Thursday, May 22, 2008

I3A Imaging Innovation Conference

I3A holds its annual Imaging Innovation conference in Vancouver on June 4-6. It stages a few interesting discussions, with image quality criteria being one of them:

"Eye of the Beholder" - Can We Measure Subjective Image Quality?

Moderator: Larry Scarff, Vista Point Technologies
Panelists: Frederic Cao, DxO Labs; Jim Clark, HP; Elaine Jin, Aptina Imaging; Jonathan Phillips, Eastman Kodak Company

When a human observer views an image, a series of psychological responses is evoked, ranging from pleasure to distaste. The physical, biological and psychological processing which results in a particular human response to an image can be quantified and the resulting measurements are referred to as subjective measurements. Find out how we can take and use these measurements to predict future responses.

Rapid Changes in Imaging Technology Require a Move to Programmable Image Processors

Nicolas Touchard, DxO Labs

As camera phone resolution increases above 3 megapixels and as the camera module form factor continues to shrink with new innovations such as wafer level optics, the traditional sensor SoC architecture is being challenged to meet both the size and image quality requirements. At the same time, the emergence of highly specialized process technology, such as back side illumination, to improve small pixel performance is causing the growing number of logic gates on the sensor chip to become a cost liability.

To complicate matters, during the time it takes for a processor to be designed and ramp to production, the ISP requirements can change several times due to new pixel and optics innovations.

In this session, we will show the divergence in requirements, where new pixels and optics technologies are being introduced more frequently and with greater change than can be accommodated by a fixed ISP on a system chip, driving the need for a programmable ISP architecture. Filling this need is a challenge however - combining high processing performance with scalability and power requirements is far beyond the capabilities of general purpose cores available today, demanding new innovative architectures.


As much as I like DxO's approach and its attention to details, I wish it finds a right proportion between flexibility and silicon real estate. If it delivers area and power efficient solution, it could be a killer product on the market.

No comments:

Post a Comment

All comments are moderated to avoid spam and personal attacks.