MOBILE OBJECT RECOGNIZING DEVICE, MOBILE OBJECT RECOGNIZING METHOD, AND COMPUTER PROGRAM THEREOF

• US 20090252377A1
• Filed: 10/04/2007
• Published: 10/08/2009
• Est. Priority Date: 10/06/2006
• Status: Active Grant

• Current Assignee: Aisin Seiki Co., Ltd. (Aisin Corp.)
• Original Assignee: Aisin Seiki Co., Ltd. (Aisin Corp.)
• Inventors: Akita, Tokihiko, Hayashi, Keita
• Granted Patent: US 8,290,209 B2
• Time in Patent Office: Days
• US Class Current: 382/106
• CPC Class Codes: G06T 2207/30241 Trajectory
• CPC Class Codes: G06T 7/269 using gradient-based methods
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
• CPC Class Codes: G06V 20/58 Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
• CPC Class Codes: G06V 2201/08 Detecting or categorising vehicles

Abstract

A mobile object recognizing device comprises a camera (2) for taking time-series images, a feature point extracting unit (23) for extracting the feature points of the individual time-series images taken by the camera (2), an optical flow creating unit (24) for comparing the feature points of the time-series images between different images, to create an optical flow joining the feature points having the same pattern, and a grouping operation unit (25) for selecting that optical flow as one belonging to one mobile object, the prolonged line of which intersects with one vanishing point within a predetermined error range and in which the external ratio of a segment joining the other end point and the vanishing point with one end point of the optical flow being the externally dividing point in the extension of the optical flow. The mobile object recognizing device may further comprise an image correcting unit (22) for correcting the image taken by the camera (2), into a transparent diagram in accordance with the characteristics of the lens of the camera (2).

Claims

1. A mobile object recognizing device comprising;

an image-taking means taking images in time series;

a feature point extracting means extracting a feature point of each image taken in time series by the image-taking means;

an optical flow creating means creating an optical flow corresponding to a vector formed by connecting one feature point in one image with another feature point in another image, the connected two feature points being identified to have the same pattern by comparing the feature points with one another between different images; and

a grouping means selecting and grouping the optical flows created by the optical flow creating means which belong to a particular mobile object, wherein each of the selected optical flows is defined as having an extended line intersecting at one focus of expansion within a predetermined error range, and having an equal external ratio of each line segment connecting one of the end points of the optical flow and the focus of expansion within a predetermined error range when the other one of the end points of the optical flow is an externally dividing point on the extended line.

2. The mobile object recognizing device according to claim 1, further comprising;

an image correcting means correcting the image taken by the image-taking means in accordance with characteristics of a lens of the image-taking means to obtain a perspective image, wherein

the feature point extracting means extracts the feature point of the image corrected by the image correcting means.

3. The mobile object recognizing device according to claim 1, further comprising;

a distance calculating means calculating a distance from the image-taking means to the mobile object by using the optical flows grouped by the grouping means.

4. A mobile object recognizing device comprising;

an image-taking means taking images in time series;

a feature point extracting means extracting a feature point of each image taken in time series by the image-taking means;

an optical flow creating means creating an optical flow corresponding to a vector formed by connecting one feature point in one image with another feature point in another image, the connected two feature points being identified to have the same pattern by comparing the feature points with one another between different images;

a distance detecting means detecting a distance from the image-taking means to at least one feature point extracted by the feature point extracting means;

a displacement calculating means calculating a displacement of the feature point in each optical flow based on the distance to the feature point detected by the distance detecting means and the optical flow including the feature point; and

an equal displacement point selecting means selecting the feature points having an equal displacement calculated by the displacement calculating means as the feature points belonging to one mobile object.

5. The mobile object recognizing device according to claim 4 further comprising an image correcting means correcting the image taken by the image-taking means in accordance with characteristics of a lens of the image-taking means to obtain a perspective image, wherein

the feature point extracting means extracts the feature point of the image corrected by the image correcting means.

6. The mobile object recognizing device according to claim 4 further comprising;

a grouping means selecting and grouping the optical flows created by the optical flow creating means which belong to a particular mobile object, wherein each of the selected optical flows is defined as having an extended line intersecting at one focus of expansion within a predetermined error range, and having an equal external ratio of each line segment connecting one of the end points of the optical flow and the focus of expansion within a predetermined error range when the other one of the end points of the optical flow is an externally dividing point on the extended line; and

a distance calculating means calculating a distance from the image-taking means to the mobile object by using the optical flows grouped by the grouping means, wherein the distance calculated by the distance calculating means is the distance to the feature point.

7. The mobile object recognizing device according to claim 4, wherein the distance detecting means measures the distance to the mobile object by a physical means.

8. A mobile object recognizing method comprising;

an image-taking step inputting images in time series;

a feature point extracting step extracting feature point of each image inputted in time series in the image-taking step;

an optical flow creating step creating an optical flow corresponding to a vector formed by connecting one feature point in one image with another feature point in another image, the connected two feature points being identified to have the same pattern by comparing the feature points with one another between different images; and

a grouping step selecting and grouping the optical flows, wherein each of the selected optical flows is defined as having an extended line intersecting at one focus of expansion within a predetermined error range, and having an equal external ratio of each line segment connecting one of the end points of the optical flow and the focus of expansion within a predetermined error range when the other one of the end points of the optical flow is an externally dividing point on the extended line.

9. The mobile object recognizing method according to claim 8 further comprising;

an image correcting step correcting the image inputted in the image-taking step in accordance with characteristics of a lens of the image-taking means to obtain a perspective image, wherein

the feature point extracting step extracts the feature point of the image corrected in the image correcting step.

10. A computer program causing a computer to function as;

an image-inputting means inputting images in time series from an image-taking means;

a feature point extracting means extracting a plurality of feature points of each image taken in time series by the image-inputting means;

an optical flow creating means creating an optical flow corresponding to a vector formed by connecting one feature point in one image with another feature point in another image, the connected two feature points being identified to change positions thereof on the image and have the same pattern by comparing the feature points with one another between different images; and

a grouping means selecting and grouping the optical flows created by the optical flow creating means which belong to a particular mobile object, wherein each of the selected optical flows is defined as having an extended line intersecting at one focus of expansion within a predetermined error range, and having an equal external ratio of each line segment connecting one of the end points of the optical flow and the focus of expansion within a predetermined error range when the other one of the end points of the optical flow is an externally dividing point on the extended line.