Signal processing apparatus, signal processing method, and storage medium

Patent metadata

Field	Value
Publication number	US-10715192-B2
Application number	US-201715470023-A
Country	US
Kind code	B2
Filing date	Mar 27, 2017
Priority date	Mar 31, 2016
Publication date	Jul 14, 2020
Grant date	Jul 14, 2020

Abstract

A signal processing apparatus includes a unit configured to generate noise cut data by deducting a predetermined noise value from values of respective signals constituting input data and a stochastic resonance processing unit configured to subject the noise cut data to a predetermined stochastic resonance processing. The predetermined stochastic resonance processing is processing to output, in a method of synthesizing a result of parallelly performing steps of adding new noise to the noise cut data to subject the resultant data to a binary processing, a value obtained in a case where the parallel number is infinite.

First claim

What is claimed is:

1. A signal processing apparatus comprising:
circuitry or at least one processor, configured to act as a plurality of units comprising:
(1) an acquisition unit configured to acquire input data;
(2) a unit configured to generate noise cut data by deducting uniformly a predetermined fixed value from values of respective signals of the input data;
(3) a stochastic resonance processing unit configured to subject the noise cut data to a predetermined stochastic resonance processing to thereby generate stochastic resonance data; and
(4) an output unit configured to output the result of the predetermined stochastic resonance processing,

wherein the predetermined stochastic resonance processing is processing to output, in a method of adding different noises to the same noise cut data and performing binary processing in a plurality of branch paths respectively, a convergence value obtained in a case where a parallel number is increased,
wherein the predetermined stochastic resonance processing is performed by using the following formula to calculate processed data J(x) from the input data I(x):

J
⁡

(
x
)

=

{

1

T
<

I
⁡

(
x
)

0

I
⁡

(
x
)

<

T
-
K

1
-

(

T
-

1
⁢

(
x
)

)

/
K

T
-
K

≦

I
⁡

(
x
)

≦
T

where T is a threshold value to quantize the input data and K is a noise strength,
or the predetermined stochastic resonance processing is performed by using the following formula to calculate processed data J(x) from the input data I(x):

J
⁡

(
x
)

=

{

1

T
<

I
⁡

(
x
)

0

I
⁡

(
x
)

<

T
-
K

1
-

1

1
+

exp
⁢

{

-

α
⁡

(

(

T
-

I
⁡

(
x
)

)

K

-
0.5

)

}

Assignees

• Canon Kk

Inventors

• Suwa Tetsuya
• Ishikawa Tomokazu
• Tanaka Wakako
• Hori Shinjiro

Classifications

• H04N1/409 Primary

  Edge or detail enhancement; Noise or error suppression · CPC title

• G06V10/30

  Noise filtering · CPC title

• G06F2218/16

  by matching signal segments · CPC title

• G06F2218/04

  Denoising · CPC title

• H04B1/10 Primary

  Means associated with receiver for limiting or suppressing noise or interference · CPC title