Information processing apparatus, control method, and storage medium that generate learning data used in learning a classification rule for classifying an input image

Patent metadata

Field	Value
Publication number	US-9602685-B2
Application number	US-201514810514-A
Country	US
Kind code	B2
Filing date	Jul 28, 2015
Priority date	Aug 7, 2014
Publication date	Mar 21, 2017
Grant date	Mar 21, 2017

Abstract

An information processing apparatus that generates learning data used in learning a classification rule for classifying an image that is input from an image input device. A reception unit receives an initial image that is input from the image input device. A first generation unit generates an image that is different from the initial image, by correcting the initial image using a coefficient determined based on information about a type of the input device. A second generation unit generates the learning data using the image generated by the first generation unit and the input image. A learning unit generates the classification rule using the learning data generated by the second generation unit.

First claim

What is claimed is:

1. An information processing apparatus that generates learning data used in learning a classification rule for classifying an image that is input from an image input device, the apparatus comprising:
at least one processor operatively coupled to a memory, serving as:
(a) a reception unit configured to receive an initial image that is input from the image input device;
(b) a first generation unit configured to generate an image that is different from the initial image, by correcting the initial image using a coefficient determined based on information about a type of the input device;
(c) a second generation unit configured to generate the learning data using the image generated by the first generation unit and the input image; and
(d) a learning unit for generating the classification rule using the learning data generated by the second generation unit.

2. The information processing apparatus according to claim 1 , wherein
the coefficient is a correction parameter for correcting an image depending on the characteristic of an image input device, and
the first generation unit generates the image that is different from the initial image by performing correction processing of an image using the correction parameter.

3. The information processing apparatus according to claim 2 , wherein the correction processing of an image using the correction parameter includes at least one of rotation processing, shift processing, scaling processing, and blurring processing of an image.

4. The information processing apparatus according to claim 3 , wherein, if the correction processing of an image using the correction parameter includes rotation processing of the image, the parameter of the rotation processing is a rotation amount, and the first generation unit determines:
(i) if the image input device is an image processing apparatus, the rotation amount based on a standard for skew of an image reading unit included in the image processing apparatus, and
(ii) if the image input device is a mobile terminal, the rotation amount based on information on image deterioration caused by original copy extraction from an image photographed by the mobile terminal.

5. The information processing apparatus according to claim 3 , wherein the parameter of a shift processing is a shift amount if the correction processing of an image using the correction parameter includes the shift processing of the image, and the first generation unit determines:
(i) if the image input device is an image processing apparatus, the rotation amount based on a standard for the positional shift of an image reading unit included in the image processing apparatus, and
(ii) if the image input device is a mobile terminal, the shift amount based on information on image deterioration caused by original copy extraction from an image photographed by the mobile terminal.

6. The information processing apparatus according to claim 3 , wherein the parameter of scaling processing is a scaling rate if the correction processing of an image using the correction parameter includes the scaling processing of the image, and the first generation unit determines:
(i) if the image input device is an image processing apparatus, the scaling rate based on the standard about the reading magnification of an image reading unit included in the image processing apparatus, and
(ii) if the image input device is a mobile terminal, the scaling rate based on information on image deterioration caused by original copy extraction from an image photographed by the mobile terminal.

7. The information processing apparatus according to claim 3 , wherein if the correction processing of an image using the correction parameter includes blurring processing of the image, the blurring processing is processing for generating a blurred image depending on blurring and shaking amounts determined by a kernel size of a Gaussian filter, the parameter of the blurring processing being kernel size, and the first generation unit determines:
(i) if the image input device is an image processing apparatus, the blurring amount and thus the kernel size based on corresponding information between a Multi Function Peripheral (MTF) characteristic of a line sensor included in the image processing apparatus and a blurring amount, and
(ii) if the image input device is a mobile terminal, the blurring amount and thus the kernel size based on corresponding information between the MTF characteristic of a lens included in the mobile terminal and a blurring amount.

8. The information processing apparatus according to claim 1 , wherein the learning data is a feature amount of an image generated by the first generation unit.

9. A method of controlling an information processing apparatus that generates learning data used in learning a classification rule for classifying an image that is input from an image input device, the method comprising:
receiving an initial image that is input from the image input device;
generating, in a first generation step, an image that is different from the initial image, by correcting the initial image using a coefficient determined based on information about a type of the input device;
generating, in a second generation step, the learning data using the image generated in the first generation step and the input image; and
generating, in a learning step, the classification rule using the learning data generated in the second generation step.

10. The method of controlling an information processing apparatus according to claim 9 , wherein the learning data is a feature amount of an image generated by the first generation unit.

11. A non-transitory storage medium storing a readable program for causing a computer to execute a controlling method executed in an information processing apparatus that generates learning data used in learning a classification rule for classifying an image that is input from an image input device, the method comprising:
receiving an initial image that is input from the image input device;
generating, in a first generation step, an image different from the initial image, by correcting the initial image using a coefficient determined based on information about a type of the input device; and
generating, in a second generation step, the learning data using the image generated in the first generation step and the input image; and
generating, in a learning step, the classification rule using the learning data generated in the second generation step.

12. The method of controlling an information processing apparatus according to claim 9 , wherein
the coefficient is a correction parameter for correcting an image depending on the characteristic of an image input device, and
the first generation step generates the image that is different from the initial image by performing correction processing of an image using the correction parameter.

13. The method of controlling an information processing apparatus according to claim 12 , wherein the correction processing of an image using the correction parameter includes at least one of rotation processing, shift processing, scaling processing, and blurring processing of an image.

14. The method of controlling an information processing apparatus according to claim 13 , wherein, if the correction processing of an image using the correction parameter includes rotation processing of the image, the parameter of the rotation processing is a rotation amount, and the first generation step determines:
(i) if the image input device is an image processing apparatus, the rotation amount based on a standard for skew of an image reading unit included in th

Assignees

• Canon Kk

Inventors

• Miyauchi Takashi
• Arakawa Junya

Classifications

• H04N1/00336 Primary

  with an apparatus performing pattern recognition, e.g. of a face or a geographic feature (image or video recognition or understanding of scenes G06V20/00) · CPC title

• G06V30/19147 Primary

  Obtaining sets of training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06V30/40

  Document-oriented image-based pattern recognition · CPC title

• G06K9/00442

  Physics · mapped topic