| 摘要 |
An image processing device calculates a coordinate transformation coefficient for each of a plurality of regions, and, for at least one point of interest in a local region, calculates a correction amount that is used when blur in a target image is corrected using a coordinate transformation coefficient for the local region and a correction amount that is used when the blur in the target image is corrected using a coordinate transformation coefficient for a region neighboring the local region. In this case, if the correction amount difference does not fall within a predetermined limit value, a coordinate transformation coefficient is newly obtained for each of the plurality of regions including the local region so that the correction amount difference falls within the predetermined limit value, and the target image is corrected for each region using the newly obtained coordinate transformation coefficient. |
