CVMT colloquia 2003

December 3rd, 2003: Salvatore Livatino

Title: The Augmented Actor

The use of computer generated characters, "virtual actors", in the motion picture production increases every day. While the most known computer graphics techniques have largely been adopted successfully in nowadays fictions, it still remains very challenging to implement virtual actors which would resemble, visually, human beings.
Interestingly, film directors have been looking at the recent progress achieved by the research community in the field of realistic virtual view synthesis, and they have successfully implemented state of the art research approaches in their productions. An innovative concept is then gaining consensus: the augmented actor.
Examples from the recent movie sequel: "The Matrix" and "The Matrix Reloaded", will briefly be presented.

November 14th, 2003: Erik Granum

Title: (What about the) Presence of Vision (in Cognitive Vision Systems)

Presence, as the sense of "being there", is the theme of a current European research initiative, where 10 projects are funded to develop theory and novel media for investigation of the concept of presence.  Descriptions of the goals of this initiative were given as an example of a larger interdisciplinary research context in which the visual faculty plays an important role.
As a reference for a discussion of the content and progress in cognitive vision - a vision project of the early nineties - was briefly revisited (VAP, Vision as Process, 1989 - 1995).  "Vision in context" was considered also at that time as relating vision to the environment in which is was operating as well as relating vision to other faculties and cognitive facilities within the operating (vision) system.
In an attempt to get closer to a definition of "cognitive vision", Webster's dictionary had been consulted.  This revealed "cognitive" as what could be based on, and capable of being reduced to, empirical factual knowledge. Correspondingly, "cognition" was explained as "the act or process of knowing, including awareness and judgement".
The point was made, that some contributions to define the field - also given at this seminar - suggested a much more comprehensive content of a cognitive vision system, than the term cognitive apparently could account for.  Hence the claimed characteristics like: embodiment, external action control, and action dependent vision, may be very relevant for a cognitive system, which is supported by the visual faculty, but not necessarily useful for description of a vision system itself.  Rather, the above concepts emerge when vision is active within a cognitive system and in internal interaction with other human like functionalities, such as consciousness.
In conclusion it was suggested that the vision community joined up with other disciplines of relevance for the pursuit of the role of vision in cognitive system.  Doing it in our own were not likely to reveal the insight we are aiming at.  We must be conscious and explicit about how vision is present in its context, and that is externally as well as internally.

October 24th, 2003: Salvatore Livatino

Title: Future Involvement of CVMT in EU Research Projects

In this colloquium possibilities will be presented for project proposals under EU FP6 (and others) in the near future. This will form the basis for a discussion about our status (need for new projects) and interests (which calls, topics, research frameworks are most suitable, or of interest to CVMT).
PowerPoint presentation containing facts about deadlines, calls, links, etc.

August 26th, 2003: Moritz Störing

Title: Graphics Technology -- Standard object colour spectra database for color reproduction evaluation (SOCS)

In ISO/TR 16066:2003 provides a database of typical and difference sets of existing object colour spectral data that are suitable for evaluating the colour reproduction of image input devices. It also includes the spectral reflectance and transmittance source data from which these data sets have been derived.
 

April 30th, 2003: Yong Liu

Title: The Perspective Three Points Problem Revisited

In this talk, an extended P3P (Perspective Three Points) problem is investigated. It is formulated as fitting three moving vertices along their associated optical rays to a known triangle structure. This allows the three optical rays to come from one, two or three cameras respectively. The classical P3P problem for only single camera is considered as a special case of the extended P3P problem.
 

April 9th, 2003: Claus Madsen

Title: Estimating real scene light sources for real-time augmented reality

In augmented reality virtual objects must be subjected to virtual lighting in a manner which is consistent with the real scene lighting. State-of-the-art approaches do not allow for real-time applications since virtual objects are rendered with ray-tracing techniques to account for real scene light conditions. In this talk I will describe how it is possible to estimate the parameters for a simplified lighting environment suitable for real-time with e.g., the OpenGL Phong shading model. The proposed approach allows the user to determine how complicated the lighting model should be. OpenGL supports ambient light plus up to 8 directional/positional light sources. The figure illustrates white virtual balls in an estimated lighting environment composed of an ambient term plus 4 directional light sources. The approach estimates colors, intensities and directions of the required sources.
 

March 26th, 2003: Moritz Störring

 
Title: Segmenting skin and fruit in near-infrared images

February 19th, 2003: Michael Nielsen

Title: Gesture HCI Procedure

In relation to ARTHUR and FGNET it was necessary to choose a gesture vocabulary. Rather than choosing these from a technical point of view, we decided to base it on human aspects, optimizing memorability,   learnability, ergonomics, logical interpretations, etc. For this purpose a formal procedure is developed for choosing gestures and testing the gesture vocabulary, using user tests and scenarios.
 
 

February 5th, 2003: Henning Nielsen

Title: Free Software Packages for Image Processing

The presentation will describe and demonstrate the use of four free software packages for image processing:
 

1. Scion Image
2. ImageJ
3. Filter Meister
4. CVIPtools


Together these four cover practically every type of image processing and analysis. The packages can be downloaded from this page .

January 22nd, 2003: Hans Jorgen Andersen

Title: Classifying Body and Surface Reflections using Expectation-Maximization

This paper presents a new method for classification of dielectrical object's RGB values into their body and surface reflections. Instead of segmenting into the two reflection components a weight is estimated that a given pixel belongs to one of them. A weighting value may be useful for classification of body and surface reflections in combination with other analysis methods. The method operates globally on the pixel points using expectation maximization for fitting the body and surface vectors in the case of one highlight reflection. In the case of multiple highlights it is shown that it is possible to relax the method by fitting one surface vector to multiple highlights. The method was empirically validated on real image data captured using a high dynamic imaging sensor (120dB). Promising results show that the method is capable of classifying the two reflection components.