Method of compressing image signals using vector quantization

• US 4,853,778 A
• Filed: 02/25/1988
• Issued: 08/01/1989
• Est. Priority Date: 02/25/1987
• Status: Expired due to Term

• Current Assignee: Fujifilm Corporation (Fujifilm Holdings Corporation)
• Original Assignee: Fuji Photo Film Co Limited (Fujifilm Holdings Corporation)
• Inventors: Tanaka, Nobuyuki
• Primary Examiner(s): Britton, Howard W.
• Assistant Examiner(s): Kostak, Victor R.
• Application Number:
  US07/160,377
• Time in Patent Office: 523 Days
• Field of Search: 358/133, 358/135, 358/136, 358/260, 358/262
• US Class Current: 348/422.1
• CPC Class Codes: G06T 9/008 Vector quantisation
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/94 Vector quantisation

Abstract

A method of compressing image signals comprises the steps of dividing original image signals representing a two-dimensional image into blocks each of which is composed of a plurality of samples, converting the original image signals at respective picture elements in each of the blocks into n-valued system where n denotes an integer of 2 or larger, calculating means values m₁, m₂, . . . , m_n of the original image signals at the respective picture elements in the block that have the common n-valued signals, and encoding the mean values m₁, m₂, . . . , m_n and the n-valued signals. The mean values m₁, m₂, . . . , m_n and/or the n-valued signals in each of the blocks are then subjected to vector quantization.

Claims

1. A method of compressing image signals by:

(i) dividing original image signals representing a two-dimensional image into blocks, each of said blocks being composed of a plurality of samples,

(ii) converting the original image signals at respective picture elements in each of the blocks into an n-valued system where each of the plurality of samples has one of the n-values and, where n denotes an integer of 2 or larger,

(iii) calculating mean values m₁, m₂, . . . , m_n of the original image signals at the respective picture element in the block that have the common n-valued signals, and

(iv) encoding the mean values m₁, m₂, . . . , m_n and the n-valued signals,

wherein the improvement comprises the steps of;

(a) carrying out vector quantization of said mean values m₁, m₂, . . . m_n and/or said n-valued signals in each of said blocks.

2. A method as defined in claim 1 wherein said vector quantization comprises the step of selecting a vector that corresponds with the minimum distortion to the set of said mean values m₁, m₂, . . . , m_n in each of said blocks from a code book comprising a plurality of vectors different from one another and each prepared in advance by defining "n" number of vector elements.

3. A method as defined in claim 1 or 2 wherein said vector quantization comprises the step of selecting a vector that corresponds with the minimum distortion to the set of said n-valued signals in each of said blocks from a code book comprising a plurality of vectors different from one another and each prepared in advance of defining vector elements in a number equal to the number of said samples in each of said blocks.