Target recognizing device and target recognizing method

• US 7,298,907 B2
• Filed: 02/13/2002
• Issued: 11/20/2007
• Est. Priority Date: 02/19/2001
• Status: Active Grant

• Current Assignee: Honda Giken Kogyo Kabushiki Kaisha (Honda Motor Company)
• Original Assignee: Honda Giken Kogyo Kabushiki Kaisha (Honda Motor Company)
• Inventors: Hasegawa, Yuji
• Primary Examiner(s): LE, BRIAN Q
• Application Number:
  US10/468,229
• Publication Number:
  US 20040066952A1
• Time in Patent Office: 2,106 Days
• Field of Search: 382/103, 382/107, 382/165, 382/170, 382/209, 382/215, 382/216, 382/217, 382/218, 382/219, 382220-223
• US Class Current: 382/209
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06T 7/215 Motion-based segmentation
• CPC Class Codes: G06T 7/269 using gradient-based methods
• CPC Class Codes: G06V 10/772 Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries

Abstract

A target recognizing device that creates a template concerning a target by itself is provided. The device comprises a robot camera for creating continuous images showing a target and an arithmetic operation unit for controlling the robot camera. A robot camera acquires continuous image information by means of a camera rotatable by a motor. The arithmetic operation unit comprises a template management section for allowing an update part to update a template stored in a template storage part and a target recognition section having an image storage part where the acquired image information is stored, a recognizing part for recognizing a target by using the template, and a motion command part for rotating a camera according to the result of recognition.

Claims

1. A target recognizing device comprising:

an image acquiring device which acquires images including a target, while performing rotational movement;

a driving device which rotates the image acquiring device;

a template storing device which stores a template used for recognizing the target, the template including a plurality of pixels each having a pixel-gradation value and a weighting value;

a template updating device which updates the template by updating the pixel-gradation value and the weighting value, when the image acquiring device directs a predetermined direction, or in accordance with a control signal input from outside;

a recognition processing device which compares a newly obtained image and an updated template for each pixels, and calculates the pixel-gradation value that can be obtained for each pixels, and an evaluation function value based on an optical flow; and

an operation instruction device which outputs an operation instruction to the driving device in accordance with the direction of the image acquiring device and the evaluation function value, wherein

the recognition processing device calculates;

a degree of approximation of the each pixels by comparing the pixel-gradation value of the each pixels of the images of the pixel-gradation value of the corresponding pixels of the template;

the optical flow of the pixels based on the continuous images,

a local shift quantity of the each pixels by multiplying the degree of approximation by the optical flow; and

the evaluation function value by summing up all values obtained by multiplying the local shift quantity and the weighting value for the each pixels.

2. The target recognizing device according to claim 1, wherein

the target recognizing device comprises a group of templates for images acquired while rotating the image acquiring device;

and

the group of template is updated according to the rotation direction of the image acquiring device.

3. The target recognizing device according to claim 2, wherein

the group of templates which are updated while rotating in the same direction comprises at least two types of templates;

and

one of which is added with a positive-polarity weighting value, while the other one of which is added with a negative-polarity weighting value.

4. A target recognition program stored in a computer-readable medium, for executing in a computer comprising:

a driving process of rotating an image acquiring device which acquires images including a target while performing rotational movement;

a template storing process of storing a template used for recognizing the target, the template including a plurality of pixels each having a pixel-gradation value and a weighting value;

a template updating process of updating the template by updating the pixel-gradation value and the weighting value, when the image acquiring device directs a predetermined direction, or in accordance with a control signal input from outside;

a recognition process of comparing a newly obtained image and an updated template for each pixels, and calculating the pixel-gradation value for the each pixels, and an evaluation function value based on an optical flow; and

an operation instruction process of outputting an operation instruction to the driving device in accordance with the direction of the image acquiring device and the evaluation function value, wherein

in the recognition process

a degree of approximation of the each pixels is calculated by comparing the pixel-gradation value of the each pixels of the images and the pixel-gradation value of the corresponding pixels of the template,

the optical flow of the pixels is calculated based on the continuous images,

a local shift quantity of the each pixels is calculated by multiplying the degree of approximation by the optical flow, and

the evaluation function value is calculated by summing up all values obtained by multiplying the local shift quantity and the weighting value of the each pixels.

5. The target recognition program stored in a computer-readable medium, according to claim 4, wherein

a plurality of template groups are provided for use in updating templates according to the rotation directions at the time of image acquiring processing.

6. The target recognition program stored in a computer-readable medium, according to claim 5, wherein

the plurality of templates to be updated at the time of rotating in the same direction include at least two templates, and

one of which is added with a positive-polarity weighting value while the other one of which is added with a negative-polarity weighting value.

7. A target recognizing method comprising:

a driving step of rotating an image acquiring device which acquires images including a target while performing rotational movement;

a template storing step of storing a template used for recognizing the target, the template including a plurality of pixels each having a pixel-gradation value and a weighting value;

a template updating step of updating the template by updating the pixel-gradation value and the weighting value, when the image acquiring device directs a predetermined direction, or in accordance with a control signal input from outside;

a recognition processing step of comparing a newly obtained image and an updated template for each pixels, and calculates the pixel-gradation value for the each pixels, and an evaluation function value based on an optical flow; and

an operation instruction step of outputting an operation instruction to the driving device in accordance with the direction of the image acquiring device and the evaluation function value, wherein

in the recognition processing step,

a degree of approximation of the each pixels is calculated by comparing the pixel-gradation value of the each pixels of the images and the pixel-gradation value of the corresponding pixels of the template,

the optical flow of the pixels is calculated based on the continuous images,

a local shift quantity of the each pixels is calculated by multiplying the degree of approximation by the optical flow, and

the evaluation function value is calculated by summing up all values obtained by multiplying the local shift quantity and the weighting value of the each pixels.