JP2019117522A - Obstacle detection device - Google Patents

Abstract

To provide an obstacle detection device capable of improving detection accuracy of an obstacle.SOLUTION: An obstacle detection device detecting an obstacle within an imaging range of a stereo camera comprises: a position measurement part measuring a position of an edge obtained using stereo vision from a pair of images imaged by the stereo camera; a data processing part performing reduction processing for reducing measurement points in a processing target area according to the number of measurement points measured by the position measurement part present in the processing target area indicating an arbitrary area in the imaging range; and a detection part detecting the obstacle based on the positions of the measurement points after being reduced by the data processing part.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an obstacle detection device.

  Conventionally, an obstacle detection device is disclosed that detects an obstacle by measuring the distance of the obstacle using stereo vision from a pair of images captured by a stereo camera (see, for example, Patent Document 1). ).

  Specifically, in the obstacle detection device of this type, an edge indicating the outline of the obstacle is extracted from each of the pair of images, and an edge extracted from one of the images and the other image corresponding to the edge The parallax between both images for the edge of the same part is determined from the edge of. Then, an obstacle is detected by calculating the distance of the edge of the same part using this parallax.

Patent No. 5073700

  By the way, in the above-mentioned obstacle detection device, in an image picked up by a stereo camera, a portion having a large difference in luminance value between adjacent pixels in the left and right direction is recognized as an edge. Therefore, there is a possibility that a texture such as fallen leaves on the ground or a tile pattern of a road surface may be recognized as an edge, and an obstacle may not be detected correctly.

  Further, in the obstacle detection device, the edge of the other image corresponding to the edge extracted from one image (hereinafter referred to as “corresponding edge”) is searched for corresponding points by template matching. However, when there are a plurality of regions having high similarity to the edge extracted from one image in the other image, the corresponding point search may fail. In this case, an edge not corresponding to an edge extracted from one of the images is extracted as a corresponding edge. Therefore, if an obstacle is present at a position where there is no obstacle, the obstacle is detected correctly. You may not be able to

  The present invention has been made in view of such circumstances, and an object thereof is to provide an obstacle detection device capable of improving the detection accuracy of an obstacle.

  One aspect of the present invention is an obstacle detection device for detecting an obstacle within an imaging range of a stereo camera, which is a position of an edge obtained using stereo vision from a pair of images imaged by the stereo camera The measurement points in the processing target area are determined according to the number of measurement points measured by the position measuring section present in the processing target area indicating a position measuring section and an arbitrary area in the imaging range. Obstacle detection characterized by comprising: a data processing unit that executes reduction processing to be reduced; and a detection unit that detects the obstacle based on the position of the measurement point after being reduced by the data processing unit. It is an apparatus.

  One embodiment of the present invention is the obstacle detection device described above, wherein the data processing unit sets a predetermined area on the same plane as the processing target area as the reduction processing, and the set processing target area If the number of measurement points is equal to or greater than the first threshold value, the number of measurement points in the processing target area is reduced to a number equal to the first threshold value.

  One embodiment of the present invention is the obstacle detection device described above, wherein the data processing unit sequentially sets one measurement point as a target measurement point among a plurality of measurement points, and sets the target measurement point set. When the number of measurements in the processing target area at the center is equal to or greater than the first threshold, the processing for deleting the target measuring point is executed for each target measuring point to obtain the processing target area. The number of measurement points is reduced to a number equal to the first threshold.

  One embodiment of the present invention is the obstacle detection device described above, wherein the data processing unit performs the reduction processing when the number of the measurement points does not exist in the processing target area as a second threshold or more. The number of measurement points in the processing target area is reduced to a number equal to a predetermined value.

  One aspect of the present invention is the obstacle detection device described above, wherein the predetermined value is zero.

  One embodiment of the present invention is the obstacle detection device described above, wherein the data processing unit sequentially sets one measurement point as a target measurement point among a plurality of measurement points, and sets the target measurement point set. When the number of measurements in the processing target area at the center does not exist more than the second threshold, the processing of deleting the target measuring point is executed for each target measuring point to obtain the processing target area. The number of measurement points is reduced to a number equal to the predetermined value.

  One embodiment of the present invention is the obstacle detection device described above, wherein the data processing unit sets a predetermined region on the same plane as a first processing target region, and the set first processing target region If the number of measurement points is equal to or greater than the first threshold value, the first process of reducing the number of measurement points in the first processing target area to a number equal to the first threshold predetermined value is performed. In the case where the number of the measurement points does not exceed the second threshold in the second processing target area different from the first processing target area after the first processing unit to be executed and the first processing. The number of measurement points in the second processing target area is reduced to a number equal to a predetermined value and a predetermined value.

  As described above, according to the present invention, the detection accuracy of an obstacle can be improved.

It is a figure showing an example of the schematic structure of the obstacle detection system concerning one embodiment of the present invention. It is a figure which shows each positional information on several edge measured by the position measurement part 33 which concerns on one Embodiment of this invention. It is a figure which sets up the 1st processing object field H1 in the 1st reduction processing concerning one embodiment of the present invention. It is a diagram illustrating the deletion of interest measurement point P ₀₁ in the first reduction process according to an embodiment of the present invention. It is a figure which shows the measurement point P after the 1st reduction process which concerns on one Embodiment of this invention. It is a figure which sets up the 2nd processing object field H2 in the 2nd reduction processing concerning one embodiment of the present invention. It is a diagram illustrating the deletion of interest measurement point P ₀₂ in the second reduction process according to an embodiment of the present invention. It is a figure which shows the measurement point P after the 2nd reduction process which concerns on one Embodiment of this invention. It is a figure which shows the flow of the detection method of the obstruction of the obstruction detection apparatus 2 which concerns on one Embodiment of this invention.

  Hereinafter, an obstacle detection system according to an embodiment of the present invention will be described using the drawings.

  For example, an obstacle detection system according to an embodiment of the present invention is a system provided in a mobile body capable of autonomous traveling and detecting an obstacle in the forward direction of the mobile body. The moving body may be an autonomously traveling vehicle or a robot. Also, the mobile body may be manned or unmanned.

  FIG. 1 is a view showing an example of a schematic configuration of an obstacle detection system A according to an embodiment of the present invention. As shown in FIG. 1, an obstacle detection system A includes a stereo camera 1 and an obstacle detection device 2.

  The stereo camera 1 is provided on the moving body so as to capture the traveling direction of the moving body. The stereo camera 1 acquires a pair of images including parallax in the traveling direction of the moving body, and outputs the pair of images to the obstacle detection device 2. Specifically, the stereo camera 1 includes a first camera 11 and a second camera 12 which are a pair of cameras.

  The first camera 11 is mounted on a mobile body, and is provided on the right side in the forward direction of the mobile body. The first camera 11 captures the traveling direction of the moving object, and outputs the captured image (hereinafter, referred to as “first captured image”) to the obstacle detection device 2.

The second camera 12 is mounted on the moving body, and is provided on the left side in the forward direction of the moving body. The second camera 12 captures the traveling direction of the moving object, and outputs the captured image (hereinafter, referred to as “second captured image”) to the obstacle detection device 2.
The first camera 11 and the second camera 12 capture images at the same timing. However, "the same" is not limited to the completely same time.

  Thus, in the stereo camera 1, obstacles present in the traveling direction of the moving object by the first camera 11 and the second camera 12 arranged separately in the horizontal direction are divided into a plurality of directions having Image from the direction) at the same time. Then, the stereo camera 1 outputs, to the obstacle detection device 2, the first captured image and the second captured image obtained by simultaneously capturing images. In addition, when not distinguishing each of a 1st captured image and a 2nd captured image, it only calls it a "captured image."

The obstacle detection device 2 measures the distance of an obstacle in front of the moving object (an obstacle within the imaging range of the stereo camera 1) using stereo vision from the first captured image and the second captured image. It is an apparatus which detects the said obstacle.
Specifically, as shown in FIG. 1, the obstacle detection device 2 includes a measurement unit 3, a data processing unit 4 and a detection unit 5.

  The measurement unit 3 includes an edge detection unit 31, a search unit 32, and a position measurement unit 33.

  The edge detection unit 31 acquires a first captured image and a second captured image from the stereo camera 1. And the edge detection part 31 detects the edge which shows the outline of an obstruction in each of the acquired 1st captured image and 2nd captured image. The edge is a portion in the captured image where the difference in luminance value between pixels adjacent in the left-right direction is equal to or greater than a predetermined value.

  The search unit 32 uses, for example, template matching or the like for an edge of the second captured image (hereinafter referred to as “corresponding edge”) corresponding to the edge of the first captured image detected by the edge detection unit 31. Find corresponding points with. Then, the search unit 32 obtains the parallax between both the images with respect to the edge of the same part obtained by searching the corresponding points.

  The position measurement unit 33 measures the position information of the edge from the parallax obtained by the search unit 32 using the principle of triangulation. FIG. 2 is a diagram showing each piece of position information of a plurality of edges measured by the position measurement unit 33 according to an embodiment of the present invention.

  As shown in FIG. 2, this position information is three-dimensional coordinates with the center of the stereo camera 1 as the origin 0 in the three-dimensional space. Each measurement point P indicates the position (position information) in the three-dimensional space of each edge measured by the position measurement unit 33. The Y direction indicates the traveling direction of the moving body, the X direction indicates the lateral direction of the moving body orthogonal to the traveling direction, and the Z direction is a direction orthogonal to both the X direction and the Y direction. , Vertical direction. Further, the center of the stereo camera is the center between the position of the first camera 11 and the position of the second camera 12.

  As described above, the position measurement unit 33 acquires, as three-dimensional coordinates, the position of an edge obtained using stereo vision from a pair of captured images captured by the stereo camera 1.

  The data processing unit 4 performs reduction processing for reducing the measurement points P in the processing target area according to the number of measurement points P present in the processing target area indicating an arbitrary area within the imaging range of the stereo camera 1 Run. The configuration of the data processing unit 4 according to an embodiment of the present invention will be described below.

  The data processing unit 4 includes a first processing unit 41 and a second processing unit 42.

The first processing unit 41 acquires position information (three-dimensional coordinates) of the three-dimensional space at each measurement point P measured by the position measurement unit 33. The first processing unit 41 sets a predetermined area on the same plane as the processing target area in the first processing target area H1 in the three-dimensional space. Then, when the number of measurement points P in the set first process target area H1 is equal to or greater than the first threshold value P _th1 , the first processing unit 41 measures the first process target area H1. A first reduction process is performed to reduce the number of points to a number equal to the first threshold P _th1 .

  The first reduction process is a process of removing an edge which is not an obstacle such as fallen leaves on the ground or a texture (pattern of a tile on the ground, etc.).

Hereinafter, the first reduction process will be described with reference to FIGS. In FIG. 3 to FIG. 5, .circle- _solid . Indicates a measurement point P ( _P.sub.out1 ) existing outside the first processing target area H1, and .smallcircle. Indicates a measurement point P existing in the first processing target area H1 It shows _Pin1 ).

First, as shown in FIG. 3, the first processing unit 41 selects one arbitrary measurement point from the plurality of measurement points P measured by the position measurement unit 33, and measures the selected measurement point P to be focused on Set to point P ₀₁ . Then, the first processing unit 41 sets a first processing target area H1 centered on the set target measurement point P ₀₁ in the three-dimensional space.

  The first processing target area H1 may be a predetermined area on the same plane in a three-dimensional space, and is not particularly limited to the shape of the area. Also, the position on the same plane does not have to match the position in the Z direction, and even if the area in the Z direction has a width that can be regarded on the same plane (for example, a degree that can include a predetermined error). Good. In the present embodiment, the first processing target area H1 is an area indicated by a rectangular parallelepiped having a length ΔX1 in the X-axis direction, a length ΔY1 in the Y-axis direction, and a length ΔZ1 in the Z-axis direction.

The first processing unit 41 is illustrated in FIG. 4 when the measurement point P _in1 present in the set first process target area H1 is equal to or more than the first threshold P _th1 (for example, P _th1 = 5). As such, the target measurement point P ₀₁ is deleted. The first processing unit 41 performs the process of deleting the target measurement point P ₀₁ when the measurement point P _in1 present in the first processing target area H 1 is equal to or more than the first threshold P _th1. Is executed (execution of the first process), the number of measurement points P in the first processing target area H1 is adjusted to a number equal to the first threshold value P _th1 .

In other words, the first processing unit 41 sequentially sets each measurement point P as the target measurement point P ₀ among the plurality of measurement points P, and sets the first processing target centered on the set target measurement point P _01. measurement number _{P in1} in the region H1 is the case where the first threshold value _{P th1} or more, a process of deleting a target measurement point _{P 01,} run for each target measurement point _{P 01.} Thereby, as shown in FIG. 5, the first processing unit 41 sets the number of measurement points P _in1 in the first processing target area H ₁ , that is, the number of non-obstacles such as fallen leaves on the ground to the first. It can be reduced to a number equal to the threshold P _th1 .

The second processing unit 42 sets the number of measurement points P to a second threshold value P _{th2 in} the second processing target area H2 different from the first processing target area H1 in the three-dimensional space after the first processing. If the above does not exist, the second reduction process is executed to reduce the number of measurement points P in the second threshold value P _th2 to a number equal to the predetermined value. The predetermined value is a value less than the second threshold value P _th2 and is, for example, zero.

  The second reduction process is a process of removing the erroneously detected edge when the corresponding point search fails and the position of the edge is erroneously detected.

Hereinafter, the second reduction process will be described with reference to FIGS. In FIGS. 6 to 8, a _{black circle} indicates a measurement point P (P _out2 ) existing outside the second processing target area H2, and a _circle indicates a measurement point P existing in the second processing target area H2 Indicates P _in2 ).

First, as shown in FIG. 6, the second processing unit 42 selects one arbitrary measurement point from among measurement points P remaining in the three-dimensional space after the first reduction processing is executed, and selects the selected one. The obtained measurement point P is set to the target measurement point P ₀₂ . Then, the second processing unit 42 sets the second processing target region H2 around the target measurement point P ₀₂ which is set in the three-dimensional space.

  The second processing target area H2 is an area indicated by a rectangular solid having a length ΔX2 in the X-axis direction, a length ΔY2 in the Y-axis direction, and a length ΔZ2 in the Z-axis direction. The shape of the second processing target area H2 may be different depending on whether the edge detected by the edge detection unit 31 is a vertical edge or a horizontal edge. For example, as in the present embodiment, when the edge detected by the edge detection unit 31 is a vertical edge, the length ΔZ2 in the Z-axis direction is the length ΔX2 in the X-axis direction and the length ΔY2 in the Y-axis direction. Set the box to be longer than it. On the other hand, when the edge detected by the edge detection unit 31 is a horizontal edge, a rectangular solid whose length ΔX2 in the X axis direction is longer than the length ΔY2 in the Y axis direction and the length ΔZ2 in the Z axis direction Set to However, the shape of the second processing target area H2 is not limited to a rectangular solid, and may have a certain area, for example, may be a cube or an elliptical shape.

If the measurement point P _in2 present in the set second process target area H2 does not exist equal to or more than the second threshold P _th2 (for example, P _th2 = 8), the second processing unit 42 performs the process shown in FIG. As shown in, the target measurement point P ₀₂ is deleted. The second processing unit 42 applies, to all measurement points P, the process of deleting the target measurement point P ₀₂ when the measurement point P _in2 present in the second processing target area H ₂ is not the second threshold P _th2 or more. The number of measurement points P in the second processing target area H2 is adjusted to a predetermined value (for example, zero) by executing the processing (execution of the second processing).

In other words, the second processing unit 42 sequentially sets each measurement point P as the target measurement point P ₀₂ among the plurality of measurement points P, and the second processing target centered on the set target measurement point P ₀₂ when the measurement number _{P in2} in region H2 is not present the second threshold value _{P th2} or more, the process of deleting a target measurement point _{P 02,} run for each target measurement point _{P 02.} As a result, as shown in FIG. 8, the second processing unit 42 sets the second processing target area H2 in the second processing target area H2 in which the measurement point P _in2 having the second threshold P _th2 or more does not exist. The number of measurement points P _in2 , that is, the number of erroneously detected edges can be reduced.

  The detection unit 5 detects an obstacle based on the position information of the measurement point P remaining in the three-dimensional space after performing the second reduction process. For example, the detection unit 5 may detect the obstacle by determining that an obstacle is present at the position of the measurement point P remaining in the three-dimensional space. In this case, the distance information of the obstacle can be obtained from the position information of the measurement point P corresponding to the obstacle.

  Next, a method of detecting an obstacle in the obstacle detection device 2 according to an embodiment of the present invention will be described with reference to FIG. FIG. 9 is a diagram showing a flow of a method of detecting an obstacle in the obstacle detection device 2 according to an embodiment of the present invention.

  The edge detection unit 31 acquires a first captured image and a second captured image which are a pair of images captured simultaneously by the stereo camera 1 from different directions in front of the moving body (step S101). Then, the edge detection unit 31 detects an edge in each of the acquired first and second captured images (step S102).

  The search unit 32 identifies the corresponding point edge which is the edge of the second captured image corresponding to the edge of the first captured image detected by the edge detection unit 31 by performing so-called corresponding point search (step S103). . Then, the search unit 32 obtains the parallax between both images with respect to the edge of the same part using the edge of the first captured image and the corresponding point edge of the second captured image corresponding to the edge. Then, the position measuring unit 33 measures edge position information using the principle of triangulation from the parallax obtained by the searching unit 32 (step S104).

Next, first processing unit 41, among the plurality of measurement points P position measuring unit 33 has measured, selects one arbitrary measurement point, and sets the selected measurement points P in the target measurement point P ₀₁ (Step S105). Then, the first processing unit 41 sets, in the three-dimensional space, a first processing target area H1 indicating a predetermined area on the same plane with the set target measurement point P ₀₁ as a center (step S106).

The first processing unit 41 determines whether the number of measurement points P in the set first processing target area H1 is equal to or greater than a first threshold P _th1 (step S107). If the first processing unit 41 determines that the number of measurement points P in the first processing target area H1 is equal to or greater than the first threshold value P _th1 , the first processing unit 41 deletes the target measurement point P ₀₁ and proceeds to step S109. Proceed (step S108). On the other hand, when the first processing unit 41 determines that the number of measurement points P is less than the first threshold P _{th1 in} the first processing target area H1, the first processing unit 41 does not delete the target measurement point P _01. The process proceeds to step S109 (step S108).

The first processing unit 41 determines whether or not the processing of steps S106 to S108 has been performed on all the measurement points P present in the three-dimensional space. That is, at the measurement point P existing in the three-dimensional space, the first processing unit 41 determines whether or not there is a measurement point not set as the target measurement point P ₀₁ (hereinafter referred to as “unset measurement point”). Is determined (step S109). The first processing unit 41, when it is determined that there is not set measurement points of interest measurement point P _01, of the unset measuring point, by setting one of the measurement points in the interest measurement point P ₀₁ (step S110), the process proceeds to step S106. By this process, the process of steps S106 to S108 can be performed on each of all the measurement points P. Therefore, non-obstacle edges such as fallen leaves on the ground and textures (patterns of ground tiles, etc.) can be removed.

When it is determined that the first processing unit 41 determines that there is no unset measurement point of the measurement point of interest P _{01 at} the measurement point P existing in the three-dimensional space, the second processing unit 42 Among the remaining measurement points P, one arbitrary measurement point is selected, and the selected measurement point P is set as the target measurement point P ₀₂ (step S 111). Then, the second processing unit 42 sets a second processing target area H2 centered on the set target measurement point P ₀₂ in the three-dimensional space (step S112).

The second processing unit 42 determines whether or not the measurement point P present in the set second processing target area H2 is _{equal to} or more than a second threshold value P _th2 . That is, the second processing unit 42 determines whether or not the measurement point P present in the second processing target area H2 is smaller than the second threshold P _th2 (step S113).

If the second processing unit 42 determines that the measurement point P present in the second processing target area H2 is less than the second threshold value P _th2 , the second processing unit 42 deletes the attention measurement point P ₀₂ and proceeds to step S115. The process proceeds to step S114. On the other hand, when the second processing unit 42 determines that the measurement point P existing in the second processing target area H2 is equal to or more than the second threshold value P _th2 , the second processing unit 42 does not delete the target measurement point P _02. The process proceeds to step S115.

The second processing unit 42 determines whether or not the processing of steps S112 to S114 has been performed on all the measurement points P existing in the three-dimensional space. That is, the second processing unit 42 determines at the measurement point P existing in the three-dimensional space, whether or not there are still set measuring points that have not been set as interest measurement points P ₀₂ (step S115). The second processing unit 42, when it is determined that there is not set measurement points of interest measurement point P _02, of the unset measuring point, by setting one of the measurement points in the interest measurement point P ₀₂ (step In step S116, the process proceeds to step S112. By this process, the processes of steps S112 to S114 can be performed on each of all the measurement points P remaining in the three-dimensional space after the first reduction process. Therefore, when the corresponding point search fails and the edge position is erroneously detected, the erroneously detected edge can be removed.

If the second processing unit 42 determines that there is no unset measurement point of the measurement point P ₀₂ of interest at the measurement point P existing in the three-dimensional space, the detection unit 5 remains in the three-dimensional space An obstacle existing in the forward direction of the mobile object is detected based on the position information of the existing measurement point P (step S117).

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes design and the like within the scope of the present invention.

(Modification 1) In the above embodiment, the second reduction processing is performed after the first reduction processing, but the present invention is not limited to this. For example, in the present embodiment, the data processing unit 4 may execute at least one of the first reduction process and the second reduction process. For example, the data processing unit 4 executes a first reduction process on each measurement point P measured by the position measurement unit 33. Then, the detection unit 5 detects an obstacle based on the measurement points P remaining in the three-dimensional space after the first reduction processing is performed. This makes it possible to remove edges that are not obstacles such as fallen leaves and textures (patterns of ground tiles, etc.) on the ground. Therefore, the detection accuracy of the obstacle can be improved.
Also, for example, the data processing unit 4 executes the second reduction process on each measurement point P measured by the position measurement unit 33 without executing the first reduction process. Then, the detection unit 5 detects an obstacle based on the measurement points P remaining in the three-dimensional space after the second reduction process is performed. Thereby, when the corresponding point search fails and the position of the edge is erroneously detected, the erroneously detected edge can be removed. Therefore, the detection accuracy of the obstacle can be improved.

(Modification 2) Although the first camera 11 and the second camera 12 are arranged horizontally in the above embodiment, the present invention is not limited to this. For example, the first camera 11 and the second camera 12 may be vertically aligned. In this case, the horizontal edge is extracted by so-called vertical stereo vision, and the position of the horizontal edge is measured by the position measurement unit 33.

(Modification 3) Although the above embodiment has described the case where the stereo camera 1 includes two cameras (the first camera 11 and the second camera 12), the present invention is not limited to this. For example, the stereo camera 1 may have two or more cameras.

  As described above, the obstacle detection device A according to the present embodiment is an obstacle detection device that detects an obstacle within the imaging range of the stereo camera 1. And obstacle detection device A performs reduction processing which reduces measurement points P in processing object areas according to the number of measurement points P which exist in processing object areas which show the arbitrary fields in the imaging range. Do.

  According to such a configuration, it is possible to remove an edge indicating fallen leaves scattered on the ground or an edge existing alone due to false detection. This makes it possible to remove non-obstacles and to improve the detection accuracy of obstacles.

For example, as the reduction processing, the data processing unit 4 sets a predetermined area on the same plane as the processing target area, and the number of measurement points P in the set processing target area is equal to or more than the first threshold P _th1. In this case, the number of measurement points in the processing target area is reduced to a number equal to the first threshold value P _th1 .

  According to such a configuration, it is possible to delete an edge indicating a texture such as fallen leaves on the ground or a tile pattern of a road surface. Therefore, it is possible to suppress the recognition of a texture such as fallen leaves on the ground or a tile pattern of a road surface as an obstacle and correctly detect the obstacle.

For example, as the reduction processing, the data processing unit 4 sets the number of measurement points P in the processing target area to a predetermined value (for example, when the number of measurement points P does not exist in the processing target area as the second threshold P _th2 or more). Reduce to zero).

  According to such a configuration, it is possible to delete an edge that is erroneously detected in the corresponding point search. Therefore, it is possible to suppress false detection when there is an obstacle at a position where there is no obstacle, and the obstacle can be detected correctly.

As described above, in the present embodiment, it has been found that the edge such as the texture of the ground exists on the same plane, and the edge erroneously detected by the corresponding point search is dispersed over a wide space. Then, the obstacle detection device A executes the process of at least one of the first process and the second process on the edge that does not exist collectively from such a feature, so that the non-obstacle edge is detected. Remove.
Note that by executing the second process after executing the first process, it is possible to remove both an edge such as a texture on the ground and an edge that is erroneously detected by a corresponding point search, so that obstacle detection accuracy Can be further improved.

  Therefore, detection and false detection of an object that does not need to be recognized as an obstacle are reduced, and when the obstacle detection system A is mounted on a vehicle, it is safe and does not adversely affect the autonomous traveling of the vehicle. Stable obstacle detection will be possible.

  Note that each unit included in the obstacle detection device A according to the present embodiment performs various processes related to control of at least one of the first process and the second process of the obstacle detection apparatus A, which is computer readable A program recorded on a recording medium may be installed and configured to be executed by causing a computer to execute the program. That is, by causing the computer to execute a program that performs various processes related to control of at least one of the first process and the second process of the obstacle detection device A, the computer can be used as each unit included in the obstacle detection device A. The obstacle detection device A may be configured by functioning. The computer has various memories such as a CPU, ROM, RAM, EEPROM (registered trademark), a communication bus, and an interface, and the CPU reads processing programs stored beforehand in the ROM as firmware and sequentially executes them to detect obstacles. It functions as the device A.

A Obstacle detection system 1 stereo camera 2 obstacle detection device 3 measurement unit 4 data processing unit 5 detection unit 31 edge detection unit 32 search unit 33 position measurement unit 41 first processing unit 42 second processing unit 11 first camera 12 Second camera

Claims (7)

An obstacle detection device for detecting an obstacle within an imaging range of a stereo camera, comprising:
A position measurement unit that measures the position of an edge obtained using stereo vision from a pair of images captured by the stereo camera;
The reduction processing for reducing the measurement points in the processing target area is executed according to the number of measurement points measured by the position measuring unit present in the processing target area indicating an arbitrary area in the imaging range. A data processing unit,
A detection unit that detects the obstacle based on the position of the measurement point after being reduced by the data processing unit;
An obstacle detection device comprising:   The data processing unit sets a predetermined area on the same plane as the processing target area as the reduction processing, and the number of measurement points in the set processing target area is equal to or more than a first threshold. The obstacle detection device according to claim 1, wherein the number of measurement points in the processing target area is reduced to a number equal to the first threshold.   The data processing unit sequentially sets one measurement point as a target measurement point among a plurality of measurement points, and the number of measurements in the processing target area centering on the set target measurement point is equal to or more than a first threshold In this case, the number of measurement points in the processing target area is reduced to a number equal to the first threshold value by executing the process of deleting the target measurement points for each target measurement point. The obstacle detection device according to claim 2, characterized in that   The data processing unit determines, as the reduction process, the number of measurement points in the processing area equal to a predetermined value when the number of measurement points does not exist in the processing area at the second threshold or more. The obstacle detection device according to claim 1, wherein the obstacle is reduced to   The obstacle detection device according to claim 4, wherein the predetermined value is zero.   The data processing unit sequentially sets one measurement point as a target measurement point among the plurality of measurement points, and the number of measurements in the processing target area centering on the set target measurement point is equal to or more than a second threshold If not present, the number of measurement points in the processing target area is reduced to a number equal to the predetermined value by executing the process of deleting the measurement points of interest for each measurement point of interest. The obstacle detection device according to claim 4 or 5. The data processing unit
When a predetermined area on the same plane is set as a first processing target area, and the number of measurement points in the set first processing target area is equal to or more than a first threshold, the first A first processing unit that executes a first process of reducing the number of measurement points in the processing target area to a number equal to the first predetermined threshold value;
After the first processing, when the number of measurement points is not more than a second threshold in a second processing target area different from the first processing target area, the second processing target area The obstacle detection device according to claim 1, wherein the number of measurement points within is reduced to a number equal to a predetermined value and a predetermined value.

Images