Object recognition device and object recognition method

• US 8,818,026 B2
• Filed: 04/24/2009
• Issued: 08/26/2014
• Est. Priority Date: 04/25/2008
• Status: Active Grant

• Current Assignee: Hitachi Astemo Ltd. (Hitachi, Ltd.)
• Original Assignee: Hitachi Automotive Systems Limited (Hitachi, Ltd.)
• Inventors: Kiyohara, Masahiro, Irie, Kota, Kourogi, Naoki, Monji, Tatsuhiko, Takahashi, Yoshitaka, Muramatsu, Shoji
• Primary Examiner(s): Bitar, Nancy
• Application Number:
  US12/918,042
• Publication Number:
  US 20120269382A1
• Time in Patent Office: 1,950 Days
• Field of Search: None
• US Class Current: 382/103
• CPC Class Codes: G01C 21/26 specially adapted for navigation in a road network
• CPC Class Codes: G06T 2207/30256 Lane; Road marking
• CPC Class Codes: G06T 5/80 Geometric correction
• CPC Class Codes: G06T 7/254 involving subtraction of images
• CPC Class Codes: G06V 20/588 Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
• CPC Class Codes: G08G 1/09623 Systems involving the acquisition of information from passive traffic signs by means mounted on the vehicle G08G1/0967 takes precedence
• CPC Class Codes: G08G 1/165 for passive traffic, e.g. including static obstacles, trees

Abstract

An object recognition device includes; an image-capturing unit mounted to a mobile body; an image generation unit that converts images captured by the image-capturing unit at different time points to corresponding synthesized images as seen vertically downwards from above; a detection unit that compares together a plurality of the synthesized images and detects corresponding regions; and a recognition unit that recognizes an object present upon the road surface from a difference between the corresponding regions.

Claims

1. An object recognition device, comprising:

an image-capturing unit mounted to a mobile body;

a condition setting unit that specifies a halation region upon images including a road surface captured by the image-capturing unit at different time points, and that executes masking processing for labeling a mask region upon the halation region;

an image generation unit that converts the images including the mask region labeled by the condition setting unit to corresponding synthesized images as seen vertically downwards from above;

a detection unit that compares a plurality of the synthesized images and detects corresponding road surface regions within the plurality of synthesized images; and

a recognition unit that recognizes a manhole present upon the road surface based upon the corresponding road surface regions of the plurality of synthesized images while ignoring the mask region;

wherein the recognition unit recognizes the manhole that has a circular or quadrilateral shape by specifying a region based upon a difference in luminance among the corresponding road surface regions of the plurality of synthesized images, and performing circular or quadrilateral shape detection in the specified region, and

wherein the recognition unit recognizes the manhole that has the circular shape by determining an external circumferential contour of the specified region, superimposing a standard circle upon the specified region, calculating an overlap ratio of the standard circle with the external circumferential contour, and deciding whether the specified region is circular based on the calculated overlap ratio.

2. An object recognition device according to claim 1, wherein the detection unit detects the corresponding road surface regions from a shift amount of the mobile body between the plurality of synthesized images.

3. An object recognition device according to claim 2, wherein the detection unit comprises a sensor that detects at least one of vehicle speed, acceleration, and angular velocity, and a shift amount of the mobile body is calculated by using at least one of the sensors.

4. An object recognition device according to claim 2, wherein the detection unit detects a characteristic feature included in the plurality of synthesized images, and calculates a shift amount of the mobile body from a shift amount of the characteristic feature.

5. An object recognition device according to claim 1, wherein the recognition unit generates an added image by adding together the corresponding road surface regions, and recognizes the manhole present upon the road surface from the added image.

6. An object recognition device according to claim 1, wherein the image-capturing unit captures the images including the road surface at the different time points, when a position information of the manhole existing within a predetermined range from the mobile body is supplied from a car navigation device.

7. An object recognition device according to claim 1, further comprising a position information generation unit that detects information for the recognized manhole from manhole information specifying at least positions and shapes of manholes and generates position information of the manhole based on the detected information.

8. An object recognition device according to claim 1, wherein the recognition unit generates a plurality of differential images for the corresponding road surface regions of the plurality of synthesized images, generates a consolidated image from the plurality of differential images, and recognizes the manhole present upon the road surface from the consolidated image.

9. An object recognition device according to claim 8, wherein the recognition unit recognizes the manhole that has a circular or quadrilaterial shape by performing circular or quadrilateral shape detection upon the consolidated image.

10. An object recognition device according to claim 9, wherein the recognition unit recognizes the manhole that has the circular or quadrilateral shape by extracting a region from the consolidated image based upon pixel values of the extracted region, determining an external circumferential contour of the extracted region, superimposing a standard circle upon the extracted region, calculating an overlap ratio of a standard circle with the external circumferential contour, and deciding whether the extracted region is circular based on the calculated overlap ratio.