WEAV video compression system

• US 9,106,925 B2
• Filed: 01/11/2011
• Issued: 08/11/2015
• Est. Priority Date: 01/11/2010
• Status: Active Grant

• Current Assignee: Patential Holdings, Inc., Typenex Co-Investment, LLC
• Original Assignee: Ubiquity Holdings, Inc.
• Inventors: Carmichael, Christopher, Jordan, Connie, Sandberg, Kristian
• Primary Examiner(s): Czekaj, Dave
• Assistant Examiner(s): ABIMBOLA, KEHINDE
• Application Number:
  US13/004,592
• Publication Number:
  US 20110170607A1
• Time in Patent Office: 1,673 Days
• Field of Search: 375/240.18
• US Class Current: 1/1
• CPC Class Codes: H04N 19/132 Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
• CPC Class Codes: H04N 19/139 Analysis of motion vectors, e.g. their magnitude, direction, variance or reliability
• CPC Class Codes: H04N 19/172 the region being a picture, frame or field
• CPC Class Codes: H04N 19/179 the unit being a scene or a shot
• CPC Class Codes: H04N 19/51 Motion estimation or motion compensation
• CPC Class Codes: H04N 19/527 Global motion vector estimation

Abstract

Compression of a video by segmentation, transformation, transform encoding, difference encoding, and muxing. The segmentation looks at energies of the frames, in order to find three different things: black frames which have energy less than the specified amount, consecutive frames which have energy difference is less than the specified amount, and changes in scene which have energy difference is more than a specified amount. If there is little or no energy or energy difference, then the frame is replaced by an indication rather than information about the frame itself; either an indication of a black frames or an indication of no change between two adjacent frames. If there is much energy, then this is taken as the beginning of a new scene, and used as an area to change group of frames.

Claims

1. A method, comprising:

using a computer for compressing a video formed of multiple frames, said compressing comprising;

segmenting pieces from the video into groups of frames, each group of frames having only information from a single scene, and each group of frames having less than 2^K frames, where K is an integer, said segmenting comprising measuring an energy of a difference between consecutive frames, and when the energy of the difference exceeds a fixed threshold, then defining one of the pieces as belonging to a new group of frames,

said segmenting also finding substantially black frames that have a total amount of energy that is less than a specified threshold representing a substantially black frame and removing said substantially black frames,

to form groups of frames where no group of frames is formed by black frames,

and where each said group of frames only has frames that less than a specified energy difference between them;

applying a transform to said groups of frames that have been extracted by the segmenting and from which the black frames have been removed, to form transformed pieces;

encoding the transformed pieces to form encoded transformed pieces; and

merging the encoded transformed pieces into a bitstream.

2. The method as in claim 1, further comprising identifying frames which have energies which differ from less than one another by a specified amount, and wherein said segmenting marks said frames in a way that prevents said frames from being separately transformed.

3. The method as in claim 1, further comprising extracting the group of frames using said segmenting.

4. The method as in claim 3, further comprising downsampling said video by removing some frames from said group of frames.

5. The method as in claim 1, wherein said applying a transform comprises applying a Poisson Transform which extracts boundary values of a set formed by said pieces, along with a Laplacian function

6. The method as in claim 5, wherein said Poisson transform is compressed by a combination of quantization, wavelet transformation, and using a line representation.

7. A method, comprising:

using a computer for compressing a video formed of multiple frames, said compressing comprising;

segmenting the video into pieces;

applying a transform to said pieces that have been extracted by the segmenting, said transform being a Poisson Transform which extracts boundary values of a frame set formed by said pieces, along with a Laplacian

8. The method as in claim 7, wherein said segmenting comprises finding substantially black frames that have a total amount of energy that is less than a specified threshold representing a substantially black frame, and removing said substantially black frame from the pieces that are formed by the segmenting, and also where said pieces are within frames that have less than a specified energy difference between two consecutive frames.

9. The method as in claim 7, further comprising identifying frames which have energies which differ from less than one another by a specified amount, and wherein said segmenting marks said frames in a way that prevents said frames from being separately transformed.

10. The method as in claim 7, further comprising extracting a group of frames using said segmenting.

11. The method as in claim 10, further comprising downsampling said video by removing some frames from said group of frames.

12. The method as in claim 7, wherein said Poisson transform is compressed by a combination of quantization, wavelet transformation, and using a line representation.

13. The method as in claim 12, wherein said line representation is data that is represented as a list of lines, where each line is represented by at least a center of the line, a length of the line, an angle of the line, and a list of wavelet coefficients for the data sampled along the line.

14. A method, comprising:

using a computer for playing back a video formed of multiple frames, said decompressing said multiple frames, said decompressing comprising;

obtaining a coded portion indicative of a piece of said video;

desegmenting pieces from the video from groups of frames, each group of frames having only information from a single scene, and each group of frames having less than 2^K frames, where K is an integer, said desegmenting comprising receiving frames where an energy of a difference between consecutive frames of a group of frames does not exceed a fixed threshold, and wherein if the energy of a difference between consecutive frames exceeds the difference, then the frames are defined as belonging to a new group of frames,

identifying a first code within said coded portion which represents that one of said frames is substantially black frame and adding a black frame into the decoded video such that no group of frames is formed by black frames, and

not carrying out any decoding during a time represented by said first code, but producing information that is indicative of the black frame based on said first code;

and where each said group of frames only has frames that less than a specified energy difference between them;

decoding transformed groups of frames that have been extracted by the segmenting and from which the black frames have been removed, to form

forming frames which represent decompressed frames including said black framed by detransforming said coded portion and forming a video.

15. The method as in claim 14, further comprising using said computer for finding a second code within said coded portion that identifies frames which have energies which differ from less than one another by less than a specified amount, not carrying out any decoding during a time represented by said second code, but producing information that is indicative of a repeated frame based on said first code without providing information on the frame themselves, and repeating said frames based on said second codes.

16. The method as in claim 14, wherein said decompressing comprises applying a Poisson inverse Transform.