Scene Recognition System for more accurate autofocus, auto exposure and auto white balance

Tokyo – Nikon Corporation is pleased to announce the introduction of their new Scene Recognition System, which recognizes a subject or scene prior to capture, then, using information from the 1,005-pixel RGB sensor, that enables highly precise exposure control utilizing color information, applies that information to achieve more accurate autofocus, auto exposure and auto white balance control.

The Scene Recognition System is an epoch-making function realized through the improvement of image recognition capability of the 1,005-pixel RGB sensor by employing the diffraction grating for it. This greatly increases the precision of the information to be utilized in auto exposure and auto white balance, as well as in autofocus.

Since the introduction of the Nikon FA that employed the first multi-pattern (Matrix) exposure metering system, the basic advantage of the system has been its scene recognition capability. In auto exposure, improvements achieved through adoption of the 1,005-pixel RGB sensor with the Nikon F5 are further enhanced by the advancement in hardware.

With this higher level of the improved scene recognition, even changes of subject position within the viewfinder can also be recognized as image information. This capability is utilized in autofocus as the 3D-tracking function. In auto white balance, better image recognition combined with the advanced auto white balance algorithm achieves higher-performance white balance control than ever before.

The new Scene Recognition System utilizes high-precision information from the 1,005-pixel RGB sensor to conduct subject tracking and identification, highlight analysis and light source identification. This data is then reflected in autofocus, auto exposure and auto white balance for greatly improved accuracy and control.

Autofocus applications

Subject identification: Information from the 1,005-pixel RGB sensor is used to identify sky, background and foreground, and color of the subject (a first for SLRs). This data is then applied to Auto-area AF mode, which gives greater priority to the subject's position.
Subject tracking: This function uses output from the 1,005-pixel RGB sensor to recognize the color of a subject within a user-selected focus point, then follows the subject's movement by detecting the identical color segment. This information is then used by Dynamic AF to improve backup capability when the subject is out of the selected area. Newly employed “3D-tracking” mode shifts the focus point automatically to respond to the subject's movements.

Auto exposure application

Highlight analysis: Information from the 1,005-pixel RGB sensor is utilized to precisely detect highlights within the frame. This data is then applied to color matrix metering and i-TTL balanced fill-flash, and the brightness range to be reproduced is calculated to realize more precise exposure control.

Auto white balance application

Light-source identification: Using extracted characteristics information from the 1,005-pixel RGB sensor, the scene is recognized.
Correctly identifying the light source ensures more accurate white balance, even in a situation where it is difficult to determine it with conventional auto white balance.

The new Scene Recognition System will be employed initially in the Nikon D3 and D300.