Content based video compression system

• US 5,854,856 A
• Filed: 07/19/1995
• Issued: 12/29/1998
• Est. Priority Date: 07/19/1995
• Status: Expired due to Term

• Current Assignee: Carnegie Mellon University
• Original Assignee: Carnegie Mellon University
• Inventors: Jasinschi, Radu S., Moura, Jose M. F.
• Primary Examiner(s): Couso, Jose L.
• Assistant Examiner(s): Do, Anh Hong
• Application Number:
  US08/504,379
• Time in Patent Office: 1,259 Days
• Field of Search: 382/236, 382/232, 348/415, 348/416
• US Class Current: 382/232
• CPC Class Codes: G06T 9/20 Contour coding, e.g. using detection of edges
• CPC Class Codes: H04N 19/20 using video object coding
• CPC Class Codes: H04N 19/23 with coding of regions that are present throughout a whole video segment, e.g. sprites, background or mosaic

Abstract

A content based method of compressing a segment of video is implemented in two stages. In a spatial integration stage, figures are represented in terms of compact models. In a temporal integration stage, which uses the information from the spatial integration stage, constructs, i.e., world images and data describing relationships between world images and frames, are generated. In operation, each frame in a series of frames is preprocessed to tessellate any moving figures and to obtain motion data for the moving figures and for the image background. The figure motion data is stored in a manner so as to associate the motion data with the original frame. Frame information identifying the size and position of each frame with respect to a background world image is also stored. Each tessellated figure is compared to the original frame to produce a template for each moving figure and a template for the background. Each template is compared to a world image that is associated with that template so that material common to both can be cut and the remaining material added to form a new world image. The steps are repeated until all frames in the segment are processed. The resulting world image data, figure motion data, and frame information are output for transmission or storage. The world image data may be compressed prior to transmission or storage.

Claims

1. A method of compressing a video segment comprised of a series of frames, comprising the steps of:

preprocessing a frame to obtain figure motion data for each moving figure, said preprocessing comprising the steps of;

comparing said frame to another frame;

estimating image velocity at certain locations;

estimating background velocity of said frame;

adjusting said image velocity to compensate for said background velocity;

segmenting each said moving figure; and

estimating figure velocity of each segmented moving figure of said frame;

tessellating any moving figures to produce an instance of a template for each moving figure and an instance of a background template;

storing said figure motion data in a manner so as to associate said figure motion data with said frame;

storing frame information representative of the size and position of said frame with respect to a background world image;

comparing each of said instances of a template to a world image for that template to cut material common to both from one and to add the remaining material to form a new world image;

repeating each step until all the frames in the series are processed; and

outputting each world image, said figure motion data, and said frame information.

2. The method of claim 1 additionally comprising the step of compressing each world image before said outputting step.

3. The method of claim 1 wherein said step of storing said figure motion data includes the step of storing data identifying the coordinates of a reference point for the figure, the size of the figure, and the position of the figure.

4. The method of claim 1 wherein said step of tessellating any moving figures includes the step of tessellating any moving figures using a rectangular approximation technique.

5. The method of claim 1 additionally comprising the step of using said motion data to register each said instance of a template with said world image for that template prior to cutting material common to both from one.

6. The method of claim 5 wherein said step of cutting material common to both from one includes the step of cutting the material common to both from said world image.

7. The method of claim 5 wherein said step of cutting material common to both from one includes the step of cutting the material common to both from said instance of a template.

8. The method of claim 5 further comprising the step of equalizing an illumination level of said world image with an illumination level of said template prior to cutting material common to both from one.

9. The method of claim 1 additionally comprising the step of preprocessing each frame to produce stacking information which indicates whether any moving figures occlude other moving figures, and wherein said stacking information is stored and output with said motion data.

10. The method of claim 1 wherein said outputting step includes the step of outputting each world image, said figure motion data, and said frame information for storage.

11. The method of claim 2 additionally comprising the steps of decompressing each world image and of reconstructing said frame from said decompressed world images, said motion data, and said frame information.

12. The method of claim 1 wherein said world images are generated based on content based methods.

13. An apparatus for compressing a video segment comprised of a series of frames, comprising:

means for preprocessing a frame to obtain figure motion data for each moving figure, said means for preprocessing comprising;

means for comparing said frame to another frame;

means for estimating image velocity at certain locations;

means for estimating background velocity of said frame;

means for adjusting said image velocity to compensate for said background velocity;

means for segmenting each said moving figure; and

means for estimating figure velocity of each segmented moving figure of said frame;

means for tessellating any moving figures to produce an instance of a template for each moving figure and an instance of a background template;

first means for storing said figure motion data in a manner so as to associate said figure motion data with said frame;

second means for storing frame information representative of the size and position of said frame with respect to a background world image; and

means for comparing each of said instances of a template to a world image for that template, for cutting material common to both from one, and for adding the remaining material to form a new world image.