Visual Computing

Visual computing is a cross-disciplinary emerging field of computer science that deals with information technological challenges for effectively acquiring, representing, processing and manipulating real-world "mages'' on one hand, and synthesizing stunning video-realistic "mages'' based on compact discrete mathematical representations on the other hand.

The last decade has attested the quick and massive consumer adoption of digital imaging that opened up the way to a brand new era of photography called computational photography.

At first, computational photography concentrated at catching up and removing limitations of conventional analog cameras: That is, mainly the image resolution, the noise sensitivity and the field of view.

Nowadays, full surround gigapixel imaging became painless and mainstream thanks to the robust automatic yet flexible stitching capabilities developed over these digital imaging years.

Yet, this digital paradigm shift not only fully changed the technological fundamentals, it also offered at the same time limitless creative scenarios to capture, process and depict our world beyond the conventional 2D picture.

In order to make these ideas benefitable to the 21st century's camera culture, visual computing concentrates on building efficient algorithms and data-structures for spatio-temporal processing.

The core of visual computing consists in harmonizing geometry, vision and graphics fields for a truly seamless 3D visual pipeline.

Indeed, these fields have historically been pioneered and developed independently yielding various discrepancies.

Another key aspect of visual computing is robust and real-time image understanding for smart intelligent cameras.

For example, we proposed a fast image segmentation technique that analyzes and splits images into homogeneous areas called segments.

These segments represent the first low-level semantic of the scene that are in turn handled by more complex tailored machine learning algorithms for

better higher-level semantic scene understanding.

Our segmentation algorithm relies on an original probabilistic framework that mathematically guarantees the result and its stability.

Visual computing has for sure more than face recognition and automatic smile shutter mode to offer to computational photography for the years to come!