Monday, February 5, 2007

EI wrap-up

Yesterday afternoon I drove the last VIP to the airport and also for this year EI is over—the event that is, for the new collaborations and insights sparked by the symposium will keep us busy over the next year. The main benefit of conferences and symposia are not the papers per se, those will be available online from the SPIE Digital Library in a few weeks, using the program.

Rather, the benefits are the networking with other scientists involved in similar research, talking to authors to find out why their research took a certain twist, what was tried and did not work. Cases in point are the papers by John McCann of McCann Imaging.

Almost every year John presents us a new research result in a very lucid presentation, unraveling the result step by step in a clear and logical sequence. Yet at the end of the presentation we ask ourselves: what has hit us? John just presented a totally non-obvious new deep insight, but what does it really mean? Only by sequestering John for a lunch or dinner and discussing with others, one has a chance to really grasp the impact of his work, which he has developed during many hours of hard labor and then distilled into a 20 minute presentation.

Of the three papers he presented, the one on veiling glare he wrote with Alessandro Rizzi was the most surprising one for me. Essentially it suggests that the human visual system has evolved its Retinex circuitry to compensate the eye’s strong veiling glare. The practical impact is that it teaches us how to deal with high dynamic range scenes. Since John and Alessandro put a lot of effort writing their abstract, I will just reproduce it verbatim instead of writing it in my own words at the risk of getting it wrong.

Veiling glare: the dynamic range limit of HDR images. High Dynamic Range (HDR) images are superior to conventional images. However, veiling glare is a physical limit to HDR image acquisition and display. We performed camera calibration experiments using a single test target with 40 luminance patches covering a luminance range of 18,619:1. Veiling glare is a scenedependent physical limit of the camera and the lens. Multiple exposures cannot accurately reconstruct scene luminances beyondthe veiling glare limit. Human observer experiments, using the same targets, showed that image-dependent intraocular scatter changes identical display luminances into different retinal luminances. Vision’s contrast mechanism further distorts any correlation of scene luminance and appearance. There must be reasons, other than accurate luminance, that explains the improvement in HDR images. The multiple exposure technique significantly improves digital quantization. The improved quantization allows displays to present better spatial information to humans. When human vision looks at high-dynamic range displays, it processes them using spatial comparisons.

There were many other excellent papers, for example in the Digital Publishing Special Session. However, in this limited space I will mention only one other paper, which stood out for an extraordinarily high quality experimental procedure. The research was by Kenichiro Masaoka, Masaki Emoto, Masayuki Sugawara, and Yuji Nojiri of the NHK Science and Technical Research Labs. in Japan. Here is their abstract:

Comparing realness between real objects and images at various resolutions. Image resolution is one of the important factors for visual realness. We performed subjective assessments to examine the realness of images at six different resolutions, ranging from 19.5 cpd (cycles per degree) to 156 cpd. A paired-comparison procedure was used to quantify the realness of six images versus each other or versus the real object. Three objects were used. Both real objects and images were viewed through a synopter, which removed horizontal disparity and presented the same image to both eyes. Sixty-five observers were asked to choose the viewed image which was closer to the real object and appeared to be there naturally for each pair of stimuli selected from the group of six images and the real object. It was undisclosed to the observers that real objects were included in the stimuli. The paired comparison data were analyzed using the Bradley-Terry model. The results indicated that realness of an image increased as the image resolution increased up to about 40-50 cpd, which corresponded to the discrimination threshold calculated based on the observers' visual acuity, and reached a plateau above this threshold.

As for the two papers I presented, the one on readability went particularly well, mainly because a few hours before the presentation I received from Silvia Zuffi and Carla Brambilla a new set of slides with new results on color preferences for colored text on colored background.