Although many of us use PhotoShop or other similar software in order to fix red eyes, lots of people will welcome the immediate removal of this nuisance. According to FotoNation, their patented red eye technology has been licensed for use in more than 100 million cameras and imaging products. In addition, they take pride in the fact that their product has been embedded in devices that are not digital cameras or camera-enabled cell phones such as photo printers, kiosks, and PC software.
The red-eye effect in photography is the common appearance of red pupils in color photographs of eyes. Almost all cases are the cause of a photographic flash made very close to the camera lens (as with most compact cameras), in ambient low light. Not all subjects suffer from this color-distortion; the red-eye effect appears only in the eyes of humans and animals that have no tapetum lucidum – meaning, no eyeshine.
The FotoNation Red embedded software library delivers, according to its maker, a detection rate of up to 90% with a false positive rate of less than 2%, making it one of the most accurate in the industry (the common detection rate is said to be around 80%). Such high detection rates are the result of regression tests performed on a red-eye image database of more than 5,000 images. It means that a large number of photos were analyzed in order to achieve better recognition of red eyes. The database includes a wide range of eyes under many different illumination conditions, such as incandescent, fluorescent, shade, and mixed lighting. The test images were captured using more than 30 different digital camera models, at different resolutions, at varying distances between subject and camera and with subjects of all ages. Other varied parameters that were tested are iris colors and skin types.
Unlike other solutions on the market, FotoNation Red 2.0 is the first embedded software that specifically addresses the problem of golden-eye defects. Although it could occur with any subject, the golden-eye phenomenon is most common in the eyes of Asian subjects. Since this phenomenon is relatively rare, most companies did not find dealing with this problem economical; therefore, FotoNation claims to have managed to separate itself from the crowed and gain an edge over their competitors because they fill this need.
According to FotoNation, red-eye detection and correction time can be as little as 0.3 seconds (depending on system configuration), and unlimited number of red eyes in one image can be corrected, even if faces are turned or obscured. The minimum size of red or golden eyes able to be corrected is as small as 8 pixels, and multiple image formats are supported, including RGB and YUV. Code size for a typical embedded software implementation of the FotoNation Red technology is 100KB, meaning manufacturers could implement it in various gadgets without the need for memory expansion. Moreover, its API is flexible and was specifically designed to fit a large number of camera models. The Red 2.0 technology is also highly compatible with various operating systems, including Windows WinCE, Linux, MacOS, and Symbian.
TFOT has previously covered the T5000, a camera that can detect weapons, drugs, and explosives hidden under people’s clothes from a distance of up to 25 meters. Other TFOT related stories include the Nikon CoolPix P6000, a 13.5-Megapixel camera which features a built-in global positioning system, and the Casio EX-F1, a camera capable of capturing up to 60 stills frames per second.
For more information about the FotoNation Red 2.0, see the company’s website.
