JPS6341253B

PURPOSE:To attain efficient vector quantization suitable for all input signals, by performing vector quantization for a component only corresponding to the basic structure of an input signal, in a quantizer performing quantization by blocking. CONSTITUTION:Plural input signal series are blocked in t...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
1. Verfasser: MURAKAMI ATSUMICHI
Format: Patent
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:PURPOSE:To attain efficient vector quantization suitable for all input signals, by performing vector quantization for a component only corresponding to the basic structure of an input signal, in a quantizer performing quantization by blocking. CONSTITUTION:Plural input signal series are blocked in the lump, and latched to an input vector register 2 as input vector X=(x1-xk). The input vector X is subtracted by a block mean value mA at each element at a parallel subtractor 5 and converted into a DC separation input vecto X'. This input vector X' and a normalized output vector Yi=(yi1-yik) read out sequentially from a normalized output vector code table memory 7 are obtained for the distortion of element at each element through a parallel subtractor 10 and a parallel absolute value operating device 11. The maximum element distortion and the minimum distortion are detected, an index (i) at that time is fetched to an index latch 17, an index 14 having the minimum distortion, a block mean value 15 and a standard deviation are coded in the lump for the purpose of output.