To perform recognition of a shadow in an image with a complicated background by use of a single image.

In the shadow recognition method, a color image taken by a camera is separated into three color component images of, for example, R image, G image and B image in a color space resolution part 22, and stored. When recognition processing of shadow in the image is performed, extraction processing of boundary (shadow boundary) between a shadow area (an area not directly exposed to the light from a light source) and a non-shadow area is performed based on the color component images in a shadow boundary extraction part 24. Similarly, shadow area extraction processing is performed in a shadow area extraction part 25. The shadow boundary extraction result and shadow area extraction result extracted in the shadow boundary extraction part and the shadow area extraction part are integrated together in a shadow boundary-shadow area integration part 26 to generate a shadow recognition result.

COPYRIGHT: (C)2008,JPO&INPIT