Image recognition apparatus, image recognition method, and program

Patent metadata

Field	Value
Publication number	US-10733477-B2
Application number	US-201715819489-A
Country	US
Kind code	B2
Filing date	Nov 21, 2017
Priority date	Nov 28, 2016
Publication date	Aug 4, 2020
Grant date	Aug 4, 2020

Abstract

In the present disclosure, an image parameter of an input image is changed, features is extracted from each of a plurality of generated images, a category of each region is determined based on the features in each image, and the results are integrated.

First claim

What is claimed is:

1. An image recognition apparatus comprising:
one or more processors; and
a memory coupled to the one or more processors, the memory having stored thereon instructions which, when executed by the one or more processors, cause the apparatus to:
acquire an image;
generate a plurality of variation images by changing a parameter of the acquired image into each of different parameters;
extract features from each of the plurality of variation images;
perform regcognition on each of the plurality of variation images based on the extracted features; and
integrate recognition results of the plurality of variation images into an integrated recognition result,
wherein the recognition results of the plurality of variation images are integrated by using a classifier that has been trained to decrease a difference between an output of the classifier and a supervisory value of learning data.

2. The image recognition apparatus according to claim 1 , wherein the recognition is performed to discriminate categories of regions included in the image.

3. The image recognition apparatus according to claim 1 , wherein the recognition is performed to discriminate types of scenes of the image.

4. The image recognition apparatus according to claim 1 , wherein the recognition is performed to detect a specific object included in the image.

5. The image recognition apparatus according to claim 1 , wherein the recognition is performed to detect a main object included in the image.

6. The image recognition apparatus according to claim 1 , wherein the classifier has learned in advance to further use any of the parameter about an image, camera information during image capturing, or a recognition result of an image as input and a supervisory value as a target value, and to output the integrated recognition result.

7. The image recognition apparatus according to claim 1 , wherein recognition results of the plurality of images are integrated stepwise.

8. The image recognition apparatus according to claim 1 , wherein the change of parameter includes at least one of a change of an exposure value of the image, a change of a brightness value of the image, transformation of the image, clipping of the image, addition of noise to the image, addition of blur to the image, and a change of a focal position of the image.

9. The image recognition apparatus according to claim 1 , wherein the memory further stores instructions which, when executed by the one or more processors, cause the apparatus to acquire a plurality of images by changing the parameters of the images during image capturing.

10. The image recognition apparatus according to claim 1 , wherein the memory further stores instructions which, when executed by the one or more processors, cause the apparatus to acquire a plurality of images by changing the parameters of the images after image capturing.

11. An image recognition method, comprising:
acquiring an image;
generating a plurality of variation images by changing a parameter of the acquired image into each of different parameters;
extracting features from each of the plurality of variation images;
performing recognition on each of the plurality of variation images based on the extracted features; and
integrating recognition results of the plurality of variation images into an integrated recognition result, wherein the recognition results of the plurality of variation images are integrated by using a classifier that has been trained to decrease a difference between an output of the classifier and a supervisory value of learning data.

12. A computer readable storage medium storing a program which causes a computer to execute a method of image recognition, the method comprising;
acquiring an image;
generating a plurality of variation images by changing a parameter of the acquired image into each of different parameters;
extracting features from each of the plurality of variation images;
performing recognition on each of the plurality of variation images based on the extracted features; and
integrating recognition results of the plurality of variation images into an integrated recognition result, wherein the recognition results of the plurality of variation images are integrated by using a classifier that has been trained to decrease a difference between an output of the classifier and a supervisory value of learning data.

Assignees

• Canon Kk

Inventors

• Tate Shunta
• Matsugu Masakazu
• Komori Yasuhiro

Classifications

• G06V10/772 Primary

  Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries · CPC title

• G06F18/214

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

• H04N23/611

  where the recognised objects include parts of the human body · CPC title

• H04N23/80

  Camera processing pipelines; Components thereof · CPC title

• G06F18/28

  Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries · CPC title