Object Detection Device

• US 20180137339A1
• Filed: 06/15/2016
• Published: 05/17/2018
• Est. Priority Date: 06/30/2015
• Status: Active Grant

• Current Assignee: Hitachi Astemo Ltd. (Hitachi, Ltd.)
• Original Assignee: Hitachi Automotive Systems Limited (Hitachi, Ltd.)
• Inventors: OSATO, Takuma, TAKEMURA, Masayuki, OTSUKA, Yuji, NAGASAKI, Takeshi
• Granted Patent: US 10,360,436 B2
• Time in Patent Office: Days
• CPC Class Codes: B60R 21/00 Arrangements or fittings on vehicles for protecting or preventing injuries to occupants or pedestrians in case of accidents or other traffic risks safety belts or body harnesses in vehicles B60R22/00; devices, apparatus or methods for life-saving in general A62B; safety devices for propulsion unit control specially adapted for, or arranged in, vehicles B60K28/00; seats constructed to protect the occupant from the effect of abnormal g-forces, e.g. crash or safety seats, B60N2/42; energy-absorbing arrangements for hand wheels for steering vehicles B62D1/11; energy-absorbing arrangements for vehicle steering columns B62D1/19; harnessing in aircraft B64D25/00
• CPC Class Codes: G01C 3/06 Use of electric means to obtain final indication
• CPC Class Codes: G06T 1/00 General purpose image data processing
• CPC Class Codes: G06T 7/00 Image analysis
• CPC Class Codes: G06V 20/56 exterior to a vehicle by using sensors mounted on the vehicle
• 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 20/64 Three-dimensional objects
• CPC Class Codes: G08G 1/04 using optical or ultrasonic detectors
• CPC Class Codes: G08G 1/16 Anti-collision systems road vehicle drive control systems for predicting or avoiding probable or impending collision otherwise than by control of a particular sub-unit B60W30/08

Abstract

Provided is an object detection device capable of reliably recognizing a vehicle. This object detection device detects multiple three-dimensional objects from a left image and a right image captured with a left imaging unit and a right imaging unit (S103), extracts, as combination candidates from among the multiple three-dimensional objects, two three-dimensional objects which exist with an interval to the left/right, and determines whether a sparse parallax region, which is a region having a smaller parallax density than left/right regions, exists in an intermediate region between the two three-dimensional objects extracted as combination candidates. Then, the regions of two three-dimensional objects (for which it has been determined that a sparse parallax region exists in the intermediate region, and which have tentatively been identified as a single three-dimensional object) are extracted respectively from the left and right images and compared to each other, a determination is made regarding whether the perspective is the same, and when it is determined that the perspective is the same, the two three-dimensional objects are determined to be a single three-dimensional object.

Claims

1. An object detection device comprising:

a three-dimensional object detection unit that detects a plurality of three-dimensional objects from a left image and a right image imaged by a left imaging unit and a right imaging unit;

a combination candidate extraction unit that extracts two three-dimensional objects existing at an interval to the left and right from among the plurality of three-dimensional objects as a combination candidate;

a sparse region determination unit that determines whether or not there is a sparse parallax region in an intermediate region between the two three-dimensional objects extracted as the combination candidate, the sparse parallax region having a parallax density smaller than that of left and right regions;

a matching unit that extracts regions, obtained at the time of assuming the two three-dimensional objects for which it is determined that the sparse parallax region exists in the intermediate region as one three-dimensional object, from the left and right images, respectively, and compares the regions with each other to determine whether perspectives thereof are equal; and

a three-dimensional object combining unit that determines the two three-dimensional objects as the one three-dimensional object when the matching unit determines that the perspectives are equal.

2. The object detection device according to claim 1, wherein

the sparse region determination unit determines that the sparse parallax region exists when the parallax density is equal to or lower than a first threshold.

3. The object detection device according to claim 1, wherein

the matching unit determines that the perspective is equal when a correlation value obtained by adding absolute values of differences of the regions extracted from the left and right images is equal to or smaller than a second threshold.

4. The object detection device according to claim 1, further comprising:

a size determination unit that determines whether or not a size obtained when detection regions of the two three-dimensional objects are set to both ends of a predetermined detection target corresponds to a size of the detection target determined based on an imaging distance; and

a distance determination unit that compares imaging distances of the two three-dimensional objects extracted as the combination candidates and determines whether or not the imaging distances are equal to each other, wherein

the sparse region determination unit performs determination on whether or not the sparse parallax region exists when it is determined to correspond to the size of the detection target by the size determination unit and it is determined that the respective imaging distances are equal by the distance determination unit.

5. The object detection device according to claim 4, wherein

the distance determination unit determines that the respective imaging distances are equal when a difference between the respective imaging distances of the two three-dimensional objects is equal to or smaller than a third threshold.

6. The object detection device according to claim 4, further comprising:

a vehicle recognition unit that recognizes whether or not the combined three-dimensional object determined as the one three-dimensional object by the three-dimensional object combining unit is a vehicle; and

a combination threshold adjustment unit that adjusts a determination threshold of the three-dimensional object combining unit according to a size or vehicle type information of the vehicle recognized by the vehicle recognition unit.

7. The object detection device according to claim 1, further comprising

a combination determination target region detection unit that extracts a sparse region of which parallax density on an image is equal to or lower than a fourth threshold set in advance, as a combination determination target region, wherein

the three-dimensional object combining unit performs the determination only for the combination determination target region as a processing target.

8. The object detection device according to claim 1, wherein

the three-dimensional object combining unit sets only a three-dimensional object on a subject vehicle traveling path, estimated based on CAN information and road surface information, as a combination target.

9. The object detection device according to claim 1, further comprising

a vehicle control unit, wherein

information to distinguish a three-dimensional object detection result obtained by the three-dimensional object detection unit from a detection result obtained by the three-dimensional object combining unit is added to an external communication signal.

10. An object detection device comprising:

a three-dimensional object detection unit that detects a three-dimensional object from left and right images imaged by left and right imaging units;

a division candidate point search unit that determines whether or not a region having a sparse parallax density than left and right regions exists in an intermediate region of the three-dimensional object and sets end portions of a dense parallax region as division candidate points when it is determined that the region exists;

a matching unit that extracts a region from which the three-dimensional object is detected from the left and right images, and compares the regions with each other to determine whether or not perspectives are different; and

a three-dimensional object division unit that divides the three-dimensional object at the division candidate points when it is determined that the perspectives are different by the matching unit.