for 3D navigation, but is also used for detecting redundancies and suppressing noise in 2D images. Also, the focus area is in the field of view of both eyes giving extra detail and resolution, while the edges of the image are only ever in the field of view of one eye. This again hints at a non-cartesian image representation with detail fall-off towards the edges. Maybe sensors could be built or refitted to produce stereo vision images as well.

- Similar to the center of attention mentioned earlier humans seem to have something like locally dominant colors. We can only sense so many colors at once (a relatively small number) so we see certain areas as one homogenous color even if they are not. Similarily our ability to discertain colors directly depends on how many different colors are visible at once. It is easier to spot a smooth color gradient than to discertain shades of the same color in a chaotic patch of different colors. This could probably be modelled by a clustering algorithm (k-means, meanshift, self organizing maps, ...) in color space, finding the locally dominant colors.

- While computer image analysis usually only works on one image at a time humans naturally blend several impressions of the same visual stimulus over time. I don't know whether the human eye has a concept of "frame rate/frames per second" or whether that would be constant for all receptors or different (it would be possible for example, that the gray receptors have faster circuit times than the color receptors, thus giving light intensity higher time resolution). All this is important in analysing motion of course, but even for static images it makes sense to view it over time. A human seeing an image for some time will gradually notice more and more detail. Due to the imperfect nature of our sensors (eyes) still frames will have a certain amount of noise and inaccuracies in them. Interpolating over time will eliminate most of these. Also, it allows one to focus on different or more details once the basics have been analyzed. Putting all these factors together it seems humans have a hiararchic system over time, reducing noise and redundancies and analyzing in increasing amount of detail.

- Texture. Texture is an exceedingly complicated and overloaded term. There is not even a standard way to define what texture means. Yet at the same time humans have no trouble whatsoever recognizing objects of same texture. Why is that? Texture describes some spatial relationship between colors and intensities, a pattern. It is not necessarily exactly repeating (the texture of a plant leaf is always the same, yet at the same time it is unique for each leaf and from one square centimetre to the next). It is not necessarily of uniform scale. It is not easy to describe in simple geometric terms (clouds). It does not necessarily appear uniform in color or intensity. Nevertheless, there must be at least some characteristics appearing constant and non-varying to the human observer, so he can group parts of an image by their texture. Texture is a hierarchical term again. Viewed from a distance at coarse detail a patch of lawn will appear textured uniformely. Viewed in close detail the individual blades of gras will appear totally chaotic.
There are multitudes of approaches to capturing texture in mathematical terms, most of them deriving some form of statistical features from gray value intensities. There are fourier descriptors, wavelet transforms, coocurance matrices, color clusterings, bayesian predictors, neural networks, genetic algorithms, self organizing maps, active contours/snakes, ... Each one of these is a valid approach and fit at capturing one aspect of texture. However none of them is adequat for the general problem.



To sum up: obviously the human visual system is a highly sophisticated, highly dynamic and hierarchically organized system. It seems difficult if not impossible to model all of it's capabilities with only one algorithm. It seems more logical to start at the bottom of the hierarchy, with the simple or even trivial transformations and interpretations and work the way up. Ideally keeping the interfaces open such that at each step each algorithm is pluggable and replacable by a different, possibly better, one.
Another very important preprocessing step is determing what will be the basic inputs to the whole system. These inputs should be modelled as close to the outputs of the receptors in the human eye as possible.

I'm a big fan of emergent behaviour and believe a lot of nature's power comes from systems organized in such a way. Very basic and simple basic blocks in large quantities work together to archieve more than the sum of their parts. Examples for systems like these are all of a human's senses, each of which is built from tiny receptors, the human nervous system and brain and, on a larger scale, ant colonies. These systems exhibit a lot of very desirable properties: they scale well, they are robust to noise, they are robust to defects in individual agents, they do not have a single point of failure or a single mastermind, they are easily parallelizable and they are easy to construct (since each individual agent is trivial). Artificial examples for algorithms with these qualities would include neural nets, self organizing maps, autonomous agent systems (ants) and genetic algorithms. The idea is to build a hierarchy of lots of very simple building blocks and hope that something useful will evolve from that.

Ok, now I'll start working on my genetically evolving, self organizing, hierarchical neural network based on fuzzy logic maths - it will call you and tell you when it's done and ready to take over the world ;-)


Information about the eye and vision:
http://webvision.med.utah.edu/

Irregular grid sampling:
http://www.ee.surrey.ac.uk/Basic/

Perceptual image pyramids:
http://www.cs.ucla.edu/~siavosh/iccv05.pdf

Log-Polar image representation:
http://users.isr.ist.utl.pt/~alex/Projects/TemplateTracking/logpolar.htm

Locally dominant color adaptive segmentation:
http://www.ece.northwestern.edu/~pappas/papers/jqchen_tip05.pdf

Experimental Determination of Visual Color and Texture Statistics for Image Segmentation:
http://commnet.ece.northwestern.edu/~jqchen/docs/spie05.pdf

Human visual system foreground background detection:
http://www.apa.org/journals/releases/xge1312194.pdf

Stereo vision correspondence comparison:
http://cat.middlebury.edu/stereo

Comparison and benchmarking datasets for different segmentation strategies:
http://www.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/segbench

Autonomous agents and pheromone maps for digital image segmentation (this one is just soo cool ;-) ):
http://alfa.ist.utl.pt/~cvrm/staff/vramos/Vramos-WCLC05b.pdf

Problems with the CIE Lab color space:
http://www.aim-dtp.net/aim/evaluation/cie_lab/index.htm

plus a gazillion more in my bookmarks and littered all over the place. I hope I'll come around to organizing them.

I've just read on the google blog about a new project called Google Transit. It's a nice little feature that allows you to get local transit schedules and routes (displayed on google maps - go AJAX go! ;-) ) via a simple search engine like free text query. Cool. But even cooler: they include the price for the trip and compare it to the price of driving the same route by car. Cars lose again.

Here's an example query:
http://www.google.com/transit?q=pdx+to+100+nw+couch+st,+portland,+oregon&hl=en

... and Münster descends into chaos! Oh such fun - I love winter ;-) Anita and I went to a musical last night, Jeckyll and Hyde - awesome - go see it - which was 4 hours long. Coming out of the theatre all cars were burried in 20cm of fresh snow. Total chaos on the parking lot, no one could get out. Anita and I were on foot ;-)
Today, almost no trains are running (which is relevant for my brother who tries to get to Cologne to see Tracy Chapman in concert), 40km traffic jams on the Autobahns because trucks can't make the easiest of slopes, non stop emergency sirens because people are just too damn stupid to deal with snow, power outings, ... I cycled 6km through the snow, took me almost twice as long as usual but was great fun...

Go winter go!

BTW, a cool non-car transportation system (that'd probably work in winter): http://www.skywebexpress.com/

Just wanted to let you know that the new Edible-Rock CD music4earthlings is sent to the factory! Finally done! ;-) The official release party will be held on november the 27th, details to be announced on the website.
As an appetizer I've snatched two songs from the new album:
H.P.V.
(of course I had to have this one since I kinda inspired it with my bike ride ;-) )
and
Writings On The Wall

And of course you can already listen to the album and see the cover art on Last.fm.

you heard it here first!

I've said it before when the service was still called audioscrobbler: last.fm totally, absolutely, fantastically rocks!
It's another one of these up and coming social computing websites. You download a music player and optionally a plugin for your current music player of choice. The plugin will upload your musical choices from now on (if you allow it to) - thus creating a musical profile. From that info it generates a list of musical neighbours, a list of recommendations (people who shop on amazon should know them: "people who bought X were also interested in Y") and all kinds of custom playlists. But the real bummer is yet to come: via their free player you can listen to dynamically generated radio playlists all day long - specifically tailored to your tastes! If you don't like it you can just skip a song and go to the next. Or say I only want to listen to music similar to "Jack Johnson" (insert your favourite artist). I love it.

BTW, my profile and playlists are at: http://www.last.fm/user/BuschnicK/

Anita and I arrived home from Canada last monday. We had a total blast. We spent 7 weeks driving, canoeing and hiking around in the Great Outdoors that is British Columbia.

I have updated my Amazon associate links on my bookreview pages. If you happen to be a regular shopper on Amazon why don't you use this link instead of the regular Amazon one. Thank you.

I'm currently loking for more software development contract work, so if you have any work/tips for me please contact me.

Spent two days cycling along the Ems river to the North Sea. Visiting my girl friend at her parents house. 265 kilometres, 115 the first day, 150 the second. I camped on a clearing near Meppen and woke up at exactly midnight to a fantastic lightning storm. Almost windstill, but continuous thunder and ligthning. My tent was bright for seconds on end. Nature at her finest. While cycling I only had the best weather though. I still have the sun tan line on my legs from my 6 months bicycle tour a year ago. Because of that I didn't have to use any sun screen this time and am back to my old dark "cyclist skin" ;-)

It seems like my job will soon be a bit of C++ again - yeah! I had a couple of interesting discussions and language lawyer puzzles recently. Fun, fun, fun. Virtual inheritance, dominating methods, ... ahh the complexities and subtleties of the language never cease to amaze and intrigue me - I'm a whacko I guess ;-)

I've started packing for Anita and my Canada trip this year. Plane leaves on Sunday...

I used to post my book reviews on Amazon.com and received some very positive feedback on them too (almost 100% helpful votes and one author even published my review on his website). Unfortunately my original Amazon account got lost and I don't know how to continue with a new account under the old name. I got totally lost in the Amazon html maze trying to figure that out. Ah well. Starting afresh then. Anyways, here is my latest review:


Coder To Developer
Tools and Strategies for Delivering Your Software

Mike Gunderloy
ISBN: 078214327X

The book tries to cover the whole of the software development process from planning, team management, coding best practices and finally creating an installer and releasing the product. With such a broad range of topics each one is only treated very superfluosly and shallow. The author has a very tool centric view on things and as such many chapters are just a market overview of available software for the task at hand. I don't think this is of much use for the reader since that is exactly the kind of information you can gather in half an hour of internet research with google - and even after reading the book you'd still have to do this research anyways in order to gather current prices for the latest gadgets. All software solutions presented in the book are for windows only and Microsoft's tools seem to get extra focus and attention. The intended target audience for the book are independant developers and small software shops. As such the author assumes that you are wearing multiple hats and are filling all kinds of different roles from designer to coder to management. I very much liked this perspective on the software world because for one thing I am one such lone wolf developer and second because there are already tons of software books for the large corporate software developer. Those books typically assume loads of process and management and different departments etc which all don't apply for the single developer. Two important things missing in the book's coverage are two chapters: One for the time before concrete planning actually begins on the question of "what to develop" and determining markets. Another one for the other end of the road on how to market your software, how to price it and how to present and distribute it. If those were included I think the book would truly cover the complete process a lone developer goes through from idea to product.
All in all, the book gave me little new information but a good checklist to work through on a project.

Take a look at this guy. He has left a sprinkler running all through an Alaska winter. The result is a ~30m high artificial tower of ice he uses for climbing. Humans need hobbies I guess ;-) It's worth reading the article from the beginning, the writing is hilarious and witty.

In my arrogant moments I like to think I'm a free electron... but then reality usually catches up ;-)

My brother's band won the first round in the international Emergenza newcomer festival. Congratulations! Now on to the next couple of rounds ;-)

The festival covers the US, Canada and Europe. There are several qualification rounds, the first couple are decided by audience vote alone, while the next few will have a jury. Tonight 8 bands were competing and Edible Rock won first place with 80 votes. Second was Fortitude with 56. I believe in Germany alone 2500 bands have signed up for the competition...

If you haven't been there yet, head straight over to the Edible Rock homepage and download some cool music.

...or how big money has won over Democracy yet again. For the really optimistic there's still a glimmer of hope left, but seriously...

http://www.groklaw.net/article.php?story=20050307095336843

Finally we get some real snow... I love it.

I've just read that the night from sunday to monday was the coldest this winter with temperatures just above -14 degrees celcius. Funnily enough we heated our outdoor bathtub that night. So I was lying around naked in the snow - not feeling cold at all ;-)

Debugging society - part one:

Thoughts on free markets and a capitalist economy

I believe our current economy and capitalist system to be deeply flawed, unjust and ultimately doomed. Let me state a couple of the reasons why:

It's flawed because it encourages all the wrong mechanisms and motivations. Free market advocates quote the fundamental mechanism of supply and demand and competition as the driving forces that keep the system healthy and alive.

Competition is what I call a destructive motivator. Meaning that to thrive in a competitive market it is to your advantage if your competitors (opponents?) struggle. This is not a healthy situation, not productive for society as a whole. Cooperation on the other hand is a constructive motivator. It is good for you if your partners thrive. If competition is the only viable motivator (as seems to be assumed) how can it be that even in a capitalist system the greatest leaps in productivity come when people cooperate? Rebuilding after a war, causing ?economic miracles?? Building up to a war? Why, if competition is the only viable option, do we expect friends and family to work by wholly different and opposite principles?

Supply and demand is a broken system as well. There are lots of things that are in huge demand and are not in the hands of capitalism to supply (rightly so). Think about our natural resources like air and water. The demand is huge and never-ending. Yet at the same time they have no monetary value attached to them. Which means it?s not, cannot be, in a capitalist?s interest to invest in, or even just to conserve them. But the alternative also isn?t possible. We cannot attach a price value to natural resources as that?d mean poor folks would only get three breaths a day or something?

Another key observation is an individual worker's productivity. In previous centuries, when work mainly consisted of manual labour, all workers were pretty much equal. Sure, if you put in twice as many hours as your neighbour or are twice as strong as him you might manage to work twice the land he does. But basic physical principles prevent huge differences in work productivity. Not any longer. Technology has changed all that. A farmer using modern machinery can easily be a hundred times as productive as one that doesn't. A logger with modern equipment cuts down a whole forest in the same time even Hercules needs to cut down a single tree with his axe. A good software developer is light-years ahead of a bad one. This increase in an individual's productivity is a good thing. It is enabling. Only because humanity invents ever more powerful tools do we progress and are able to sustain our numbers. And yet, capitalism punishes these advances. Oh sure, the one productive farmer is rewarded for his work. Yet the 99 others that are now without work are now without work- read: unemployed. This too, should actually be a good thing, because they are now free to pursue other venues. Yet their choice of alternative occupations is severely limited by what the market supports. Unemployment is only bad because it is made so. Think about it for a second: How often did you say ?If I only had more free time I could??. Unemployment is the ultimate in free time. Granted you need a system in place to keep the 1 worker that supports the 99 unemployed happy and working. The current system?s answer is to punish the 99 and make them all compete for the 1 worker slot. Is there really no alternative? More on this in another post (as it is a topic that interests me very much because my job as a software developer is basically making other people?s jobs superfluous).

It?s unfair because the old saying of money breeds money is just all too true. I think I don?t even have to argue this point much. Once you have a million it?s easy to make the next. Even if you didn?t even deserve (earned with your own labour) the first to begin with. It?s unfair because it actively sustains and supports the current global imbalance. It?ll never be in our best monetary interest to share our resources with the poor, yet it is in our interest to use their labour. It?s unfair because work is valued crassly to the advantage of the already rich. The further away your job is from work that is necessary, from work that actually must be done, the more money you?ll get for it. Imagine that! The more needless and dispensable your job is the more money you?ll get for it. Someone working the fields until his fingers bleed every day of his life just to support his family doesn?t even compare to someone working the stock market 20 years of his life and retiring early. Someone selling lottery tickets earns more than the local bakery? I?m lucky; I?m pretty much at the top of this chain. I enjoy a very high luxury standard of living for typing stuff on a keyboard each day without any real risk of starving or even losing my standards. But is this fair? Should this be so?

It's doomed because of all the reasons stated above (individual productivity increasing through technology, money piling in fewer and fewer places) and because it?s already failing in a downward spiral. Take a look at the world today. How many are we? 6.000.000.000? Common wisdom has us believe that free market capitalism is the best option, the survivor, the winner, the dominant. I?d go so far to even argue that point. Seriously, of those 6.000.000.000 people ? how many do even participate in capitalism? How many of them have more than a dollar a day at their disposal? Less than that and you do not really participate. I bet if you view it that way capitalism doesn?t come out all that well. And if I had to guess again I?d say the group that?s not actively participating in, but just being controlled by, capitalism is increasing faster than the other one.

Disclaimer: I don?t have the slightest clue about what I?m talking here. These are all just personal observations and first stabs at refining them into some more or less coherent arguments. If you agree/disagree/have strong opinions on the topic yourself ? feel invited to cooperate and shape my views ;-)

First update in a long time. I have been very busy recently, working two contract jobs, hunting for a new appartment, helping a friend with his move to Stuttgart, and getting all the insurance and tax paperwork done for my independant work.

It's fun learning new stuff! My job used to be programming C++ pretty exclusively. I'm still doing C++ work, but my primary tools at the moment are Java, VisualBasic, PHP, MySQL and Halcon Script - with a major project in each. PHP and MySQL are an insanely productive combination in their problem domain and real fun to work with. Java feels like a toy language coming from C++ - it's similar enough so you are able to dive right in and hack away and forgiving enough to let you get away with it. The refactoring tools available for it are powerful and easy to use. I don't like it's exception handling and object lifetime semantics though. VisualBasic and HalconScript are clumsy and basically a huge PITA to work with. Dunno why BASIC ever got as popular as it is.

Working independantly takes getting used to. For one thing I find working from my home office more exhausting than sitting at the company's desk. Maybe it's just me, but when working from home I only bill actual working hours. For an eight hour work day that feels like taking an eight hour exam and I'm completely wasted afterwards. Working at the company's office on the other hand you seldom if ever get eight hours of high concentration work done on any single day. Colleagues, phones, meetings etc interrupt often enough that the typical eight hour day is more like five or so...

Cool new website find of the month: spurl
A page that manages your bookmarks and finds related links and information. I wonder when the ultimate-uber-search engine will emerge. A mix of technorati, google and the likes. I wanna have my own customizable web spiders. Continuously monitoring and crawling the web for me and presenting relevant results. I'm already using a baby version of a service like that, which lets you craft a search and updates you via email or rss when something relevant to your search happens. But it's not good enough.

While toying around with research for various game ideas I stumbled across the Nasa pages, especially the Blue Marble and World Wind projects. The Blue Marble is an incredibly detailed set of texture maps of the earth. Stitched together from thousands of satellite images, color corrected and deskewed it's the result of obscene amounts of work I'm sure ;-) Available free of charge.
World Wind is an open source project that uses the Blue Marble texture as a basemap for a full 3D rotating globe. But it doesn't stop there: Through this 3D globe interface you have access to 20TB (that's terabyte as in 1000 Gigabytes!) of satellite imagery and data. Meaning you can zoom way in and the program will download missing datasets from the central Nasa servers. Combined with live weather data, height maps, place names and video overlays it's a very powerful and fun tool. Big brother is watching you!

An article on MSNBC about coming home for Christmas and fix your parents' computers. All too true. There's something seriously wrong with technology... Apple is on the right track fixing the usability problems, I have to give them that - and I hate the company.

Anyway, I myself have been wasting hours toying around with technology lately. Set up my own webserver and remote shell access to my computer at home. The whole setup was up and running relatively soon, though I didn't realize it because my stupid router doesn't support loopback. Meaning the webserver was working properly as expected but I wouldn't know because it wasn't reachable under it's domain name from inside my LAN. Solution? Either buy a new router or do as I do now: use a free proxy server. Get this: I'm accessing the web server running on my computer through a proxy server in southern Thailand. sick
Why am I doing this you wonder? Well... wait and see ;-)

according to "The Sect of Homokaasu" website, which, incidentially, is also the home of the "Kill everyone project". Quite tasteless actually - but then again, I like that kind of humor and killed a couple myself };->

Because of that total harddisc crash (a RAID system no less!) while I have been away cycling all my mp3 music collection was lost. I'm currently in the process of re-ripping all the CDs. I'm currently at ~60GBs again. This time I've switched to Ogg Vorbis as my audio format. Unlike mp3 it's unemcumbered by software patents while at the same time offering better audio quality with better compression rates. It's not as ubiquitous as mp3s yet, but it's getting there with some hardware players (DvD players, portable players etc) already supporting it.

During the ripping and cataloguing of all those CDs I've stumbled across these two very cool websites:
Musicplasma is a very artsy website that lets you visually navigate through a network of bands where close proximity between two nodes indicates musical similarity. (read: if you enter your favourite band as a starting point you'll get bands that are musically alike)
AudioScrobbler comes as a plugin for your favourite music player and monitors your playing choices. What you get in return is a list of "musical neighbours" of people also using the service with similar musical tastes like your own. Kinda like Amazons "people who bought X also bought Y" service - only better ;-) It also produces a list of recommondations after you've used it for a while. Very good! And it's also open, meaning everyone has access to their database and code.

Firefox and AdBlock with one of these filters. This gives you the bare minimum of a reasonable secure, fully featured browser with efficient popup and advertisement blocking.
Next you want Thunderbird for your email, newsgroup and RSS feed needs. Forget about it's built in spam filtering though and head straight over to spambayes. This is the single best personal anti spam solution I've tested.
Of course all this is pointless if you are running on one of the "swiss cheese" OSes, so as a Windows user at least upgrade to XPsp2. It's built in one way firewall plus your routers firewall protection plus a software firewall like Kerio should keep most intruders out of the system.
You also want a couple of utility web sites, two particularly useful ones are bugmenot, which bypasses annoying but mandatory one time registrations, and mailinator for the times bugmenot fails and you actually have to give out an email address.
For completeness and because everyone thinks you need one include AntiVir and Avg as your antivirus solution. I just do a scheduled nightly scan with these and turn off their realtime protection features because I find them terribly annoying and slowing down my system (life file system watch while compiling anyone?). They find at least a dozen virii and trojans per day but all of these sitting harmlessly in my spam email attachment folder and getting deleted no fuss.
There are more tools I use and recommend, but these should give you an enjoyable online experience. Also, while this is a white-list of things I recommend, the black-list of things not to install is far far far longer. As a rule of thumb I try to stick with free open source software wherever possible and am very careful around file sharing tools and software from "evil" companies that try to lock you into their philosophy and try to talk you into installing far more than you originally asked for.
Most important of all tools: common sense, a brain and the ability and willingness to read. It amazes me time and time again just how fucking stupid some people seem to be falling for the most naiive of traps and scams.

BTW, I wonder how long the war between spammers, advertisers and users can continue the way it does currently. I remember back in the days I signed a petition for a commercial free internet, which would be absolutely laughable today. Today's reality has half of the internet financed by advertisements. Which puts me in a sweet spot because with my tools I don't see any of the annoying website banners and almost no spam mails at all. At the same time Joe average user isn't interested or educated enough to defend himself against the flood and thus pays for the ad sponsored services I'm benefiting from as well. Thank you Joe! I wouldn't want to live without my free search engines, free emails accounts, free message boards and blog services...
Now imagine the anti ad tools getting better, easier to use and ubiquitous. Suddenly ads as a revenue stream and financial model are gone. Oh oh.

the german magazine Der Spiegel runs an article about US Americans pretending to be Canadians because they are ashamed of their own country's politics and don't wanna face embarassing questions about Bush when travelling. I thought that was really funny - and sad really. Here's the article. For english readers the same from the times .
How did that graphical joke about Canada and the US go? A map of both countries and Canada saying: "We are bigger and we are on top - in prison you'd be our bitch."

BuschnicK, proud to be Canadian ;-)

ey

...our home network administration with a meager 5 PCs drives me crazy at times. Now compare that with >50.000 consumer PCs running under heavy load 24/7. Ouch! I don't wanna know.

...at innergeek and scored 37.27811% which makes me a major geek. Guess I'll have to be - otherwise I wouldn't be proud of the result and trying to improve on it - or would I? ;-)

Read an interesting article about the development of Audion.

Got my credentials from my previous employer and am currently looking for a new job. In the meanwhile I'm doing Java contract work for a small start-up. Sabre-Security produces binary debugging, audit and verification tools which is a whole new field for me. It's also a whole new perspective on software and bugs: usually I'm the "try to find the bug by reading the code" type instead of the "start the debugger and see what happens" type. Of course when you don't even have access to the code my approach doesn't work any more...

Jumped on the FireFox bandwaggon. Like it so far. Especially impressed by the easy migration process from internet explorer. Tabbed browsing is cool. Adblock is waaaaay cool. Was using the proxomitron for ad filtering before which is a PITA in comparison.

Went climbing tonight. My fingers didn't hurt too much, so I hope their condition is improving (my knuckles were the reason I ended my bicycle tour early).

James from the States is visiting us at the moment - hi James ;-)

I have a couple of free accounts for the new google mail service (www.gmail.com) currently in beta to give away. If you are interested in one drop me a line.

I'm curiously awaiting the reactions and success of google's take on email. 1GB of free storage and powerful search functions (you'd expect that from google - right?). No folders to sort your mails into but labels instead (you can tag mails, which, in a way, is more powerful than fo