No abstract

The media and communications providers share an increasing interest in 3D models of people, objects, and scenes. The Shifts in 3DTraditionally, 3D acquisition technology has been developed for visual inspection and robot guidance. These two domains have for a long time also been the two major driving forces behind computer vision in general. Nowadays however, telecommunications and other domains with a need for realistic visualisation -and this encompasses the media in general -have become the primary motor behind developments in computer vision.Quite a few of these newer developments involve solutions to problems nobody had really been dealing with before. Think of database retrieval, visual speech, or the recognition of emotions from facial expressions. By contrast, 3D modeling had quite a history already and consequently it was tempting to use existing technology for the creation of 3D models in these new application areas. As a result, laser scanners are still the 3D acquisition technology of choice for many.Yet, the novel applications of 3D acquisition technology come with requirements of their own:For one thing, the extraction of surface texture is a must. Visualisation of untextured 3D surfaces is anything but unacceptable for most applications in the (tele)communications arena. For instance, people find it even surprisingly difficult to recognize a person's face when shown a detailed 3D but untextured representation.The absolute scale of shapes is of much less importance. Their images may be shown at quite different scales anyway, There is a larger variety of object sizes to be handled. Whereas the traditional industrial setting would allow for a predefined and probably rather narrow range of sizes, applications that require 3D modelling for visualisation typically deal with substantial variation of sizes. Think of the different objects used in special effects for the movies. In 'Twister' we saw spiral anything from bricks to large fuel trucks into the sky.There is a tremendous advantage to be gained if dynamic 3D can be extracted, ie. 3D shapes together with their changes over time. This is a dream come true for people in animation, who now have to go to great length in order to extract natural motions from observed markers in performance animation or through the tedious modeling of muscle and skin dynamics.As the users will probably be postproduction people, artists, or even participants in games rather than engineers from the quality control department whose primary job it is to operate the 3D acquisition system, these new 3D acquisition systems should be easier to use and easier to install. No mindboggling calibration procedures here... This is all the more important as for these visualisation oriented markets it is more often necessary to freely move around the whole apparatus.Traditional 3D acquisition systems do not quite match this shopping list. Laser scanners and stereo systems are most widespread. The former have a working volume that is restricted primarily by the need for precise ...

No abstract

No abstract