| 摘要 |
The invention relates to a method and a device for the search for a prediction block in predictive video compression. In the method, a desired search area is searched for an image block resembling an image block (B) to be encoded as closely as possible for use as a prediction block for said image block. In order that the best possible prediction block could be located with considerably less calculation than previously, (a) image blocks (A) acting as prediction block candidates within the search area are classified by deriving an identifier (CA) from each image block (A) acting as a prediction block candidate, the identifier describing the image content of the concerned image block, and storing the location of the image block acting as a prediction block candidate in a memory (21) at an address determined by said identifier, whereby the storing takes place such that at least one location data item remains in the memory for each identifier; (b) an identifier (CB) is similarly derived from each image block (B) to be encoded, the identifier describing the image content of said block; and (c) the location of the prediction block for the block (B) to be encoded is fetched from said memory (21) from a memory position determined by the identifier (CB) describing the image content of the block. <IMAGE> |
