| 摘要 |
PROBLEM TO BE SOLVED: To retrieve a correct motion vector at the first stage of retrieval by extracting whether or not a characteristic is from all the screens of an image and its coordinate, so as to detect a motion vector in addition to detection of a conventional motion vector. SOLUTION: This detector is provided with an original image steerage means 1, a reference image storage means 2 and a motion vector detection circuit 3 detecting motion of the original image with respect to a reference image. In this case, the dector is provided also with a large area motion vector detection circuit 10, including at least a characteristic setting means 11 that sets and records a characteristic area capable of distinguishing from another part in a major object in the image, a characteristic coordinate extract means 12 that seeks a characteristic from the original image to extract the characteristic coordinate and a characteristic coordinate storage means 13 that stores the characteristic coordinate. A large area motion vector is detected from a difference between the characteristic coordinate stored in the storage means 13 and the extracted characteristic coordinate based on characteristic presence/absence information outputted from the characteristic coordinate extract means 12, when the means 12 extracts the characteristic coordinate from the original image and when the characteristic coordinate extract means 12 cannot extract the characteristic, a normal motion vector is detected. |
