| 摘要 |
A system and method for de-interlacing, motion compensation and frame rate conversion of digitally encoded video signals. The invention utilizes multiple motion vectors per pixel of the final image. In a preferred embodiment, this is accomplished by increasing the size of the image, with the amount of size increase depending upon the degree of sub-pixel accuracy desired. In a preferred embodiment, image size is doubled. Thus, four motion vectors are generated for each pixel. Each motion vector is found by searching independently for the best match with previous and subsequent frames. In one aspect, the invention includes a method of temporal processing of motion picture image frames each comprising a plurality of pixels, including the steps of comparing each pixel of a current frame to at least one previous or subsequent image frame; determining at least one motion vector corresponding to each such pixel relative to such at least one previous or subsequent image frame; and saving the determined motion vectors. The invention also includes the steps of applying motion vectors corresponding to multiple image frames to define a new pixel for each pixel of the current frame, and outputting all of such new pixels as a constructed image frame. The invention includes system and computer program implementations. |
