Approaching-Object Detection System and Vehicle

• US 20160098606A1
• Filed: 05/29/2014
• Published: 04/07/2016
• Est. Priority Date: 07/03/2013
• Status: Active Grant

• Current Assignee: Clarion Co., Ltd. (Faurecia SA)
• Original Assignee: Clarion Co., Ltd. (Faurecia SA)
• Inventors: NAKAMURA, Katsuyuki, AKIYAMA, Yasuhiro, IRIE, Kota
• Granted Patent: US 9,811,745 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: B60R 2300/30 characterised by the type of image processing
• CPC Class Codes: G06T 2207/30261 Obstacle
• CPC Class Codes: G06V 10/247 by affine transforms, e.g. correction due to perspective effects; Quadrilaterals, e.g. trapezoids
• 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 20/584 of vehicle lights or traffic lights

Abstract

An approaching-object detection system detects an approaching object based on images captured by an imaging device. The approaching-object detection system comprises an extraction unit, a distortion correction unit, an object detection unit, and an approaching-object classification unit. The extraction unit extracts a partial image of one far side and a partial image of the other far side from each of the images. The distortion correction unit corrects distortion of the partial images. The object detection unit detects an object from the corrected partial images through pattern matching with reference to preinstalled image information on the object. The approaching-object classification unit classifies the detection result of the object detection unit as an approaching object or not.

Claims

1. An approaching-object detection system for detecting an approaching object based on images captured by an imaging device, the approaching-object detection system comprising:

an extraction unit configured to extract a partial image of one far side and a partial image of the other far side from each of the images;

a distortion correction unit configured to correct distortion of the partial images extracted by the extraction unit;

an object detection unit configured to detect an object from the partial images corrected in distortion by the distortion correction unit through pattern matching with reference to preinstalled image information on the object; and

an approaching-object classification unit configured to classify the detection result of the object detection unit as an approaching object or not.

2. The approaching-object detection system according to claim 1, wherein the approaching-object classification unit is configured to:

calculate a feature value of the object detected by the object detection unit; and

classify the object as approaching object or not based on a variation with time in the calculated feature value of the object.

3. The approaching-object detection system according to claim 2, further comprising an orientation estimation unit configured to estimate an orientation of the imaging device,

wherein the approaching-object detection unit is configured to:

locate a grounding point of the object detected by the object detection unit in the partial images corrected in distortion by the distortion correction unit;

calculate a relative distance from the imaging device to the object based on the orientation of the imaging device estimated by the orientation estimation unit and the located grounding point;

estimate a pixel count of the object based on the calculated relative distance and a kind of the object detected by the object detection unit; and

determine whether the object is an approaching object or not in consideration of a difference between the estimated pixel count and an actual pixel count of the object.

4. The approaching-object detection system according to claim 2, wherein the approaching-object classification unit is configured to calculate, as the feature value of the object, at least one of a width of the object in the partial images, a height of the object in the partial images, coordinates of the grounding point of the object, a motion vector of the object, an edge strength, a texture frequency, a rectangle expansion rate, and an actual pixel count of the object in the partial images.

5. The approaching-object detection system according to claim 1, further comprising:

an object tracking unit configured to track an object by comparing the partial images corrected in distortion by the distortion correction unit with an object previously classified as an approaching object; and

a combined-determination unit configured to determine whether there is consistency between a detection result of the object detection unit and a tracking result of the object tracking unit based on the detection result and the tracking result,

wherein the combined-determination unit is configured to;

determine that there is no consistency between the detection result and the tracking result in a case where the object detection unit newly detects an object and the object newly detected by the object detection unit is an object previously classified as an approaching object and is being tracked by the object tracking unit;

estimate a current position of the object previously classified as an approaching object using a position as of the previous classification in a case where the combined-determination unit determines that there is no consistency between the detection result and the tracking result; and

output a result indicating the object positioned closer to the estimated current position between the detection result of the object detection unit and the tracking result of the object tracking unit to the approaching-object classification unit.

6. The approaching-object detection system according to claim 1,

wherein the image information on the object to be referred to by the object detection unit includes image information for daytime and image information for nighttime,

wherein the object detection unit is configured to:

determine whether a current time is daytime or nighttime based on gain information of the imaging device;

detect an object with reference to the image information for daytime in a case where a determination result is daytime; and

detect an object with reference to the image information for nighttime in a case where a determination result is nighttime.

7. The approaching-object detection system according to claim 1,

wherein the object detection unit includes:

input layers to be fed with a plurality of feature values of the partial images corrected in distortion by the distortion correction unit;

hidden layers configured to non-linearly transform linearly-coupled outputs from the input layers and weighting coefficients; and

output layers configured to calculate classification probabilities for objects based on transformation results of the hidden layers, and

wherein a number of outputs of the output layers is equal to a number of kinds of objects to be detected by the object detection unit.

8. A vehicle on which the approaching-object detection system according to claim 1 is mounted, wherein the imaging device is attached on at least one of front, rear, and sides of the vehicle.

9. The vehicle according to claim 8, further comprising a warning notification unit for issuing a warning in a case where the approaching-object classification unit classifies the detection result as an approaching object.

10. The vehicle according to claim 8,

wherein a plurality of imaging devices are attached on the vehicle, and

wherein the vehicle comprises an imaging device selection unit for selecting an imaging device to capture the images to be used in approaching-object classification in the approaching-object detection system based on speed information and steering angle information on the vehicle.