Systems and methods for decoding of partially corrupted reversible variable length code (RVLC) intra-coded macroblocks and partial block decoding of corrupted macroblocks in a video decoder
View Patent ↗The invention is related to methods and apparatus that recover usable video data from partially corrupted data. Embodiments inspect corrupted data packets and identify the location or locations of an error, whether the corrupted data packet contains data expected to be error-free, and whether the error-free data should be used. Decoding of a packet in both the forward direction and the backward direction can be used to locate a position of an error. Intra-coded macroblocks can also be recovered. A decoder can elect to use or to drop an intra-coded macroblock recovered from a corrupted data packet according to further criteria that is applied to the recovered intra-coded macroblock. One embodiment inspects video bitstream data that has been encoded with an optional data partitioning feature enabled, and retrieves specified data in areas of a corrupted packet that are expected to be free from error.
1. A method of recovering useful data from a video packet that has been corrupted, the method comprising:
receiving the video packet;
ending without recovering data when corruption is detected in a video packet header of the video packet;
ending without recovering data when corruption is detected in a DC portion of the video packet;
ending without recovering data when corruption is detected in a motion vector portion of the video packet;
initiating decoding of the video packet in a forward direction;
maintaining a first count of a number of macroblocks decoded without error in the forward direction;
storing codewords decoded in the forward direction;
storing a first bit location when an error is first detected in the forward direction;
initiating decoding of the video packet in a reverse direction;
maintaining a second count of a number of macroblocks decoded without error in the reverse direction;
storing codewords decoded in the reverse direction;
storing a second bit location when an error is first detected in the reverse direction;
determining if there is an overlapping region, where the overlapping region corresponds to a region identified in both the forward direction and in the reverse direction as having an error;
if there is an overlapping region, discarding the data in the overlapping region and using the data in a remaining portion of the video packet; and
if there is no overlapping region, discarding the data between a first backtracking amount ahead of the first error location in the forward direction and a second backtracking amount behind the second error location in the first location, and recovering the remaining portion of the video packet.
2. The method as defined in claim 1 , wherein the first error location and the second error location correspond to bit locations.
3. The method as defined in claim 1 , wherein the first error location and the second error location correspond to macroblock boundaries.
4. The method as defined in claim 1 , wherein the first backtracking amount and the second backtracking amount are each to a next valid macroblock boundary.
5. The method as defined in claim 1 , wherein the first backtracking amount and the second backtracking amount are about 90 bits.
6. The method as defined in claim 1 , further comprising discarding recovered data from a corrupted video packet that corresponds to an intra-coded macroblock.
7. The method as defined in claim 1 , further comprising:
determining whether AC prediction was disabled by the encoder;
using a recovered intra-coded macroblock if the intra-coded macroblock is recovered from a portion of the video packet that is ahead of a DC marker in the forward direction, where the video packet was encoded with AC prediction disabled; and
otherwise discarding recovered data corresponding to an intra-coded macroblock.
8. The method as defined in claim 1 , further comprising using recovered data corresponding to a first intra-coded macroblock only if no other intra-coded macroblock exists to the immediate left of the first intra-coded macroblock and no other intra-coded macroblock exists immediately above the first intra-coded macroblock in the image.
9. The method as defined in claim 1 , further comprising concealing errors with gray pixels for portions of the video packet that were not recoverable.