Video encoding methods and devices
View Patent ↗A dual, and possibly multiple, frame approach is used by the invention. Embodiments of the invention include making a decision to use a long term reference frame, which is a frame other than an immediate past reference frame, to conduct INTER coding, or to conduct INTRA frame coding. Other embodiments include use of long and short term reference blocks, and make a decision between two types of INTER coding blocks and INTRA coding. In accordance with embodiments of the invention, a long term frame is a high quality frame. The high quality frame may be used as a reference frame under particular conditions.
1. A video encoder comprising:
a coder for encoding vectors to describe at least an image block with respect to at least a reference block;
a short term reference block buffer storing at least one short term reference block; and
at least one long term reference block buffer storing at least one long term reference block;
wherein the coder selectively chooses between encoding with respect to the at least one short term reference block in the short term reference block buffer and the at least one long term reference block in the long term reference buffer based upon one or more factors examined at the time of encoding to improve one of compression, video quality, and a metric balancing compression and video quality,
wherein the at least one long term reference block buffer comprise a multiple frame buffer, and the encoder coder selectively chooses between coding using the reference block (INTER coding) and using INTRA coding, and
wherein the coder chooses between two types of INTER coding and the INTRA coding, the two types of INTER coding comprising coding using the at least one short term long term reference block (ST) and the at least one long term long term reference block (LT), and wherein;
the coder computes moments for the INTRA coding and the ST block using a recursive optimal per pixel estimate treating elements of a previous block as a random variable; and
the coder computes moments for the LT block using a recursive optimal per pixel estimate treating elements of a previous block as a random variable.
2. The encoder of claim 1 , wherein the coder for encoding selectively chooses the at least one long term reference block when a connection used by the video encoder changes to a lower quality.
3. The encoder of claim 1 , wherein the coder for encoding selectively chooses the at least one long term reference block when a connection used by the video encoder is anticipated to be changing to a lower quality.
4. The encoder of claim 1 , wherein the coder for encoding selectively chooses the at least one long term reference block to encode background data and selectively chooses a more recent reference block to encode foreground data.
5. The encoder of claim 1 , wherein the coder conducts a fractional pixel accuracy encoding, by, determining, for the at least one long term reference block and on a fractional pixel grid,
original pixel positions including pixels that coincide with an actual pixel position;
horizontally or vertically interpolated pixel positions including pixels that lie between two original pixel positions; and
diagonally interpolated pixel positions.
6. The encoder of claim 1 , wherein:
first moments of the horizontally or vertically interpolated pixel positions and the diagonally interpolated pixel positions are calculated directly; and
second moments of the horizontally or vertically interpolated pixel positions and the diagonally interpolated pixel positions are estimated.
7. The encoder of claim 1 , wherein the at least one long term reference block comprises a block in a region of interest.
8. The encoder of claim 1 , wherein the at least one long term reference block comprises a composite frame predicted from both the short term frame memory and the long term frame memory.
9. The encoder of claim 1 , wherein the at least one long term reference block comprises a long term reference frame and the coder encodes a frame on a block by block basis.