Object detection device, vehicle, and object detection process

1. An object detection device configured to be mounted on a vehicle, comprising:

a camera that is a stereo camera located on a rear side of the vehicle and configured to capture a video that includes an image which is an aerial view image of a road surface, on which the vehicle travels, from an upper side of the rear side of the vehicle;

a memory that stores a program;

a processor that executes the program stored in the memory, the processor configured to:

extract a point group that is a set of points, from among a plurality of particular points, representing a part of an object as the object in a world coordinate system in a real space, wherein an X-axis extends in a vehicle width direction of the vehicle in a horizontal direction and a Y-axis extends orthogonal to the X-axis in the horizontal direction;

from among a plurality of objects that have been extracted, determine a neighboring object in an XY-plane of the world coordinate system, the XY-plane being partitioned in such a manner that a first block whose dimension along the Y-axis is larger than that along the X-axis is located in a front area and second blocks whose dimensions along the Y-axis each are smaller than that along the X-axis are arranged in a direction in which the Y-axis extends and located in turning areas, the front area facing the vehicle in the direction in which the Y-axis extends, the turning areas being adjacent to the front area in a direction in which the X-axis extends, wherein the neighboring object is the object, from among the plurality of objects, which is the closest to the vehicle in each of the first block and the second blocks for each of the first block and the second blocks;

transform coordinates of the neighboring object in the world coordinate system into coordinates of the neighboring object in the image captured by the camera; and

perform a person detection process in which it is determined whether or not the neighboring object is a person based on the coordinates of the neighboring object in the image.

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

the first block includes a plurality of first blocks, and

the front area is partitioned into the plurality of the first blocks which are arranged in the direction in which the X-axis extends in the world coordinate system.

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

the processor is further configured to:

acquire a disparity image whose pixels are correlated with a disparity obtained by comparing a first image and a second image from the video captured by the stereo camera; and

derive coordinates of the points in the world coordinate system from the disparity image, and

transform the coordinates of the neighboring object in the world coordinate system into the coordinates of the neighboring object in the image captured by the stereo camera.

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

the number of the neighboring objects on which the person detection process is performed has an upper limit, and

the processor is configured to perform the person detection process on the neighboring objects in the first blocks, and then, on the neighboring objects in the second blocks in an order from the neighboring object of the second block that is closer to the vehicle.

5. A vehicle on which an object detection device according to claim 1 is mounted.

6. The vehicle of claim 5 which is a forklift.

7. An object detection process performed by an object detection device mounted on a vehicle, wherein the object detection device comprises a memory that stores a program and a processor that executes the program stored in the memory, and a camera that is a stereo camera located on a rear side of the vehicle and configured to capture a video that includes an image which is an aerial view image of a road surface, on which the vehicle travels, from an upper side of the rear side of the vehicle, the process comprising the steps of:

extracting a point group that is a set of points, from among a plurality of particular points, representing a part of an object as the object in a world coordinate system in a real space, wherein an X-axis extends in a vehicle width direction of the vehicle in a horizontal direction and a Y-axis extends orthogonal to the X-axis in the horizontal direction;

from among a plurality of objects that have been extracted, determining a neighboring object in an XY-plane of the world coordinate system, the XY-plane being partitioned in such a manner that a first block whose dimension along the Y-axis is larger than that along the X-axis is located in a front area and second blocks whose dimensions along the Y-axis each are smaller than that along the X-axis are arranged in a direction in which the Y-axis extends and located in turning areas, the front area facing the vehicle in the direction in which the Y-axis extends, the turning areas being adjacent to the front area in a direction in which the X-axis extends, wherein the neighboring object is the object, from among the plurality of objects, which is the closest to the vehicle in each of the first block and the second blocks for each of the first block and the second blocks;

transforming coordinates of the neighboring object in the world coordinate system into coordinates of the neighboring object in the image captured by the camera; and

performing a person detection process in which it is determined whether or not the neighboring object is a person based on the coordinates of the neighboring object in the image.

8. The object detection process according to claim 7 , wherein

the first block includes a plurality of first blocks, and

the front area is partitioned into the plurality of the first blocks which are arranged in the direction in which the X-axis extends in the world coordinate system.

9. The object detection process according to claim 7 ,

further comprising:

acquiring a disparity image whose pixels are correlated with a disparity obtained by comparing a first image and a second image from the video captured by the stereo camera; and

deriving coordinates of the points in the world coordinate system from the disparity image, and

transforming the coordinates of the neighboring object in the world coordinate system into the coordinates of the neighboring object in the image captured by the stereo camera.

10. The object detection process according to claim 7 , wherein

the number of the neighboring objects on which the person detection process is performed has an upper limit, and

performing the person detection process on the neighboring objects in the first blocks, and then, on the neighboring objects in the second blocks in an order from the neighboring object of the second block that is closer to the vehicle.