Friday, June 29, 2007

Recent progress in wet color perception research

The June 15 issue of Science has tow very interesting papers describing new results in perception. Both group of researchers studied the synchronization of gamma waves to investigate how various regions of the brain communicate. Their research indicates that top-down signals between brain regions regulate the flow of information, and that a distributed neural networks that use oscillatory dynamics support a broad spectrum of neural processing and behavior.

You may want to start by reading the introductory overview article Neural Networks Debunk Phrenology, by Robert T. Knight on page 1578. You may want to read second the earlier related paper Top-Down Versus Bottom-Up Control of Attention in the Prefrontal and Posterior Parietal Cortices, by Timothy J. Buschman and Earl K. Miller that was published in the 30 March issue on p. 1860.

In Modulation of Neuronal Interactions Through Neuronal Synchronization, Thilo Womelsdorf, Jan-Mathijs Schoffelen, Robert Oostenveld, Wolf Singer, Robert Desimone, Andreas K. Engel, and Pascal Fries recorded multiunit activity and local field potentials simultaneously from four to eight electrodes while the neurons were visually stimulated with moving gratings. They tested the hypothesis that the phase relation between the rhythmic activity of groups of neurons determines the strength of their mutual influence and found a heavy dependence of the correlation on phase. These results suggest a mechanism by which signals are matched and coupled during complex perceptual and cognitive operations. The fact that in the V4 area the synchrony occurs at 60 Hz, which is higher than the frequency ranges for the various attention functions, color identification is relatively effortless.

In Neural Mechanisms of Visual Attention: How Top-Down Feedback Highlights Relevant Locations, Yuri B. Saalmann, Ivan N. Pigarev, and Trichur R. Vidyasagar simultaneously recorded from two areas in the dorsal stream of the visual pathway, the posterior the lateral intraparietal area (LIP, which is critical for spatial attention) in the parietal cortex and the medial temporal area (MT, which plays a role in the perception of motion), while subjects performed a delayed match-to-sample memory task. Activity in LIP predicted activity in MT when the receptive fields of the LIP and MT neurons were in the same place and when the monkey was attending to that place. LIP feedback can thus account for attention-enhanced MT responses, in simpler words, parietal neurons may thus selectively increase activity in earlier sensory areas to enable focused spatial attention.

Tuesday, June 26, 2007

No pictures this Sunday

This is an administrative post. Due to a power shutdown on Sunday, July 1 in our building, you will not be able to see the pictures, graphs, formulae, etc. on my blog between 6:00am and 6:00pm PDT.

Sunday, June 24, 2007

Chaotic light sources

My previous post was on light sources, and I hope it cleared up the difference between the various sources. Today I am opening Pandora's box: chaotic light sources. Let us see what the experts have to say about this subject:

On page 674 of Optical coherence and quantum optics, Mandel and Wolf write:

We shall call radiation that is derivable from blackbody radiation by any linear filtering process thermal radiation. It has sometimes also been called chaotic radiation.

On page 31 of A guide to experiments in quantum optics, Bachor and Ralph instead write:

In most practical cases classical light will be noisy. Only a perfect oscillator would emit an electro-magnetic wave with perfectly constant amplitude and phase. There are some relatively simple models of realistic light sources. Chaotic light is the idealized approximation for the light generated by independent sources emitting resonance radiation. A practical example is a spectral lamp. In contrast, a thermal light is an approximation of the light emitted by many interacting atoms that are thermally excited and together emit a broad, non resonant spectrum of light. A practical example is a hot, glowing filament.

On page 109 (among others) of Quantum optics—An introduction, Mark Fox finally writes:

The light emitted by a mercury lamp originates from many different atoms. This leads to fluctuations in the light intensity on time-scales comparable to the coherence time. These light intensity fluctuations originate from fluctuations in the number of atoms emitting at a given time, and also from jumps and discontinuities in the phase emitted by the individual atoms. The partially coherent light emitted from such a source is called chaotic to emphasize the underlying randomness of the emission process at the microscopic level.

If we take a vote, then electric discharge lamps (mercury vapor) are chaotic light sources. But now consider this quote from the paper Experimental study of the momentum correlation of a pseudothermal field in the photon-counting regime by Scarcelli, Valencia, and Shih at the University of Maryland:

In principle the term "thermal radiation" should refer only to radiation coming from a blackbody in thermal equilibrium at some temperature T. But in reality, some characteristics of true thermal fields, like the extreme shortness of their coherence time, have imposed serious obstacles to their use in actual experiments, and therefore since the early days of quantum optics there has been a great interest in the realization of more utilizable sources that could simulate the behavior of true thermal fields (gas discharge lamps, randomized lasers, etc.). We usually describe this kind of source as pseudothermal; they are actually all chaotic light sources that can be modeled as a collection of independent atoms emitting radiation randomly […]. The principle of the generation is very simple: coherent laser radiation is focused on a rotating ground glass disk so the scattered radiation is chaotic with a Gaussian spectrum.

So here a electric discharge lamp (e.g., mercury vapor) is pseudothermal. If laser light (which is Poisson) is focused on a rotating ground glass disk it produces a chaotic source with a Gaussian spectrum — very different from the line spectrum of a mercury lamp as plotted below. I do not know about you, but I am confused. Are "gas discharge lamps" and "electric discharge lamps" the same thing? What is a chaotic light source? What is the role of coherence length?

If you can shine some light on this in form of a comment, I would appreciate it very much.

mercury vapor lamp

PS: Links cited in the comments:

Friday, June 22, 2007

Hot body, excited particles, and the north sky

At the beginning of color perception there is radiant energy. Because its treatment in color science is slightly different from what we learned in high school physics — it can be limited to the visible domain — I give a brief overview of the sources of radiant energy of interest to CIE (Commission Internationale de l'Éclairage) colorimetry. The material is excerpted from the Wyszecki and Stiles book.

The energy of a photon is Planck's constant times the photon's frequency. In color science we are interested in the radiometric quantity's spectral distribution in terms of that quantity per unit wavelength interval. By definition, a photon's frequency times its wavelength is the speed of light, which allows us to go from the formulas based on frequency common in physics, to formulas based on wavelength as we like them in color science.

For us humans, the most important natural source of radiant energy is the sun. The components of daylight that play a key role in color science are direct sunlight and sunlight scattered by the atmosphere. Consequently, when we talk about daylight, we must state whether it refers to direct sunlight, scattered light, or a combination of the two. The plot below shows the spectral irradiance for the solar disk outside the atmosphere (blue) and at the earth's surface at air mass 2 (violet).

solar irradiance

The air mass is defined as the ratio between length of path within the atmosphere for rays from the sun to the observation point on earth, to the corresponding length of path for a hypothetical position of the sun at the zenith. Therefore, air mass 2 means a sun altitude of 30º.

The CIE then recommends a calculation of the spectral radiant power distribution given the chromaticity coordinates of a particular phase of daylight. These chromaticity coordinates cluster around a curve called daylight locus, which runs slightly above and parallel the Planckian locus in the CIE chromaticity diagram. The Planckian locus is the curve generated by the chromaticities of Planckian (blackbody) radiators operating at different absolute temperatures measured in degrees Kelvin (K).

Planck’s law gives the spectral concentration of radiance of a Planckian radiator as a function of wavelength and temperature:

Planck's Law

Daylight from the sun plus the total sky, ranges in a correlated color temperature (CCT) from 5,000ºK to 7,000ºK, regardless of cloud coverage. Daylight from the north sky (i.e., from the total sky but with the sun occluded) has a CCT above 7,000ºK. Daylight from the sun disk alone or from the sun disk at low altitudes plus sky generally, corresponds to a CCT below 5,000ºK.

This is a big difficulty in designing digital camera software that handles correctly sunset pictures, because it has two handle two radically different light sources in the same image: the north sky and the sun disk.

A thermal radiator (hot body) capable of providing a spectrum dependent on the temperature alone is called a full radiator, Planckian radiator, blackbody radiator, or cavity radiator. Any surface that emits radiant energy identical in all respects with that from a small aperture in a constant temperature enclosure is also described as a blackbody radiator and is characterized similarly by one parameter, the temperature T.

The radiant energy emitted by an incandescent solid is always less that that emitted by a full radiator at the same temperature, and the ratio of thermal excitance of a hot body to that of a blackbody radiator at the same temperature is called the emissivity.

In the visible range, bodies like carbon, platinum, and tungsten have spectral power distributions approximating that of blackbodies, but the emissivity is not constant. A considerably better light output can be obtained by adding a small quantity of halogen to the filling gas. Halogens are negative elements like bromide, chlorine, fluorine, and iodine.

A completely different source of radiant energy are electric discharge lamps. Here usually the radiant energy is produced by passing an electric current through a gas or vapor. Accelerated electrons collide with the gas molecules yielding excited molecules and atoms, ions and more electrons. When the excited particles return to their normal states, the excitation energy is emitted in the form of quanta of radiant energy, i.e., photons.

In contrast to the hot bodies, these excited particles yield line spectra. Therefore, their light quality is quite different from that of natural light. Examples of discharge lamps are mercury vapor lamps, fluorescent lamps, high-pressure xenon arc lamps, and flash tubes. The figure below (data courtesy of Dr. Dmitri Boiko) shows the spectral distribution of an Ocean Optics CAL-2000 mercury lamp. On the abscissa is the mercury part of the spectrum and on the ordinate are photon counts.

CAL-2000 mercury lamp

To obtain a better light quality, the mercury vapor and fluorescent lamps are coated with layers of "fluorescent" materials (phosphors), which absorb the half of radiant power emitted in the ultraviolet domain and emit it in a continuous spectrum in the visible domain, especially when activators are added.

Today's ubiquity of display projectors has created a large market for mercury vapor lamps, from which a large cost reduction and quality improvement has ensued. The figure below (data courtesy of Jeffrey DiCarlo, Ph.D.) shows the spectral distribution of a modern halogen discharge high pressure mercury lamp. Such a lamp has a typical luminous flux of 2,600 lm, a luminous efficacy (burner) of 55 lm/W, an average luminance of 100,000 cd/m2, and a correlated color temperature of 5,800º K. Note that the axes are different from those in the previous plot.

halogen discharge high pressure mercury lamp

In colorimetry, we do not want to bother with a myriad of actual lamps. Rather, we would like to restrict ourselves to a few specific and well-defined spectral distributions of radiant power incident on the sample under study. The CIE recommends a set of such spectral radiant power distributions called the CIE standard illuminants.

In practice we need only the relative spectral radiant power distributions. The curves are normalized to 100 at 560 nm. The figure below shows to important illuminants: CIE standard illuminants A and D65. Illuminant A represents light from the full radiator at 2,856º K. Illuminant D65 represents a phase of natural daylight with a CCT of approximately 6,504º K.

CIE standard illuminants A and D65

Thursday, June 21, 2007

Save a hill in West Virginia

Whether or not you believe in global warming, Google Earth can quickly take you for example to Kayford Mountain, West Virginia where you can see for yourself the devastation of Mountaintop removal mining (MTR). Here is what you can do.

Mr. Hewlett and Mr. Packard have always been lovers of the outdoors and HP gives you advice on how you can conserve energy and help protect the earth when you use our products. The day I am writing this, HP has joined the Climate Savers Computing Initiative, which within two and a half years will cut greenhouse gas emissions in an amount equal to shutting down twenty 500-megawatt coal-fired power plants. Follow the above two links to learn what you can do to use less power with your computer system.

If you design computer devices, make sure you specify a switched power supply with power factor correction. Power factor correction adds only little to your power supply cost. If you do not know how to do it, there are excellent engineering support services here in the Silicon Valley, like for example Power Factor 1.

If you design software, make sure your code is efficient, because color algorithms can be processor hogs. In modern computers, processors are dynamically clocked down when the full power is not required. Optimize your code and give your customers a quieter computer with longer battery life … and help save a hill in West Virginia.

Kayford Mountain, West Virginia

Kayford Mountain, West Virginia. Photo by Jeffrey Gerard.

Tuesday, June 19, 2007

Art and color

The study of color is pursued in three communities of interest: basic and applied color research, industrial applications of color, and art, and design and psychology. These three communities are not complementary, on the contrary, they are heavily intertwined. Today, I would like to take an artist's point of view.

The earliest pictorial art forms were probably of liturgical nature. The earliest reference to color — more precisely to the relations among colors — can be found in the Upanishads, which go back to the eighth century BCE. In esoteric Buddhism (Tantra, Shingon), the color palette consists of five colors, like there are five basic language sounds. The actual colors in the palette may change, possibly depending on the availability of pigments, but the number is fixed. Colors have specific esoteric meanings and thus are used depending on the intended function.

The first intertwining of the color disciplines occurred in the 15th century CE during the Renaissance, when Leonardo da Vinci first applied technology to achieve color fidelity instead of using color symbolically. He used colored pieces of glass (filters) to determine color mixtures, anticipating spectral color reproduction. He also introduced the concept of color perception, color order systems, the fact that black and white are colors (chiaroscuro), that there are three pairs of opponent colors (black–white, red–green, yellow–blue), and simultaneous contrast. Of Leonardo's works I have seen in their original, the most stunning one from a color point of view is the Last Supper at Maria delle Grazie on Corso Magenta in Milano after the recent restoration.

opponent colors

The French chemist Michel Eugène Chevreul, in his role of restoring the French Gobelins tapestries, rediscovered simultaneous contrast and introduced also the concepts of color theory and of visual color mixtures. His work had a huge influence on Eugène Delacroix and sparked impressionism, neo-impressionism, and orphic cubism.

Although chemists compiled influential color order systems, most notably Chevreuil (color hemisphere) and Friedrich Wilhelm Ostwald, it was the painter Albert Henry Munsell whose Color Tree had the most important impact in color science, in the form of the NBS (now NIST) Munsell Renotation.

The artist who most laboured to find mathematical formulaæ modeling color æsthetics was the Belgian sculptor Georges Vantongerloo (see for example his 1935 sculpture métal: y = ax3 - bx2 + cx (construction: y = 2x3 - 13.52 + 21x)). Under the influence of De Stijl movement, he branched into painting trying to find formulæ that generate æsthetically pleasing color sequences. Vantongerloo gave up in frustration and returned to sculpture, but when I saw his Spring 1981 painting exhibition at the Kunsthaus Zürich, if found his work of an ethereal beauty. In may view, his use of color is strongly reflected in Joan Miró's work.

The last artist I want to mention is Max Bill, because he gave us the contemporary definition of abstract and concrete art, the former referring to an abstraction of nature and the latter referring to form and color standing by themselves, like in his Verdichtung zu caput mortuum. I hope to read the remainder of this topic in the comments below.


In case I suddenly disappear, good bye. As you have noticed, our blog system was running only halfway for a while. Our access system, which hides your identity from my curious eyes, was upgraded into a form that was incompatible with our blog editing and commenting software. Therefore, we could not access it ourselves.

Because the blog system has been sunsetted by IT, next time there is a glitch we may not be back, so I am saying good bye now and thank you for reading this blog, and last but not least, thank you for your many comments, which were my guide and inspiration for subsequent posts.

As you may have read in the press, on May 1st Shane Robison, HP executive vice president and chief strategy and technology officer, took over the responsibilities of HP's corporate marketing activities. HP Labs is also under Mr. Robison, so after the blogs are gone, you may want to refer to this link to get the latest news about what is cooking here in Labs and how we are thinking: Of course, the official communication will remain available at Labs' site.

Wednesday, June 6, 2007

Mini review. A Scientist's Guide to Talking with the Media

Science is a dialogue, and blogging is a newer dialogue technology fostering discussion in a open social network. For us scientists, blogs offer the opportunity to include the lay world in our discussions and consequently to improve society at large. This recent book published by Rutgers University Press can help you overcome any writer's blocks you may have.

I admit I am not a great communicator myself, and I have even been given the characterisation of being charismatically challenged. However, I make an effort to share my knowledge, because this sharing costs very little but has a big impact. As the Romans used to say about sharing knowledge, when a wayfarer lights the torch of a fellow wayfarer, the latter's torch shines without reducing the brilliance of the former's torch.

This is why I try to spark a dialogue in my blog posts. To overcome my shortcomings as a communicator, I try to stimulate you my reader and fellow wayfarer with controversial tiles like "Sex and Evolution" or "The End of JPEG." I also try to put something extra in the post, something that is surprising, catchy, humorous, or sobering.

Sometimes this works, and the post on "Non-Local Realism" sparked a serious discussion with 13 comments. However, most of the time it does not work, like in my last post on "Childhood Origins of Adult Resistance to Science." I was hoping that at least my fellow American scientists reading this blog would react to being kicked in their shins, alas the patient remained catatonic. No wonder, American's views of global warming equal those in underdeveloped countries. Do you remember how in the Sixties we were stimulated to outbrain the Commies? We did not, the cold war was won by outspending them in the arms race; therefore, our task is not finished.

Maybe a reason for this silence lies in the educational system, where we learn to write erudite papers for our fellow scientists, but not to communicate with humans at large. If this is the case for you, then I have some good medicine, a little book from Rutgers University Press called A Scientist's Guide to Talking with the Media.

This book was not written by scientists, but by Richard Hayes, a media and public relations specialist, and Daniel Grossman, a science journalist. They teach you, my reader, how the world functions on the other side of this screen and help you have more impact on society at large. And in the process I hope to get more comments from you to my posts.

Monday, June 4, 2007

Childhood origins of adult resistance to science

I recently posted an entry on non-local realism that sparked quite a discussion. Last week’s Science Magazine had an interesting review article bearing the above title childhood origins of adult resistance to science (subscription required), analyzing similar issues in neuroscience and evolutionary biology.

When I travel to Europe, people often ask me about certain controversies in American society, because they are not able to see the controversy. Mostly I have to pass, because I am not able to understand the controversy myself, for example the controversy about evolution and creationism.

The Science article has a two-part explanation. The first part is that resistance to certain concepts (like gravity and non-locality) has its origin in childhood, or more precisely, when an incorrect common-sense assumption is not debunked in children, it will persist in adulthood.

When I grew up, the school system in southern Switzerland was an amalgam of the renaissance system, Cartesian dualism, Bourbaki’s methodology, and Derrida's deconstructivism. Intuition was always challenged and, like my buddies, I grew up as a very critical person. By comparison, the American school system is less structured, which — according to the Science article — leads to 42% of Americans not grasping evolution, and other beliefs that look weird from a Swiss perspective, like ESP and the more widespread reliance on astrology.

The second part of the explanation is that both children and adults are sensitive to the trustworthiness of the source of information. Here in America the role of trusted religious and political authorities is much more important. I wonder what the consequences are for American politician’s propensity of giving a spin on events, of foregoing debate in favor of sound bites.

I am interested in hearing the opinion on this Science Magazine article from people who went through the American educational system. I am also curious to hear how this looks from the perspective of other cultures. For example, ‘horizontal reading’ of the 'world-text' in the cosmology of Japanese esoteric Buddhism has deep similarities with deconstruction. How does this impact resistance to science in your culture?