Quantizer designed by using human visual sensitivity

• US 5,557,276 A
• Filed: 07/27/1994
• Issued: 09/17/1996
• Est. Priority Date: 08/13/1993
• Status: Expired due to Term

• Current Assignee: KDDI Corporation
• Original Assignee: Kokusai Densin Denwa Kabushiki Kaisha
• Inventors: Matsumoto, Shuichi, Hamada, Takahiro, Sakazawa, Shigeyuki
• Primary Examiner(s): Young, Brian K.
• Application Number:
  US08/281,294
• Time in Patent Office: 783 Days
• Field of Search: 341/200, 348/405, 395/131, 395/155, 395/161
• US Class Current: 341/200
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/126 Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
• CPC Class Codes: H04N 19/527 Global motion vector estimation
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/86 involving reduction of coding artifacts, e.g. of blockiness

Abstract

A quantizer designed by using human visual sensitivity is disclosed, in which when postprocessing is done after inverse quantization an optimum visual sensitivity function for a system involving the postprocessing is obtained for doing quantization by using that visual sensitivity function.

Where the postprocessing that is executed after the inverse quantization consists of a filtering process with a characteristic L(ω) (ω being the frequency) and a 1:2 interpolation process, denoting the optimum visual sensitivity as the output result by W(ω) as shown in FIG. 2, the optimum equivalent sensitivity function W'"'"'(ω) is given as

W'"'"'(ω)=1/2[L(ω/2)W(ω/2)+L(ω/2+π)W(ω/2+π)

].

By using this equivalent sensitivity function W'"'"'(ω) for the quantizer, it is possible to obtain a visually optimum quantizer designed by using human visual sensitivity.

Claims

1. In a quantizer used for a picture signal coder in a system, in which postprocessing is carried out in a decoding process,

a quantizer designed by using human visual sensitivity, in which a human visual frequency sensitivity W1(ω

) (ω

being the angular frequency) equivalent in function to the human visual frequency sensitivity W(ω

) with respect to picture after decoding is obtained on the basis of the postprocessing in the decoding process, the obtained human visual frequency sensitivity W1(ω

) being used as the visual sensitivity prior to the decoding for quantization.

2. The quantizer designed by using human visual sensitivity according to claim 1, wherein:

when the human visual frequency sensitivity after the decoding is W(ω

) and the postprocessing in the decoding process is constituted by a 1;

M (M being a positive integer) interpolation process, the visual sensitivity W2(ω

) before the decoding process is obtained using an equation ##EQU1##

3. The quantizer designed by using human visual sensitivity according to claim 1, wherein:

when the human visual frequency sensitivity after the decoding process is W(ω

) and the postprocessing in the decoding process is constituted by a filtering process with a filter characteristic H(ω

), the visual sensitivity W3(ω

) before the decoding process is obtained using an equation

space="preserve" listing-type="equation">W3(ω

)=H(ω

)W(ω

).

4. The quantizer designed by using human visual sensitivity according to claim 1, wherein:

when the human visual frequency sensitivity after the decoding process is W(ω

) and the postprocessing in the decoding process is constituted by a 1;

M (M being a positive integer) interpolation process and a filtering process with a filter characteristic H(ω

), the visual sensitivity W1(ω

) before the decoding is obtained as a combination of the following visual sensitivities W2(ω

) and W3(ω

), as a result being W4(ω

). ##EQU2##

5. In a quantizer for a system, in which a picture signal is divided into a plurality of frequency components and also in which a decoding process is constituted by a 1:

M (M being a positive integer) interpolation process and/or a process with a filter H(ω

),

a quantizer designed by using human visual sensitivity, wherein the step size Δ

k of the quantizer with respect to the k-th (k being a positive integer) frequency component is such as to minimize the sum of the results of weighting of the mean square noise by the visual sensitivity for the individual frequency components under the condition of a constant coded information amount.