One hundred years of 2D+t is enough.
HoloCam is advancing the state of the art of image processing by and for multiple-point-of-view cameras (an MPOV camera is an array of cameras/imagers packaged into a single platform).One hundred years of 2D+t is enough. Current video presentation lacks true 3D content. The scene does not change when the viewer moves his head - he cannot 'look around corners.' And of course video is rarely supplied stereo-optically.
While there are many components to both the problem and solution of 3D+t, an MPOV system offers a promising start. With an array of imagers:
* POV selection (changing viewpoint) becomes possible; changing viewpoint can give a 3D effect (the looking-around-the-corner effect) even without stereovision;
* new spatial-temporal compression methods become available.
The advances include the development of new (real-time) MPOV algorithms to enable 3D structure extraction and image segmentation for use in video production and machine vision applications. Further, and most importantly, we plan to demonstrate that the additional information provided by the MPOV approach will always permit greater video compression than by any other means.
HoloCam's MPOV camera project is on track to create products that will experience exponential adoption in large and profitable industries.
Since the output of this camera system is a single data stream into which all of the data from all of the imagers has been compressed the viewer can, on the fly:
* choose any stereo-optic pair as his POV, and
* even change his stereo-optic distance (effective eye separation: optically increasing eye separation increases depth perception).
These characteristics enable:
* 3D video capture and viewing from different perspectives,
* video and film production efficiency - powerful digital video film production at bargain prices,
* 3D video delivered over the Internet or other narrow-band pipes,
* cost-effective, enhanced machine vision,
* an enhanced ability to overlay special effects on film and video.
Other than their obvious entertainment value, the abilities to change POV, vary stereo-optic distance and extract 3D structure data can improve productivity and performance in: medical imaging, surveillance, telepresence (human interactive robotic systems), etc. In surveillance applications the ability to extract 3D data will aid in face recognition and other Homeland Security type applications.
An example (telepresence): In a telepresence environment the data, when properly used, from the additional viewpoints provided by an MPOV camera becomes information that can be employed in two ways:
* increase the effective video data compression;
* provide augmented (enhanced 3D) telepresence.
The video data compression can be increased even more if not all of the information and data from all of the POVs is sent (which we can think of as allowing the quality of the telepresence to degrade). Also, the tradeoff between video data rate and telepresence quality can be managed dynamically as the situation changes, which leads us to the concept of: metered telepresence.