| 摘要 |
The present invention discloses a method for extracting an object region, which can precisely modify a boundary of a roughly extracted object region (reference region). The boundary of the reference region can be modified by detecting edge pixels nearest to the boundary pixel of the reference region. Thereafter, an expanded region is designated from the modified reference region, and a surrounding region having the smallest color difference is selected by using color differences between the expanded region and the surrounding regions. At last, the modified boundary of the reference region is modified by using distribution of interest region included in the selected surrounding region. As a result, the object region can be more precisely extracted by preventing confusion of the boundary by modifying the reference region twice. <IMAGE> |
