Research

PhD Alessandro Artusi

Team Leader

DeepCamera Group

Research Center on Interactive Media Smart Systems and Emerging Technologies

RISE Ltd.

Cyprus

Research

My primary research interest is the development of techniques for the acquisition, manipulation/enhancement and visualization of high-quality and realistic imagery in a real-time environment. In particular, I have been working in the following sub-areas:

Color Science

Color as an important aspect of our daily life, it needs to be accurately reproduced under different lighting conditions and devices. This is requiring to understand how colors need to be mapped on different devices colorimetric characterization, taking into account the different properties and characteristics of the device where the image will be visualized such as the color gamut as, well as we need to simulate several aspects of how our visual system is working to properly convey the appearance of the image under specific viewing conditions.

Image Processing

My mainly interest is to develop novel image processing techniques to solve problems in several other research areas such as Computer Graphics and Vision. Recently we have presented a state-of-the-art of specular removal techniques, where a deep analysis and comparison is presented (see the Website for more info). A novel image decomposition techniques, able to decompose textures while preserving edges, has been proposed and applied on several applications providing excellent results.

Computational Photography

I am mainly active in the area of High Dynamic Range Imaging (HDRI). The term “dynamic range” is used to describe the ratio between the smallest and largest possible values of a quantity. In the context of HDRI, this quantity is the light information or luminance and its dynamic range is the ratio between the brightest and the darkest values. HDRI allows to acquire the vast luminance dynamic range available in the real world, merging a series of photographs/frames taken under different exposure time. This means to represent the merged image with a bit-depth higher than the 8-bit used in the traditional imaging. This way of representing the image can not directly visualized on traditional display, and it requiring a compression step called tone mapping. We have proposed a technique called Hybrid that takes into account the best tonemapping operator in a specific zone luminance and combine their result into the final output. For accelerating the tone mapping step to a real-time performance, we have introduced two techniques. The former, adopts an efficient use of the interoperability between CPU-GPU. The later, makes use of knowledge where the strong contrast is located it is limiting the use of a computational expensive local luminance adaptation computation only on these regions. This results, in largely increasing the computational performances for very large HDR images/textures. To this techniques has been granted a UK Patent number GB2449272, Journal number 6235, and it has been integrated in a commercial software. I have been working in the development of a compression/decompression technique for HDR content. Three patent applications are pending and it has been exhibited at the emerging technology of SIGGRAPH ASIA 2009 and at the exposition of SIGGRAPH 2009 respectively. Recently, I am working on a novel HDR display technology, that won the 1st Prize at the CyEC 2010 business competition, and three patent applications are pending. I have been working also on evaluation studies of tone mapping operators, as well as inverse tone mapping operators, and on some Cultural heritage applications making use of HDR content. I am part of the management committee of the COST action IC1005, and I am co-author of the book Advanced High Dynamic Range Imaging: Theory and Practice CRC Press.