Multi-Pattern Auto White Balance Technology

When shooting scenes that include multiple light sources, such as sunlight and shade or flash, light sources are recognized automatically and the image is separated into different areas. White balance is adjusted separately in each area according to the different light sources to achieve optimal color reproduction.

Automatic recognition of sunlight and shade areas / strong and weak flash for optimal color reproduction

When shooting people in a scene that includes sunlight and shade light sources, it is difficult to achieve optimal color reproduction by the conventional method, which sets the same white balance (WB) parameters for the entire image.

If WB parameters are set for the entire image, they are set either to the background sunlit area or to people in the shade. Then, the color tone of the resulting picture is unnatural. The new method, called Multi-Pattern Auto White Balance (Multi-Pattern AWB) or Multi-pattern AUTO, sets WB parameters for sunlight and shade areas separately. Therefore the colors of both background sunlit areas and shade areas are reproduced properly.

Multi-Pattern AWB automatically sets different WB parameters for areas of different flash light intensity. As shown in Photo 1 (a), because WB is adjusted to the background area and the parameters are set for the entire image, the color of people's faces is whitish blue, where the flash is strong. In Photo 1(b), Multi-Pattern AUTO sets WB parameters for the background based on an incandescent (bulb) light source and sets WB parameters for people’s faces based on strong flash exposure. Colors of both background and people are thus properly reproduced.

(a)Set WB parameters according to background (b)Multi-Pattern AUTO
(a)Set WB parameters to background (b)Multi-Pattern AUTO

Photo 1: Multi-pattern AUTO processing

Advanced area recognition function and high-performance classifier data

To realize Multi-Pattern Auto White Balance for a scene comprising mixed sunlight and shade areas, highly accurate sunlight and shade recognition technology is necessary. With this method, selective readings are taken from both sunlight and shade to perform recognition for automatic area segmenting. Color is optimally reproduced in different areas corresponding to the various light sources.

High-quality, robust classifier data is necessary for recognition. The classifier data is gathered from an accumulated knowledge base, using many digital camera images of sunlight and shade areas. The classifier data is kept small and manageable, so that the camera can recognize sunlight and shade areas quickly and accurately to realize Multi-Pattern AWB.

This technology is applied to Ricoh's current digital camera lineup. The method yields optimal color reproduction for scenes involving people, which is difficult for conventional WB methods to achieve.

Page Top