| 摘要 |
The present invention provides a method for decoding corrupt codestreams for encoded digital imagery and video and, in particular, JPEG2000 codestreams with improved error resilience properties. The decoding techniques apply to a class of coding algorithms in which the data from underlying images are partitioned, typically to allow decoding of different spatial sections of the image, and the rules governing the dependencies within and perhaps between partitioned sets are known. Corrupt codestreams are decoded with improved image quality by observing the partitions imposed by the underlying algorithm ( 200 ), detecting an error in a partition set ( 202 ), analyzing the dependencies within and perhaps between the partitions ( 204 ), determining what sections of encoded data that follow the error in the partition set can be salvaged ( 206 ), and decoding those sections ( 208 ). Even though there might be errors in a particular partition, portions of that partition might still be completely or partially decodable, depending on certain modes or "switches" used during the creation of the codestream.
|
