Adaptive motion vector control
AssigneeThomson Licensing
InventorMary Lafuze Comer
Life of this patent
Estimating the term…
Abstract
An apparatus for reproducing a digitally encoded signal. The apparatus comprises a decoding means for decoding the digitally encoded signal to form a specific picture. A memory means is coupled to the decoding means for storing the specific picture. A control means is controllably coupled to the memory means, wherein a first mode the control means controls the memory means for storing a frame of the specific picture and responsive to a second mode the control means controls the memory means for storing only one field of the specific picture.
Claims33
An apparatus for processing a digitally encoded signal produced from a medium, comprising:
- a source of said digitally encoded signal representative of a plurality of block processed pictures, a group of said plurality of block processed pictures including at least one intra coded picture and at least one forward predicted picture;
- a decoder coupled to said source for decoding said one intra coded picture;
- and,
- processing means coupled to said decoder for processing individual blocks comprising said intra coded picture to determine values of spatial frequency components within said individual blocks,
- wherein said values within said individual blocks being classifying into types representative of picture content for use as predictors for decoding said at least one forward predicted picture.
An apparatus for processing a digitally encoded signal from a medium, comprising:
- a source of said digitally encoded signal representative of a plurality of groups of digitally encoded block processed pictures, a group of said plurality of groups including at least one intra coded picture and at least one forward predicted picture and motion compensation data;
- a decoder coupled to said source for decoding said one intra coded picture;
- and,
- processing means coupled to said decoder for processing each block of said intra coded picture to classify image content type represented by data occurring within said block from a plurality of image content types, and storing said classified image content type represented within said block for subsequent image processing.
A method for processing a digitally encoded signal having a plurality of groups of digitally encoded pictures, a group of said plurality of groups including at least one intra coded picture processed as a plurality of blocks, at least one forward predicted coded picture, comprising the following steps:
- a) decoding and storing said one intra coded picture;
- b) processing each block of said plurality of blocks to derive separate spatial frequency components therein;
- c) classifying amplitudes values of said components of said each block as a type of a plurality of types;
- d) storing in a memory said type for said each block,
- e) decoding said one predicted picture,
- f) reading from said memory a stored block type for a block corresponding to said block currently decoding in said one predicted picture;
- and
- g) modifying a motion vector for said block currently decoding to change prediction responsive to said stored block type read from said memory.
An apparatus for reproducing from a medium a digitally encoded signal representative of a plurality of groups of pictures, a group of said plurality of groups including an intra coded picture and a predictively coded picture, comprising:
- a decoder coupled to said medium for decoding said digitally encoded signal to form said intra coded picture and said predictively coded picture;
- a memory coupled to said decoder for storing only one of a first field and a second field of said intra coded picture;
- and,
- control means coupled to said decoder and said memory for controlling prediction of said predictively coded picture responsive to matching between a motion vector field selection corresponding to said predictively coded picture with said only one field, wherein responsive to a mis-match said control means controllably modifies prediction to occur from said only one field.
Dependencies are inferred from the claim text.