Object recognition system

• US 6,775,395 B2
• Filed: 02/15/2001
• Issued: 08/10/2004
• Est. Priority Date: 03/27/2000
• Status: Expired due to Term

• Current Assignee: Honda Giken Kogyo Kabushiki Kaisha (Honda Motor Company)
• Original Assignee: Honda Giken Kogyo Kabushiki Kaisha (Honda Motor Company)
• Inventors: Saka, Masakazu, Yuhara, Hiromitsu, Aoki, Tomoyoshi, Nishigaki, Morimichi
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): Akhavannik, Hussein
• Application Number:
  US09/783,613
• Publication Number:
  US 20010028729A1
• Time in Patent Office: 1,272 Days
• Field of Search: 382/104, 382/106, 382/154, 340/901, 340/903, 340/937
• US Class Current: 382/104
• CPC Class Codes: G06T 7/97 Determining parameters from multiple pictures depth or shape recovery from multiple images G06T7/55; stereo camera calibration G06T7/85
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
• CPC Class Codes: G06V 10/267 by performing operations on regions, e.g. growing, shrinking or watersheds
• 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 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: G06V 20/64 Three-dimensional objects

Abstract

An object recognition system having at least two image sensors and a controller that is adapted for measuring the distance from the system to a physical object with respect to respective windows of a n image captured by the sensors. The controller is further programmed to form clusters by uniting adjacent windows that have similar measure distances. The controller is programmed to judge whether each of the clusters is valid or invalid based on the attributes of the cluster and to recognize an object based on the clusters judged to be valid.

Claims

1. An object recognition system having at least two image sensors and a controller that is adapted for measuring distance from the system to a physical object with respect to respective windows of an image captured by the sensors,

wherein said controller is programmed to form clusters by uniting adjacent windows that have similar measured distances, judge whether each of the clusters is valid or invalid based on the attributes of the cluster,

recognize the physical object based on the clusters judged to be valid;

detect a failed state of the captured image based on the number of clusters judged to be invalid or based on a ratio of the number of clusters judged to be invalid to the total number of clusters contained in the captured image; and

disable performing the recognition of the physical object or performing control based on the recognized physical object if the failed state is detected.

2. The system of claim 1, wherein the attributes of the cluster include an area of the cluster, and

wherein said controller is programmed to calculate an area of the cluster based on the number of windows contained in the cluster and measured distance of each of the windows, and to judge that the cluster is valid if the area is larger than a predetermined threshold value.

3. The system of claim 2, wherein the area A of the cluster is calculated according to the following equation:

$A=\frac{{\mathrm{\mu <br><br>}}_h<br><br>{\mathrm{\mu <br><br>}}_v}{f^2}<br><br>D_{\mathrm{ave}}^2<br><br>N$

where μ

_h is the horizontal length of a window, μ

_v is the vertical length of the window, D_ave is the average of the measured distances of windows forming the cluster, N is the number of the windows forming the cluster, and f is the focal length.

4. The system of claim 1, wherein the attributes of the cluster include the number of windows contained in the cluster, and

wherein said controller is programmed to judge that the cluster is valid if the number of windows contained in the cluster is larger than a threshold value that is predetermined according to the measured distances of the windows contained in the cluster.

5. The system of claim 4, wherein the threshold value N_o is calculated according to the following equation:

$N_0=\frac{f^2}{{\mathrm{\mu <br><br>}}_h<br><br>{\mathrm{\mu <br><br>}}_v<br><br>D^2}<br><br>A_0$

where A₀ is a predetermined threshold value for the area of the cluster, μ

_h is the horizontal length of a window, μ

_v is the vertical length of the window, D is the measured distance of the cluster, and f is the focal length of a lens installed in the image sensor.

6. A method for recognizing a physical object in front of a vehicle, comprising:

capturing an image in front of the vehicle;

measuring distance from the vehicle to the physical object with respect to respective windows of the captured image;

uniting adjacent windows that have similar measured distances to form clusters;

judging whether each of the clusters is valid or invalid based on attributes of the cluster;

recognizing the physical object based on the clusters judged to be valid;

detecting a failed state of the captured image based on the number of clusters judged to be invalid or based on a ratio of the number of clusters judged to be invalid to the total number of clusters contained in the captured image; and

disabling performing the recognition of the physical object or performing control based on the recognized physical object if the failed state is detected.

7. The method of claim 6, wherein the attributes of the cluster include the area of the cluster, and

wherein the step of judging includes calculating an area of the cluster based on the number of windows contained in the cluster and measured distance of each of the windows, and judging that the cluster is valid if the area is larger than a predetermined threshold value.

8. The method of claim 6, wherein the attributes of the cluster include the number of windows contained in the cluster, and

wherein said step of judging includes judging that the cluster is valid if the number of windows contained in the cluster is larger than a threshold value that is predetermined according to the measured distances of the windows contained in the cluster.