摘要 |
Of an object extracted from an image, it is difficult to separately process a main object and a secondary object, such as a shade, attached to the main object. If an object extracted from an image contains a plurality of objects, these objects are difficult to process as independent objects. In this invention, an object indicating a shade is separated from an object extracted from an image to thereby classify the object into a main object and a secondary object, and suitable image processing is performed for each of the main object and the secondary object. An object extracted from an image is separated on the basis of shape information of the object. The similarity of each separated object is measured in units of frames to obtain an interframe correlation. This increases the encoding efficiency.
|