Back to results
US 7,889,791 B2Granted
Method and apparatus for scalable compression of video
OwnerDavid Taubmaninventor
Life of this patent
Estimating the term…
Loading patent…
Loading patent…
OwnerDavid Taubmaninventor
Estimating the term…
The present invention relates to a method and apparatus for producing a fully scalable compressed representation of video sequences, so that they may be transmitted over networks, such as the Internet, for example. Because the signal is scalable, users receiving the signal can obtain the signal at the appropriate resolution and quality that their system will handle or that they desire. The invention implements a “motion compensated temporal wavelet transform” in order to enable compression of the video in a scalable fashion while still taking advantage of inter-frame redundancy in a manner which is sensitive to scene and camera motion. The motion compensated temporal wavelet transform is implemented by decomposing the video sequence into a set of temporal frequency bands and then applying a sequence of motion compensated lifting operations to alternately update an odd frame sub-sequence based upon an even sub-sequence and vice versa in a manner which is sensitive to motion.
A method of compressing a video sequence to produce a compressed video signal, including the steps of forming the video sequence into an input signal, transforming the input signal into a scalable, compressed video signal by the steps of decomposing the input signal into a set of temporal frequency bands, in a manner which exploits motion redundancy, and applying scalable coding methods to the decomposed input signal to generate a scalable compressed video signal wherein the step of decomposing the input signal into a set of temporal frequency bands includes the steps of separating the input signal into even and odd indexed frames sub-sequences, by applying a sequence of motion compensated lifting operations, each of which either updates the even indexed frame sub-sequence by the addition of a linear combination of motion compensated frames from the odd indexed frame sub-sequence, or updates the odd indexed frame sub-sequence by the addition of a linear combination of motion compensated frames from the even indexed frame subsequence, in a manner which is sensitive to motion.
A system for compressing a video sequence to produce a compressed video signal, the system comprising means for forming the video sequence into an input signal, means for decomposing the input signal into a set of temporal frequency bands, in a manner which exploits motion redundancy, and a means for applying a scalable coding method to the decomposed input signal to generate a scalable compressed video signal wherein the means for decomposing the input signal into a set of temporal frequency bands includes a lifting means, arranged to separate the input signal into even and odd indexed frame sub-sequences, the means being arranged to apply a sequence of motion compensated lifting steps, each of which either adjusts the even indexed frame sub-sequence by the addition of a linear combination of motion compensated frames from the odd indexed frame sub-sequence, or adjusts the odd indexed frame sub-sequence by the addition of a linear combination of motion compensated frames from the even indexed frame sub-sequence, in a manner which is sensitive to motion.
Dependencies are inferred from the claim text.