Browse Prior Art Database

Compression of Video Data via Three Dimensional Bandwidth Reduction

IP.com Disclosure Number: IPCOM000088056D
Original Publication Date: 1977-Apr-01
Included in the Prior Art Database: 2005-Mar-04
Document File: 3 page(s) / 48K

Publishing Venue

IBM

Related People

Nadas, AJ: AUTHOR

Abstract

This article describes a technique for data compression in the transmission of digitized video data.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 3

Compression of Video Data via Three Dimensional Bandwidth Reduction

This article describes a technique for data compression in the transmission of digitized video data.

Transform image coding has been previously used for data compression of two-dimensional video images. In the present approach, a three-dimensional transform, similar to the two-dimensional singular value decomposition transform [*] is used to remove spatial, but also temporal, redundancy from a sequence of video images. In general, the present approach provides for a volumetric reduction of inherently multidimensional numerical arrays.

In particular, video data, e.g., motion pictures can be thus compressed for storage in a video disk or cassette system.

The following system may be used as a stand-alone compression device or in tandem with other compression devices such as difference techniques. While the description below is in terms of three dimensions suitable for video data, the method works equally well in higher dimensions. Let X = (x(ijk); i = 1, ..., I, j = 1,...,j, k = 1,...,K). denote the three-dimensional array formed by K consecutive I by J frames of video data (see figure). Each frame is an I by J matrix of numbers, usually "gray levels" from a logarithmically incremented set of possible values. This kind of data is inherently three-dimensional because adjacent video frames are very much alike. The volume of X is I.J.K. The compression consists of replacing X by an approximation X such that X can be constructed (in a simple way) from approximately 100 M.P.Q/I.J.K % of original volume. The approximation is constructed as follows.

Step 1 - Compute three "Marginal Gramian" matrices G(I) = (Sigma/J/(j = 1) Sigma/K/(k - 1) x(i(1) jk) x(i(2)jk;...