In this issue:

Image Integration Toolbox

Using photospace to design an exposure algorithm: Use measured distributions of picture taking preferences to maximize your customer’s likelihood of achieving high quality images

Image Integration at EI2008: Full copies of paper and slides are now available.

Tools and the Image Integration Toolbox:
Several years ago I worked at Polaroid with a young colleague on improving the image quality of new digital photo print media. Although time was always in short supply I suggest spending a good proportion of it on creating tools, since a good tool facilitates the solution of more problems than those the tool was originally intended for. One day she came into my office and wrote the characters shown above on my whiteboard

The Confucian proverb says:

If a workman wishes to improve his work, he must first sharpen his tools

Image Integration is introducing a set of software tools designed to complement and work with existing image analysis packages such as those available from Imatest and QEA, to assist the user in determining what measurements to make and what those measurements mean in terms of perceived image quality.

We have introduced the first tool, Image Phi, which we demonstrated at the 2008 Electronic Imaging Conference in San Jose. Please check out the description of Image Phi at our website.

Using Photospace to design and optimize an exposure algorithm:
Using Image Phi, we have measured the frequency of mobile phone images over the operational space — Subject Illumination and Subject-Camera Distance. Since the majority of images are made under low subject illumination, the low light performance of mobile camera phones is a prime design consideration. In addition to the primary camera characteristics — lens f/#, flash type, flash power and pixel area — the algorithm that balances ambient and flash illumination with effective ISO speed is also of great interest. The optimization depends on balancing camera-subject motion blur with image noise. In general, shutter speed required for a correct exposure varies inversely with the ambient subject illumination. At low-light conditions the shutter duration increases the negative effects of motion blur and dark current. Decreasing the shutter speed requires the increase of effective ISO speed, which is accompanied by increased noise. Other constraints are imposed by the desire to have a mixture of ambient and flash illumination [‘fill flash’] and the rapid fall-off of the flash component with distance. The goal is designing a system that produces the largest number of high quality images for the conditions that your customer prefers.

We have developed a digital still camera model that can be populated with typical characteristics of mobile camera phones to predict image quality for a given set of subject lighting and subject-camera distances. Extending the evaluation over the usage distribution [photospace] will yield a predicted average image quality [an average weighted by the photospace distribution of typical mobile phone camera usage]. We have devised simple exposure algorithms that can be applied to the limited on-camera flash capabilities of mobile phone cameras in order to determine the optimum shutter durations for ambient and flash exposures under different camera operating conditions. In the example shown below, we have assumed that the camera has autofocus capability and a small, quenched xenon flash.

If you feel that applications such as described in the example above would aid your product development, or if you would prefer assistance in determining your customer’s image usage patterns and their subjective assessment of image quality, please contact us — bror.hultgren@i-2-q.com

Image Integration at Electronic Imaging 2008:
It was good to meet with many of you in San Jose and to have a chance to demonstrate both Image Phi and a tool for evaluating subjective image quality. The abstract of the paper that we delivered in San Jose — Megapixel mythology and photospace: estimating photospace for camera phones from large image sets – is available as a pdf from the website. A full copy of the paper and/or a copy of the slides is available on request.

Newsletter Archives
The intention of these newsletters is both to inform our customers and promote our products and services. Therefore we maintain an archive of past newsletters on our website. We invite you to visit and hope that the information will be of value.

Bror Hultgren
Image Integration, Inc.
978.356.5975
bror.hultgren@i-2-q.com
ImagIntegration.com