| 摘要 |
The invention relates to a method for predicting a video image on the basis of a preceding video image in the processing order for use in compressed transmission or storing of video images, wherein a prediction is generated for a video image to be encoded in an encoder encoding the video image on the basis of the preceding image in the processing order. The invention comprises dividing the preceding image in the processing order into blocks; selecting a point from each block of the preceding image in the processing order as the basic point (A, B, C, D) of the concerned block, the basic points of the different blocks dividing the preceding image in the processing order into quadrangle-shaped basic areas; searching the image to be encoded to find points (A', B', C', D') corresponding to the basic points (A, B, C, D) of the blocks of said preceding block in the processing order, the corresponding points dividing the image to be encoded into corresponding quadrangle-shaped areas; defining a motion vector for each pair of a basic point and a corresponding point, the motion vector indicating a displacement between the points; and performing a geometric transformation on the quadrangle-shaped basic areas so as to transform them into the size and shape of their corresponding areas for generating predictions for the corresponding areas of the image to be encoded. <IMAGE> <IMAGE> |
