# US 2009/0268964 A1 · Pattern recognizing apparatus, pattern recognizing method and pattern recognizing program

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2009/0268964 A1
- Document ID: US20090268964A1
- Status: Published application
- Kind: A1
- Application number: 12/093,407
- Filed: 2006-11-15
- Published: 2009-10-29
- Earliest claimed: 2005-11-29
- Assignees: NEC Corporation
- Inventors: Katsuhiko Takahashi
- Web page: https://patentgrep.com/patent/US20090268964A1

## CPC

- G06V 30/162: Quantising the image signal
- G06V 30/18095: {Summing image-intensity values; Projection and histogram analysis}
- G06V 30/10: Character recognition
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A pattern recognizing method for matching a binary or ternary reference image and an input grayscale image, having a precision similar to the method in which a plurality of binary reference images are generated by supposing settable various binarization threshold value and comparing the distances between the plurality of binary reference images and the grayscale image, and enabling the processing time to be shorter. The first accumulated histogram of the pixels of the grayscale image corresponding to a high density region of the binary reference image and the second accumulated histogram opposite to it and of the pixels of the grayscale image corresponding to the low density region of the reference image are generated and summated. The existence or absence of a pattern is judged by comparing the minimum value of the summated accumulated histogram and a threshold value.

## Claims

### Claim at position 1

1-18. (canceled)

### Claim 19

A pattern recognizing apparatus comprising:
a storing section configured to store a reference image being binarized and having a high density region and a low density region and an input grayscale image to be matched with the reference image;
a first accumulated histogram extracting section configured to extract a first accumulated histogram generated by accumulating pixels whose density value is less than a given value among pixels of the input grayscale image and corresponding to the high density region of the reference image;
a second accumulated histogram extracting section configured to extract a second accumulated histogram generated by accumulating pixels whose density value is more than a given value among pixels of the input grayscale image and corresponding to the low density region of the reference image;
an accumulated histogram summating section configured to generate a summated accumulated histogram by summating the first accumulated histogram and the second accumulated histogram; and
a pattern matching section configured to recognize the input grayscale image by using a distance value of the reference image and the input grayscale image which is a minimum value of the summated histogram generated by the accumulated histogram summating section.

### Claim 20 (depends on 19)

The pattern recognizing apparatus according to claim 19, wherein the storing section stores a weighting image having weighting values for every pixel;
the first accumulated histogram extracting section extracts the first accumulated histogram by accumulating the weighting values of the weighting image corresponding to pixels in the high density region of the reference image and whose density values are larger than a given value; and
the second accumulated histogram extracting section extracts the second accumulated histogram by accumulating the weighting values of the weighting image corresponding to pixels in the low density region of the reference image and whose density values are smaller than a given value.

### Claim 21 (depends on 19)

The pattern recognizing apparatus according to claim 19, wherein the reference image includes a third density region as a non-calculation region,
the first accumulated histogram extracting section excludes pixels in the third density region from the extracting of the first accumulated histogram, and
the second accumulated histogram extracting section excludes pixels in the third density region from the extracting of the second accumulated histogram.

### Claim 22 (depends on 19)

The pattern recognizing apparatus according to claim 19, further comprising:
a region setting section configured to set a partial region image from an original image; and
a grayscale image scaling section configured to scale the partial region image to become a same image as the reference image,
wherein the grayscale scaling section sets the scaled partial region image as the input grayscale image to be matched with the reference image.

### Claim 23

A pattern recognizing method comprising:
(a) storing a reference image being binarized and having a high density region and a low density region and an input grayscale image to be matched with the reference image;
(b) extracting a first accumulated histogram generated by accumulating pixels whose density value is less than a given value among pixels of the input grayscale image and corresponding to the high density region of the reference image;
(c) extracting a second accumulated histogram generated by accumulating pixels whose density value is more than a given value among pixels of the input grayscale image and corresponding to the low density region of the reference image;
(d) generating a summated accumulated histogram by summating the first accumulated histogram and the second accumulated histogram; and
(e) recognizing the input grayscale image by using a distance value of the reference image and the input grayscale image which is a minimum value of the summated histogram.

### Claim 24 (depends on 23)

The pattern recognizing method according to claim 23, further comprising:
(f) storing a weighting image having weighting values for every pixel,
wherein at (b) extracting, the first accumulated histogram is extracted by accumulating the weighting values of the weighting image corresponding to pixels in the high density region of the reference image and whose density values are larger than a given value; and
at (c) extracting, the second accumulated histogram is extracted by accumulating the weighting values of the weighting image corresponding to pixels in the low density region of the reference image and whose density values are smaller than a given value.

### Claim 25 (depends on 23)

The pattern recognizing method according to claim 23, wherein the reference image includes a third density region as a non-calculation region,
at (b) extracting the first accumulated histogram, pixels in the third density region are excluded from the extracting of the first accumulated histogram, and
at (c) extracting the second accumulated histogram, pixels in the third density region are excluded from the extracting of the second accumulated histogram.

### Claim 26 (depends on 23)

The pattern recognizing method according to claim 23, further comprising:
(f) setting a partial region image from an original image; and
(g) scaling the partial region image to become a same image as the reference image,
wherein the scaled partial region image is set as the input grayscale image to be matched with the reference image.

### Claim 27

A computer program product embodied on a computer-readable medium and comprising code that, when executed, causes a computer to perform the following:
(a) storing a reference image being binarized and having a high density region and a low density region and an input grayscale image to be matched with the reference image;
(b) extracting a first accumulated histogram generated by accumulating pixels whose density value is less than a given value among pixels of the input grayscale image and corresponding to the high density region of the reference image;
(c) extracting a second accumulated histogram generated by accumulating pixels whose density value is more than a given value among pixels of the input grayscale image and corresponding to the low density region of the reference image;
(d) generating a summated accumulated histogram by summating the first accumulated histogram and the second accumulated histogram; and
(e) recognizing the input grayscale image by using a distance value of the reference image and the input grayscale image which is a minimum value of the summated histogram.

### Claim 28 (depends on 27)

The computer program product according to claim 27, further causes the computer to perform the following:
(f) storing a weighting image having weighting values for every pixel,
wherein at (b) extracting, the first accumulated histogram is extracted by accumulating the weighting values of the weighting image corresponding to pixels in the high density region of the reference image and whose density values are larger than a given value; and
at (c) extracting, the second accumulated histogram is extracted by accumulating the weighting values of the weighting image corresponding to pixels in the low density region of the reference image and whose density values are smaller than a given value.

### Claim 29 (depends on 27)

The program according to claim 27, wherein the reference image includes a third density region as a non-calculation region,
at (b) extracting the first accumulated histogram, pixels in the third density region are excluded from the extracting of the first accumulated histogram, and
at (c) extracting the second accumulated histogram, pixels in the third density region are excluded from the extracting of the second accumulated histogram.

### Claim 30 (depends on 27)

The program according to claim 27, further causes the computer to perform the following:
(f) setting a partial region image from an original image; and
(g) scaling the partial region image to become a same image as the reference image,
wherein the scaled partial region image is set as the input grayscale image to be matched with the reference image.
