| 摘要 |
A method for processing an image (3082) sa as to automatically locate a region of interest such as a human face. The image is converted to a suitable color space (3803), after which a series of seed values (3804, 3805) are inputted, each having a seed color range, a local colour difference limit and a global color difference limit. For each seed value, an output region of interest (3812) is determined by determining adjacent pixels which differ by no more than the local color dofference limit and differ from the seed colour by no more than the global colour difference limit. The resultant image object is located within the image and the image can be processed by a handheld camera utilising and the processes information.
|
