Image processor, detection apparatus, learning apparatus, image processing method, and computer program storage medium

• US 10,417,511 B2
• Filed: 09/23/2017
• Issued: 09/17/2019
• Est. Priority Date: 10/07/2016
• Status: Active Grant

• Current Assignee: Panasonic Automotive Systems Company of America (Panasonic Holdings Corporation)
• Original Assignee: Panasonic Intellectual Property Management Co., Ltd. (Panasonic Holdings Corporation)
• Inventors: Arai, Yuko, Arata, Koji
• Primary Examiner(s): Chang, Jon
• Application Number:
  US15/713,604
• Publication Number:
  US 20180101741A1
• Time in Patent Office: 724 Days
• CPC Class Codes: B60Y 2400/3015 Optical cameras
• CPC Class Codes: G06F 18/2321 using statistics or function optimisation, e.g. modelling of probability density functions
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 2207/30268 Vehicle interior
• CPC Class Codes: G06T 7/344 involving models
• CPC Class Codes: G06T 7/35 using statistical methods
• CPC Class Codes: G06T 7/75 involving models
• CPC Class Codes: G06V 10/32 Normalisation of the pattern dimensions
• CPC Class Codes: G06V 20/593 Recognising seat occupancy
• CPC Class Codes: G06V 40/11 Hand-related biometrics; Hand pose recognition
• CPC Class Codes: H04N 23/00 Cameras or camera modules comprising electronic image sensors; Control thereof

Abstract

An image processor includes an image converter. The image converter transforms data of an image that is photographed with a camera for photographing a seat, based on a transformation parameter that is calculated in accordance with a camera-position at which the camera is disposed. The image converter outputs the thus-transformed data of the image. The transformation parameter is a parameter for transforming the data of the image such that an appearance of the seat depicted in the image is approximated to a predetermined appearance of the seat.

Claims

1. A learning apparatus comprising:

an image converter configured to;

acquire data of an image including data of a two-dimensional image depicting a first state of a seat and data of a two-dimensional image depicting a second state of the seat, the two-dimensional first state image and two-dimensional second state image being photographed with a camera;

transform the data of the image, based on a transformation parameter which is calculated in accordance with a camera-position at which the camera is disposed, the transformation parameter being calculated for each of a plurality of camera-positions respectively; and

output the transformed data of the image, and

a learning unit configured to create a learning model based on a feature extracted from the transformed data of the image, wherein

the transformation parameter is a parameter for transforming the data of the image such that an appearance of the seat depicted in the image approximates a predetermined appearance of the seat.

2. The learning apparatus according to claim 1, further comprising a transformation-parameter memory configured to store the transformation parameter,

wherein the image converter is configured to acquire the transformation parameter from the transformation-parameter memory.

3. The learning apparatus according to claim 1, further comprising a transformation-parameter receiver configured to acquire the transformation parameter from an external memory,

wherein the image converter is configured to acquire the transformation parameter from the transformation-parameter receiver.

4. The learning apparatus according to claim 1, wherein the transformation parameter is a parameter for transforming the data of the image such that a coordinate of a predetermined point on the seat depicted in the image matches a predetermined coordinate.

5. The learning apparatus according to claim 1, wherein the transformation parameter is a parameter for transforming the data of the image such that an inclination of a line segment defined based on predetermined points on the seat depicted in the image matches a predetermined inclination.

6. The learning apparatus according to claim 1,

wherein the transformation parameter is calculated in accordance with at least one of an angle of the seat and a position of the seat;

the image converter configured to select the transformation parameter based on data of the camera-position at which the camera is disposed and data of at least one of an angle of the seat and a position of the seat, the angle of the seat and the position of the seat being determined when the camera photographs the image; and

the image converter configured to use the selected transformation parameter for transforming the data of the image.

7. The learning apparatus according to claim 1, further comprising an image memory configured to store, in advance, the data of the image, and to output the stored data of the image to the image converter.

8. The learning apparatus according to claim 1, further comprising an image receiver configured to receive the data of the image from an image memory, and to output the received data of the image to the image converter.

9. The learning apparatus according to claim 1, wherein the first state is a state of a person in the seat, and the second state is a state with no person in the seat.

10. The learning apparatus according to claim 1, wherein the learning model is based upon positions of cameras for different vehicles, the positions being stored in a memory.

11. A learning method comprising:

acquiring data of an image including both data of a two-dimensional image depicting a first state of a seat and data of a two-dimensional image depicting a second state of the seat, the two-dimensional first state image and two-dimensional second state image being photographed with a camera;

transforming the data of the image, based on a transformation parameter calculated in accordance with a camera-position at which the camera is disposed, the transformation parameter being calculated for each of a plurality of camera-positions respectively;

outputting the transformed data of the image, and

creating a learning model based on a feature extracted from the transformed data of the image,

wherein the transformation parameter is a parameter for transforming the data of the image such that an appearance of the seat depicted in the image approximates a predetermined appearance of the seat.

12. The learning method according to claim 11, wherein the first state is a state of a person in the seat, and the second state is a state with no person in the seat.

13. The learning method according to claim 11, the creating of the learning model is based upon positions of cameras for different vehicles, the positions being stored in a memory.

14. A nonvolatile storage medium storing a learning program causing a computer to execute:

acquiring data of an image including both data of a two-dimensional image depicting a first state of a seat and data of a two-dimensional image depicting a second state of the seat, the two-dimensional first state image and two-dimensional second state image being photographed with a camera for;

transforming the data of the image, based on a transformation parameter calculated in accordance with a camera-position at which the camera is disposed, the transformation parameter being calculated for each of a plurality of camera-positions respectively;

outputting the transformed data of the image, and

creating a learning model based on a feature extracted from the transformed data of the image,

wherein the transformation parameter is a parameter for transforming the data of the image such that an appearance of the seat depicted in the image approximates a predetermined appearance of the seat.

15. The non-volatile storage medium according to claim 14, wherein the first state is a state of a person in the seat and the second state is a state with no person in the seat.

16. The non-volatile storage medium according to claim 14, wherein the learning model is created based upon positions of cameras for different vehicles, the positions being stored in a memory.