Signal processor and signal processing method

Patent metadata

Field	Value
Publication number	US-2018096218-A1
Application number	US-201715711295-A
Country	US
Kind code	A1
Filing date	Sep 21, 2017
Priority date	Sep 30, 2016
Publication date	Apr 5, 2018
Grant date	—

Abstract

A signal processing method includes addition of noise obtained by multiplying generated random number by K to the input pixel signal I(x), a binarization process of comparing the result of the addition with two thresholds, and a process of calculating a probability. The binarization process includes a first nonlinear process and a second nonlinear process. The first nonlinear process outputs “P” in a case where I(x) after the addition of the noise is greater than the threshold T1 and less than the second threshold T2. The second nonlinear determines “1” or “0” for a processing target pixel, in which the result of the first nonlinear process is “P,” based on input pixel signals of pixels around the processing target pixel. The process of calculating a probability calculates a probability J(x) that the result of the first nonlinear process is “1,” or the result of the first nonlinear process is “P” and the result of the second nonlinear process is “1”.

First claim

What is claimed is:

1 . A signal processor comprising:
an acquisition unit configured to acquire input pixel signals I(x) corresponding to two-dimensionally arranged pixels (x), respectively; and a processing unit configured to apply a predetermined stochastic resonance process to each of the input pixel signals I(x) acquired by the acquisition unit, wherein the predetermined stochastic resonance process includes addition of noise obtained by multiplying generated random number by K to the input pixel signal I(x), a binarization process of comparing the result of the addition of the noise to the input pixel signal I(x) with a threshold T2 and a threshold T1 (where T2>T1), and a process of calculating a probability related to the result of the binarization process, the binarization process includes: a first nonlinear process of outputting “0” in a case where the input pixel signal I(x) after the addition of the noise is less than the threshold T1, “1” in a case where the input pixel signal I(x) after the addition of the noise is greater than the threshold T2, and “P” in a case where the input pixel signal I(x) after the addition of the noise is greater than or equal to the threshold T1 and less than or equal to the second threshold T2; and a second nonlinear process of determining whether to output “1” or “0” for a processing target pixel, in which the result of the first nonlinear process is “P,” based on input pixel signals of pixels around the processing target pixel, the second nonlinear process being subsequent to the first nonlinear process, and the process of calculating a probability is a process of calculating a probability J(x) that the result of the first nonlinear process is “1,” or the result of the first nonlinear process is “P” and the result of the second nonlinear process is “1” in the binarization process.

2 . The signal processor according to claim 1 , wherein the process of calculating the probability J(x) includes a process of using the following formula for the input pixel signal I(x) as a process of obtaining a probability Rate1 that the result of the first nonlinear process is “1” in a case where the random number N is greater than 0 and less than 1 and a probability that the random number N is generated is f(N):

Rate



1

=

{

1

Pixel



where



A



2

<
0

0

Pixel



where



A



2

>
1

1
-

∫

N
=
0

A



2



f


(
N
)




dN

Pixel



where



0

≤

A



2

≤
1




where



A



2

=

{

T


Assignees

• Canon Kk

Inventors

• Tanaka Wakako
• Ishikawa Tomokazu
• Mizoguchi Yoshinori

Classifications

• G06V10/28 Primary

  Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns · CPC title

• G06F2218/04

  Denoising · CPC title

• G06V30/40

  Document-oriented image-based pattern recognition · CPC title

• G06V10/758

  Involving statistics of pixels or of feature values, e.g. histogram matching · CPC title

• G06T7/0002

  Inspection of images, e.g. flaw detection · CPC title