| 摘要 |
The amount of misalignment (Dbmi) with respect to a reference image (Da0) is detected (10-1 to 10-M) for each pixel of reference images (Da1-DaM) and each pixel of the reference image is allocated in accordance with the amount of misalignment (Dbmi), whereby aligned images (Dc1-DcM) are generated (20-1 to 20-M) and weighted addition is performed using the reference image (Da0) and the aligned images (Dc1-DcM). A weight corresponding to whether each pixel of the aligned images (Dc1-DcM) is a defined pixel or not is added, or an undefined pixel is interpolated, before the weighted addition is performed. Even when the viewpoints of a plurality of input images are mutually different, it is possible to solve a problem caused by an image deformation process, and to generate a composite image having high picture quality. |
