Method and apparatus for video data compression using temporally adaptive motion interpolation
First Claim
Patent Images
1. A method for compressing video data comprising the steps of:
- determining the degree of global motion between frames of video data, where global motion is the motion between frames as a whole;
assigning reference frames based upon the global motion;
adjusting the temporal spacing between reference frames relative to the degree of global motion measured between frames relative to temporal masking in human vision, thereby determining the group of pictures (GOP) structure;
establishing different threshold magnitudes or levels of motion between frames as representing different scene change types;
assigning different bit rates to individual frames based upon said pre-established threshold levels of motion between frames;
said threshold establishing step including the steps of designating a Type 1 scene change between a pair of successive frames as occurring whenever the measured motion therebetween exceeds a value defining a T1 threshold representing a substantial scene or picture change; and
designating and defining pursuant to the value of measured motion defining a Type 1 scene change, the first occurring or prior frame of the pair as a P2 frame, and the second occurring or past frame as an I2 frame, each being of predetermined bit rates via said assigning step.
4 Assignments
0 Petitions
Accused Products
Abstract
In a system for compressing video data, temporally adaptive motion interpolation based upon temporal characteristics of human vision is used for establishing threshold levels relative to the degree of motion as a whole (global motion) between frames. The global motion between successive frames in a group of pictures (GOP) is measured to determine if the motion is less or greater than the established threshold levels for determining the designation of I, P, and B frames, spacing between I and P reference frames, and the number of bits used for each frame, and B frames therebetween.
230 Citations
44 Claims
-
1. A method for compressing video data comprising the steps of:
-
determining the degree of global motion between frames of video data, where global motion is the motion between frames as a whole; assigning reference frames based upon the global motion; adjusting the temporal spacing between reference frames relative to the degree of global motion measured between frames relative to temporal masking in human vision, thereby determining the group of pictures (GOP) structure; establishing different threshold magnitudes or levels of motion between frames as representing different scene change types; assigning different bit rates to individual frames based upon said pre-established threshold levels of motion between frames; said threshold establishing step including the steps of designating a Type 1 scene change between a pair of successive frames as occurring whenever the measured motion therebetween exceeds a value defining a T1 threshold representing a substantial scene or picture change; and designating and defining pursuant to the value of measured motion defining a Type 1 scene change, the first occurring or prior frame of the pair as a P2 frame, and the second occurring or past frame as an I2 frame, each being of predetermined bit rates via said assigning step. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37)
-
-
38. A system for compressing video data comprising video data associated with groups of pictures (GOP) including a predetermined number of frames, said system comprising:
-
motion detection means for determining the degree of global motion between said frames, where global motion is the motion between frames as a whole, wherein said motion detection means includes Type 0 scene change detector means for detecting when the cumulative motion from an immediately preceding reference frame, and a successive frame exceeds a predetermined value defined as a T0 threshold, whereby said designating means responds by designating and defining the immediately prior frame to said successive frame as a P1 frame; means responsive to said global motion measurements from said motion detection means for both designating certain of said frames as reference frames, and adjusting the temporal spacing between reference frames, thereby determining the GOP structure, wherein said designating means includes means for coding said reference frames as I and/or P, and B frames relative to global motion between frames; and encoder means for encoding said reference frames. - View Dependent Claims (39, 40, 41)
-
-
42. A method for compressing video data comprising the steps of:
-
determining the degree of global motion between frames of video data, where global motion is the motion between frames as a whole; assigning reference frames based upon the global motion; adjusting the temporal spacing between reference frames relative to the degree of global motion measured between frames relative to temporal masking in human vision, thereby determining the group of pictures (GOP) structure; establishing different threshold magnitudes or levels of motion between frames as representing different scene change types; and assigning different bit rates to individual frames based upon said pre-established threshold levels of motion between frames, wherein said assigning step further assigns bit rates to said frames in a manner utilizing a backward temporal masking effect in a resulting coding scheme, whereby in instances where a scene change involves relatively large motion between two successive frames, the immediately past frame at the scene change and the immediately following frame are coarsely coded via the assignment of a relatively low bit rate.
-
-
43. A system for compressing video data comprising video data associated with groups of pictures (GOP) including a predetermined number of frames, said system comprising:
-
motion detection means for determining the degree of global motion between said frames, where global motion is the motion between frames as a whole, wherein said motion detection means includes Type 1 scene change detector means for detecting when the measured level of global motion between two successive frames exceeds a predetermined value defined as a T1 threshold representing a substantial scene or picture change, whereby said designating means responds by designating and defining the first occurring of the two successive frames as a P2 frame, and the other or second occurring of the two successive frames as an I2 frame; means responsive to said global motion measurements from said motion detection means for both designating certain of said frames as reference frames, and adjusting the temporal spacing between reference frames, thereby determining the GOP structure; and encoder means for encoding said reference frames.
-
-
44. A method for compressing video data comprising the steps of:
-
determining the degree of global motion between frames of video data, where global motion is the motion between frames as a whole; assigning reference frames based upon the global motion; adjusting the temporal spacing between reference frames relative to the degree of global motion measured between frames relative to temporal masking in human vision, thereby determining the group of pictures (GOP) structure; establishing different threshold magnitudes or levels of motion between frames as representing different scene change types; and assigning different bit rates to individual frames based upon said pre-established threshold levels of motion between frames, wherein said assigning step further assigns bit rates to said frames by using forward temporal masking in human vision, whereby at a scene change an immediately following frame is coarsely coded.
-
Specification