JP2020003850A - Fence detector - Google Patents

Abstract

To provide a fence detector that can detect a fence that is difficult to be detected by a sonar.SOLUTION: A fence detector 10 comprises: an image acquisition unit 101 for acquiring images obtained by an imaging unit 50 that is mounted on a vehicle 1 by imaging the surroundings of the vehicle 1; an obstacle detection unit 102 for detecting an obstacle existing in the surroundings of the vehicle 1 on the basis of a plurality of images acquired by the image acquisition unit 101; an obstacle determination unit 103 for determining whether or not the obstacle detected by the obstacle detection unit 102 is a stick object; a feature-points extraction unit 104, when the obstacle is determined as a plurality of stick objects by the obstacle determination unit 103, for extracting the feature points in an image area between the determined plurality of stick objects; and a fence determination unit 105 for determining that there is a fence between the stick objects when the plurality of feature points extracted by the feature-points extraction unit 104 meet a predetermined condition.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a fence detection device.

  2. Description of the Related Art Conventionally, when performing automatic parking or the like, there is a technique for detecting an obstacle existing around a vehicle using a sonar. Specifically, a technique of irradiating an ultrasonic wave around a vehicle using a sonar and detecting a reflected wave thereof to detect a distance to an object is known (for example, see Patent Document 1).

JP 2005-306301 A

  However, in the case of a fence that is formed in a net shape and the background can be seen through, it may not be possible to detect the distance to the fence using sonar.

  The present invention has been made in view of the above problems, and has as its object to provide a fence detection device that can detect a fence that is difficult to detect with sonar.

  A fence detection device according to the present invention is based on an image acquisition unit that acquires an image obtained by imaging the periphery of the vehicle with an imaging unit mounted on the vehicle, and a plurality of images acquired by the image acquisition unit. An obstacle detection unit that detects an obstacle present around the vehicle, an obstacle determination unit that determines whether the obstacle detected by the obstacle detection unit is a rod-shaped object, and an obstacle determination unit. A feature point extracting unit that extracts a feature point in an image area between the determined plurality of bar-shaped objects when the plurality of obstacles is determined to be the bar-shaped object; When the plurality of feature points extracted by the extraction unit satisfy predetermined conditions, a fence determination unit that determines that there is a fence between the rod-shaped objects is provided.

  According to the present invention, it is possible to detect a fence that is difficult to detect with a sonar.

It is a mimetic diagram showing the composition of the parking lot to which fence detecting device 100 concerning an example is applied (the 1). It is a mimetic diagram showing the composition of the parking lot to which fence detecting device 100 concerning an example is applied (the 2). It is a hardware block diagram showing the hardware elements which constitute the fence detection system to which the fence detection device is applied. It is a flowchart which shows the flow of a series of processes performed by a fence detection apparatus. It is an explanatory view for explaining the fence detection processing (part 1). It is explanatory drawing for demonstrating the detection process of a fence (the 2). FIG. 9 is an explanatory diagram for describing predetermined conditions in a fence determination process.

  Hereinafter, specific embodiments of a fence detection device according to the present invention will be described with reference to the drawings.

  The fence detection device 100 according to the first embodiment is applied to a running vehicle 1 in a parking lot having a fence at a boundary of a parking space.

(Explanation of application scene of fence detection device)
First, an outline of a parking lot to which the fence detection device 100 (FIG. 3) according to the first embodiment is applied will be described with reference to FIGS. 1 and 2.

  In FIG. 1, a schematic diagram of the parking lot is shown on the left, but an example of an image captured by the front camera 50 is shown on the right.

  In FIG. 1, the left half of the image captured by the front camera 50 is unnecessary, and therefore will be omitted, and the description will be made using only the right half.

  The image shown in the upper right of FIG. 1 is an image captured by the front camera 50 at time t0. The image shown in the lower right of FIG. 1 is an image captured by the front camera 50 at time t1.

  FIG. 1 and FIG. 2 are schematic diagrams illustrating a configuration of a parking lot to which the fence detection device 100 according to the first embodiment is applied. At a boundary that separates a parking lot from another parking lot, pillars P (bar-shaped members) are erected at predetermined intervals. Between the columns P adjacent to each other, there is provided a net-like fence F formed by stretching, for example, a wire-like metal or a synthetic resin.

  The vehicle 1 is traveling upward in the paper surface of FIG. 1 (the direction indicated by the white arrow in FIG. 1). A front camera 50 (imaging unit) that captures an image of the periphery of the vehicle 1 including the front end of the vehicle 1 is mounted on a front portion of the vehicle 1. The star mark shown in FIGS. 1 and 2 indicates the position of the column P where the front camera 50 is focused.

  The fence detection device 100 performs a fence detection process described later based on the image captured by the front camera 50. FIG. 1 shows a state in which no obstacle other than the strut P and the fence F is present in the parking lot. However, even if another vehicle 2 (obstacle) exists in the parking lot as shown in FIG. The detection device 100 is applied.

  As shown in FIG. 2, even when another vehicle 2 is present in the parking lot, the fence detection device 100 detects only the support P and the fence F from the image captured by the front camera 50.

(Explanation of system configuration to which fence detection device is applied)
FIG. 3 is a hardware block diagram illustrating hardware components that configure the fence detection system 10 to which the fence detection device 100 is applied.

  The fence detection system 10 is installed in the vehicle 1. The fence detection system 10 includes a front camera 50, a wheel speed sensor 60, a steering angle sensor 70, and a fence detection device 100.

  Wheel speed sensor 60 detects the wheel speed of vehicle 1. The information detected by the wheel speed sensor 60 is output to the fence detection device 100 and is used in arithmetic processing in the fence detection device 100.

  The steering angle sensor 70 measures the steering angle of the steering of the vehicle 1. Information detected by the steering angle sensor 70 is output to the fence detection device 100 and is used in arithmetic processing in the fence detection device 100.

(Description of the configuration of the fence detection device)
As illustrated in FIG. 3, the fence detection device 100 includes an image acquisition unit 101, an obstacle detection unit 102, an obstacle determination unit 103, a feature point extraction unit 104, a fence determination unit 105, and a movement detection unit 106 And a storage unit 107.

  Inside the fence detecting apparatus 100, for example, a microprocessor and a program including peripheral devices, a random access memory (RAM) for storing necessary data, a read only memory (ROM), and a dedicated ASIC for performing image processing and signal processing (Application Specific Integrated Circuit) and other modules are mounted.

  The image obtaining unit 101 obtains an image obtained by imaging the periphery of the vehicle 1 (FIG. 1) with the front camera 50.

  The obstacle detection unit 102 detects an obstacle existing around the vehicle 1 based on the plurality of images acquired by the image acquisition unit 101.

  The obstacle determination unit 103 determines whether the obstacle detected by the obstacle detection unit 102 is the pillar P.

  When a plurality of obstacles are determined to be the pillars P by the obstacle determination unit 103, the feature point extracting unit 104 extracts the feature points FP in the image region between the determined plurality of pillars P. .

  The fence determination unit 105 determines that the fence F exists between the columns P when the plurality of feature points FP extracted by the feature point extraction unit 104 satisfy predetermined conditions.

  The predetermined condition is that a plurality of feature points FP are arranged substantially linearly in the vertical direction and the horizontal direction, the intervals between the feature points FP are substantially the same, and the number of feature points FP is a predetermined number or more within a predetermined area. It is a condition that exists.

Here, the predetermined value is a value set in advance, and is preferably set to, for example, 5 cm (FIG. 7). The predetermined area is an area set in advance, and is preferably set to, for example, 750 cm ² (FIG. 7). The predetermined number is a preset number, and is preferably set to 8, for example.

  The movement detection unit 106 detects that the vehicle 1 is moving, based on information on the behavior of the vehicle 1 (FIG. 1). The information on the behavior of the vehicle 1 includes the wheel speed detected by the wheel speed sensor 60 and the steering angle detected by the steering angle sensor 70.

(Flow of a series of processes performed by the fence detection device 100)
Next, a flow of a series of processes performed by the fence detection device 100 will be described with reference to a flowchart illustrated in FIG. 4 and explanatory diagrams illustrated in FIGS. 5 and 6.

  A series of processes illustrated in FIG. 4 is started, for example, when an input for instructing to start execution of the fence detection process is transmitted to the fence detection device 100 by a driver's switch operation.

(Step S1)
First, the image obtaining unit 101 obtains an image obtained by imaging the periphery of the vehicle 1 with the front camera 50.

(Step S2)
Next, the obstacle detection unit 102 detects an obstacle existing around the vehicle 1 using a known motion stereo technique. By using the motion stereo technology, the obstacle detection unit 102 can obtain a three-dimensional point cloud of the obstacle.

  Specifically, as shown in FIG. 1, the motion stereo technique uses a plurality of images I0 and I1 captured by different cameras from different viewpoints at different times t0 and t1 as shown in FIG. This is a technique for calculating the three-dimensional coordinate position of an obstacle based on the amount of movement (parallax) of the object.

(Step S3)
Next, the obstacle determination unit 103 determines whether the obstacle detected by the obstacle detection unit 102 is the pillar P. Specifically, the obstacle determination unit 103 determines that the obstacle is the pillar P when the height of the obstacle is equal to or greater than a predetermined height and the width of the obstacle is equal to or less than a predetermined length. judge.

  Here, the predetermined height is a height set in advance, and is preferably set to, for example, 0.5 m. The predetermined length is a length set in advance, and is preferably set to 3 cm.

  The obstacle determination unit 103 removes noise by detecting an obstacle other than the pillar P, such as a vehicle or a person, which is lower than the pillar P. Thereby, the pillar P can be detected more accurately.

  When it is determined that the obstacle is the pillar P (YES in step S3), the process proceeds to step S4. If it is not determined that the obstacle is the pillar P (NO in step S3), the process returns to step S2.

(Step S4)
Next, the feature point extracting unit 104 determines whether or not the obstacle determining unit 103 determines that a plurality of obstacles are the pillars P. If a plurality is determined (YES in step S4), the process proceeds to step S5. If not (NO in step S4), the process returns to step S2.

(Step S5)
The feature point extracting unit 104 extracts a feature point FP in an image region between the plurality of columns P. Here, the feature point FP is a portion where the linear members intersect and intersect. Specifically, the feature point FP is a portion where the wire-like metal or the thin synthetic resin or the like constituting the lattice-shaped fence F intersects and intersects. It is shown by the intersection of the X mark in FIG. Here, the lattice shape is a state in which a plurality of vertical lines and horizontal lines intersect, and includes a repetitive shape such as a regular hexagonal lattice (honeycomb shape) in addition to a rectangular lattice.

  The feature point extracting unit 104 scans a region occupying a predetermined area in the image region between the plurality of columns P in a substantially straight line in the vertical and horizontal directions, and extracts a change point in luminance. The extracted portion is a portion where wire-shaped metal, thin synthetic resin, and the like intersect and intersect. The feature point extraction unit 104 extracts, as feature points FP, change points at which the intervals are substantially the same. Thereafter, the process proceeds to step S6.

(Step S6)
Next, the fence determination unit 105 collates the feature points FP stored in the storage unit 107 in advance with the feature points FP by pattern recognition or template matching, and calculates a correlation value between the two. The fence determining unit 105 determines whether there is a feature point FP in which the calculated correlation value is equal to or larger than a predetermined threshold.

  If there is a feature point FP whose correlation value is equal to or larger than the threshold (YES in step S6), the process proceeds to step S7. If there is no feature point FP whose correlation value is equal to or larger than the threshold (NO in step S6), the process returns to step S5.

(Step S7)
Next, the fence determination unit 105 determines whether or not the number of feature points FP whose correlation value is equal to or greater than a threshold is equal to or greater than a predetermined number. If the number of the feature points FP is equal to or more than the predetermined number (YES in step S7), the fence determination unit 105 determines that there is a fence between the columns P (step S8). If the number of feature points FP is less than the predetermined number (NO in step S7), the process proceeds to step S9.
(Step S9)
Next, the fence determination unit 105 determines whether the determination has been completed for all the columns. When the determination has been completed for all the columns (YES in step S9), a series of processing performed by the fence detection device 100 is completed. On the other hand, if the determination for all the columns is not completed, the process returns to step S5.

  In the first embodiment described above, the fence determining unit 105 determines that the fence F exists between the columns P when the plurality of feature points FP extracted by the feature point extracting unit 104 satisfy a predetermined condition. This makes it possible to detect the fence F, which is difficult to detect with sonar.

  As described above, the embodiments of the present invention have been described in detail with reference to the drawings. However, since the embodiments are merely examples of the present invention, the present invention is not limited only to the configurations of the embodiments, and the present invention is not limited thereto. Even if there is a change in the design or the like without departing from the gist, it is needless to say that the change is included in the present invention.

  In the above-described first embodiment, an example has been described in which an obstacle is detected based on a plurality of images captured by the single front camera 50 at different times t0 and t1 while the vehicle 1 is moving. The present invention is not limited to this mode. For example, a plurality of cameras installed at a position different from the front part of the vehicle 1 may cause a failure based on a plurality of images simultaneously captured at one specific time. An object may be detected.

  In the above-described first embodiment, the example in which the imaging unit is configured by the front camera 50 attached to the front part of the vehicle 1 has been described. The present invention is not limited to this mode. For example, the imaging unit may be a rear camera attached to the rear of the vehicle 1, a right camera attached to the right side of the vehicle 1, or a left camera attached to the left side of the vehicle 1. You may comprise.

  In the first embodiment described above, an example has been described in which the feature points FP are portions where the linear members intersect and intersect. The present invention is not limited to this mode. For example, the feature point FP may be a portion having substantially the same contour shape.

1 ... vehicle 2 ... other vehicle (obstacle)
10 fence detection system 50 front camera (imaging unit)
100 fence detection device 101 image acquisition unit 102 obstacle detection unit 103 obstacle determination unit 104 feature point extraction unit 105 fence determination unit 106 movement Detector F Fence FP Feature points I0, I1 Image P Support (bar-shaped member)

Claims (4)

An image acquisition unit that acquires an image obtained by imaging the periphery of the vehicle with an imaging unit mounted on the vehicle,
Based on the plurality of images acquired by the image acquisition unit, an obstacle detection unit that detects an obstacle existing around the vehicle,
An obstacle determination unit that determines whether the obstacle detected by the obstacle detection unit is a rod-shaped object,
When a plurality of the obstacles are determined to be the rod-shaped objects in the obstacle determination unit, a feature point extraction unit that extracts a feature point in an image region between the determined plurality of the rod-shaped objects. ,
A fence detection unit that determines that there is a fence between the rod-shaped objects when the plurality of feature points extracted by the feature point extraction unit satisfy a predetermined condition. . The fence detection device according to claim 1,
The fence detection device is characterized in that the feature points are portions that cross and intersect. In the fence detection device according to claim 1 or 2,
The predetermined condition is that the plurality of feature points are arranged substantially linearly in the vertical direction and the horizontal direction, the intervals between the feature points are substantially the same, and the feature points are a predetermined number or more within a predetermined area. A fence detection device, characterized in that it is present. In the fence detection device according to any one of claims 1 to 3,
Based on the information about the behavior of the vehicle, comprising a movement detection unit that detects that the vehicle is moving,
The obstacle detection unit detects the obstacle present in the vicinity based on the plurality of images acquired at the different times when it is detected that the vehicle is moving. Fence detection device.