METHOD FOR DATA REDUCTION OF DIGITAL PICTURE SIGNALS BY VECTOR QUANTIZATION OF COEFFICIENTS ACQUIRED BY ORTHONORMAL TRANSFORMATION BY MEANS OF A SYMMETRICAL, NEARLY CYCLICAL HADAMARD MATRIX

Patent №

US 4,772,946

Granted

1988-09-20

Filed 1986

Owner

SIEMENS AKTIENGESELLSCHAFT, A GERMAN CORP.

Lab

AI components

1

vision

Assignment

Recorded

Dataset

AIPD

2023_r1 edition

Application

06938143

A method and apparatus for data reduction of digital picture signals by vector quantizing of coefficient acquired by orthonormal transformation using a symmetrically, nearly cyclical Hadamard matrix, includes means for deriving a mean value coefficient and a plurality of structure coefficients from a plurality of blocks formed from the input signal and resorted by means of a Peano curve, said coefficients being in part only scalarly quantized, and in part scalarly and vectorly quantized. The apparatus and method allows a noise amplitude of the picture signal acquired after back transformation to be adapted to visibility thresholds of the human visual system according to CCIR recommendation 500-1.

AI classification

Vision1.00
AI hardware0.23
Speech0.04
Machine learning0.00
Evolutionary computation0.00
Planning0.00
Natural language0.00
Knowledge representation0.00

Ownership

SIEMENS AKTIENGESELLSCHAFT, A GERMAN CORP.

assignment · 46400214

Assignors

HAMMER BERNARD

On an employer assignment, the assignors are typically the inventors.

© 2026 NYSGPT2525 LLC