Systems and methods for probabilistic mesh motion estimation
Disclosed herein are systems, devices and methods for probabilistic mesh motion estimation. The systems, devices and methods can involve an encoder, appliance, or other device for processing the video data to estimate motion in one or more locations of the video data using probabilistic mesh motion estimation based on a motion model. The encoder can introduce inter-probability in the motion model for appearing areas in the video data in conjunction with using the probabilistic mesh motion estimation. The encoder can introduce a probability that regions in the video data are intra-regions or inter-regions in the motion model as weights to indicate a confidence of finite elements being inter-associated. The encoder can use master and slave nodes and over-splitting for mesh continuity preservation. The encoder can use a non fully connected mesh applicable to video coding standards.
1 . A computer-implemented system for probabilistic mesh motion estimation, the system comprising:
at least one processor; and
memory in communication with the at least one processor, the memory storing software code, which when executed by the at least one processor causes the processor to process video data to estimate motion in one or more locations of the video data using probabilistic mesh motion estimation based on a motion model, the probabilistic mesh motion estimation using a mesh as a motion vector field to estimate motion at the one or more locations of the video data, the mesh defining finite elements representing defined shapes of reference areas for motion estimation, wherein the motion model comprises weights representing probabilities that the one or more locations in the video data are intra-regions or inter-regions to indicate a confidence of the finite elements being inter-associated.
2 . The computer-implemented system of claim 1 further comprising a transceiver for receiving and transmitting the video data.
3 . The computer-implemented system of claim 1 wherein the memory stores the video data.
4 . The computer-implemented system of claim 1 wherein the processor is an encoder, the encoder comprising a preprocessor, a look-ahead, and a codec.
5 . The computer-implemented system of claim 1 wherein the processor introduces the weights inter-probability in the motion model for appearing areas in the video data in conjunction with using the probabilistic mesh motion estimation.
6 . The computer-implemented system of claim 1 wherein the processor uses master and slave nodes and over-splitting for mesh continuity preservation, wherein the master nodes are movable nodes and the slave nodes are constrained nodes with motion that can be computed from their connected master nodes.
7 . The computer-implemented system of claim 4 wherein the encoder uses a non fully connected mesh applicable to video coding standards for compression and preprocessing.
8 . The computer-implemented system of claim 1 wherein the processor implements dedicated splitting for motion continuity preservation.
9 . The computer-implemented system of claim 4 wherein the encoder uses the probabilistic mesh motion estimation for compression of the video data.
10 . The computer-implemented system of claim 4 wherein the look-ahead implements the probabilistic mesh motion estimation.
11 . The computer-implemented system of claim 4 wherein the codec implements the probabilistic mesh motion estimation.
12 . The computer-implemented system of claim 4 wherein the preprocessor implements the probabilistic mesh motion estimation.
13 . The computer-implemented system of claim 1 , wherein the processor partitions the mesh to refine motion at a location in the mesh, wherein the processor uses continuous mesh splitting to partition the mesh without introducing discontinuities into the mesh.
14 . A computer-implemented method for probabilistic mesh motion estimation, the method comprising:
processing video data to estimate motion in one or more locations of the video data using probabilistic mesh motion estimation based on a motion model, the probabilistic mesh motion estimation using a mesh as a motion vector field to estimate motion at the one or more locations of the video data, the mesh defining finite elements representing defined shapes of reference areas for motion estimation, wherein the motion model comprises weights representing probabilities that the one or more locations in the video data are intra-regions or inter-regions to indicate a confidence of the finite elements being inter-associated; and
transmitting or storing the video data.
15 . The method of claim 14 further comprising partitioning the mesh to refine motion at the location in the mesh, wherein the partitioning involves using continuous mesh splitting to partition the mesh without introducing discontinuities into the mesh.