Object detection apparatus, object detection method and object detection program

• US 20080278576A1
• Filed: 05/10/2007
• Published: 11/13/2008
• Est. Priority Date: 05/10/2007
• Status: Active Grant

• Current Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Aloimonos, Yiannis, Nishigaki, Morimichi, Domke, Justin
• Granted Patent: US 8,300,887 B2
• Time in Patent Office: Days
• US Class Current: 348/94
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/20164 Salient point detection; Corner detection
• CPC Class Codes: G06T 2207/30261 Obstacle
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments

Abstract

An object detection apparatus and method for accurately detecting a movable object in a region around a vehicle from a time series of images obtained through a camera mounted on the vehicle by eliminating the influence of the movement of the camera through simple processing, and a program for making a computer execute processing in the apparatus. The object detection apparatus has a feature point extraction unit which extracts a feature point contained in a feature region of each image in the time series of images obtained through a camera mounted on the vehicle, a correspondence degree computation unit which computes the degree of correspondence for each pair of the feature points, wherein one of the feature points in the each pair is each of one or more of the feature points extracted by the feature point extraction unit from one of two images taken by the camera at different times, and another of the feature points in the each pair is each of a plurality of the feature points extracted by the feature point extraction unit from another of the two images, and a detection unit which detects the movable object on the basis of the degree of correspondence computed by the correspondence degree computation unit.

Claims

1. An object detection apparatus which detects a movable object in a region around a vehicle from a time series of images obtained through a camera mounted on the vehicle, the apparatus comprising:

a feature point extraction unit which extracts a feature point contained in a feature region of each of the images in the time series;

a correspondence degree computation unit which computes a degree of correspondence for each pair of the feature points, wherein one of the feature points in the each pair is each of one or more of the feature points extracted by the feature point extraction unit from one of two images taken by the camera at different times, and another of the feature points in the each pair is each of a plurality of the feature points extracted by the feature point extraction unit from the other of the two images; and

a detection unit which detects the movable object on the basis of the degree of correspondence computed by the correspondence degree computation unit.

2. The object detection apparatus according to claim 1, further comprising a movement estimation unit which estimates parameters indicating the state of movement of the camera on the basis of the degree of correspondence computed by the correspondence degree computation unit, wherein the detection unit detects the movable object on the basis of the parameters estimated by the movement estimation unit for indicating the state of movement, and a feature value of each of the pair having the highest degree of correspondence among each of the pair composed of the feature point extracted from one of the two images and each of a plurality of the feature points extracted from the other of the two images.

3. The object detection apparatus according to claim 2, wherein the movement estimation unit estimates the parameters indicating the state of movement of the camera on the basis of the degree of correspondence for each of the pair of the feature points satisfying an epipolar constraint condition.

4. The object detection apparatus according to claim 2, wherein the feature point extraction unit extracts the feature points by using a plurality of Gabor filters having different orientations and scales, and the correspondence degree computation unit computes the degree of correspondence by using a feature value obtained by performing filtering with the plurality of Gabor filters on the feature points extracted by the feature point extraction unit.

5. The object detection apparatus according to claim 2, wherein the feature point extraction unit extracts the feature points by using changes in colors on the images, and the correspondence degree computation unit computes the degree of correspondence by using a feature value relating to the colors of the feature points extracted by the feature point extraction unit.

6. An object detection method for detecting a movable object in a region around a vehicle from a time series of images obtained through a camera mounted on the vehicle, the method comprising:

a feature point extraction step of extracting a feature point contained in a feature region of each of the images in the time series;

a correspondence degree computation step of selecting each pair of the feature points, wherein one of the feature points in the each pair is each of one or more of the feature points extracted by the feature point extraction step from one of two images taken by the camera at different times, and another of the feature points in the each pair is each of a plurality of the feature points extracted by the feature point extraction step from the other of the two images, and computing a degree of correspondence for the selected each pair; and

a detection step of detecting the movable object on the basis of the degree of correspondence computed in the correspondence degree computation step.

7. An object detection program for making a computer execute processing for detecting a movable object in a region around a vehicle from a time series of images obtained through a camera mounted on the vehicle, the program comprising the functions to make the computer execute:

feature point extraction processing for extracting a feature point contained in a feature region of each of the images in the time series;

correspondence degree computation processing for selecting each pair of the feature points, wherein one of the feature points in the each pair is each of one or more of the feature points extracted by the feature point extraction processing from one of two images taken by the camera at different times, and another of the feature points in the each pair is each of a plurality of the feature points extracted by the feature point extraction processing from the other of the two images, and for computing a degree of correspondence for the selected each pair; and

a detection processing for detecting the movable object on the basis of the degree of correspondence computed by the correspondence degree computation processing.