JP2013182561A - Level difference detector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a level difference detector that can highly accurately detect a level difference such as a curb and the like contained in an image photographing ahead of a vehicle.SOLUTION: The level difference detector comprises: a surround environment photograph part 10 that photographs a surround image of the vehicle; and an edge detection part 20 that detects an edge in a photographed image. The edge detection part 20 extracts a first edge along a travelling road of an own vehicle from a plurality of detected edges, and further extracts a plurality of second edges crossing the first edge from the plurality of detected edges. When two second edges positioned on one linear line are cut off at a portion crossing the first edge or mutually parallel two second edges exist and each second edge has discontinuity in the same direction, the first edge is determined to be an edge of the level difference.

Description

  The present invention relates to a step detection device for detecting a step such as a curb existing in front of a host vehicle.

  A front image of the vehicle is captured by the camera, and a partition line at the road edge (hereinafter described as a white line) is extracted from the captured image, and the vehicle deviates from the lane inside the white line at the road edge. There has been proposed a system for performing control so as not to travel. In such a system, it is necessary to detect a white line existing on a road with high accuracy. For example, a system disclosed in Japanese Patent Laid-Open No. 9-35195 (Patent Document 1) has been known. In Patent Document 1, a white line is detected using a luminance value of an image captured by a camera.

JP-A-9-35195

  However, in the white line detection method disclosed in Patent Literature 1 described above, a curb existing on the road may be erroneously recognized as a white line. That is, when a curb existing on the road is photographed, the curb in the image may have a high luminance value. In this case, the curb may be mistakenly recognized as a white line. As a result, there is a problem that it is impossible to recognize partition lines with high accuracy.

  The present invention has been made in order to solve such a conventional problem, and an object of the present invention is to detect a step such as a curb included in an image obtained by imaging the front of the vehicle with high accuracy. The object is to provide a possible step detector.

  In order to achieve the above object, the present invention provides an edge detection means for detecting an edge in an image captured by an imaging means, and a first edge along a vehicle traveling path from a plurality of edges detected by the edge detection means. Edge extracting means for extracting the second edge extracting means, and second edge extracting means for extracting a plurality of second edges intersecting the first edge from the plurality of edges detected by the edge detecting means. Furthermore, two second edges located on one straight line are interrupted at a portion intersecting the first edge, or there are two second edges parallel to each other, and each second edge is in the same direction. Step determining means for determining that the first edge is an edge of a step in the case of having the discontinuity.

  In the level difference detection device according to the present invention, the first edge and the second edge are extracted from the image ahead of the host vehicle, and whether or not the second edge is interrupted at a portion where the first edge intersects, Since it is determined whether or not the first edge is an edge due to a step depending on whether or not there are a plurality of second edges, it is possible to determine the step with high accuracy.

It is a block diagram which shows the structure of the level | step difference detection apparatus which concerns on one Embodiment of this invention, and its peripheral device. It is a flowchart which shows the process sequence of the level | step difference detection apparatus which concerns on one Embodiment of this invention. It is a figure which shows the 1st example of the edge image which extracted the surrounding image imaged by the surrounding environment imaging part and the edge of the level | step difference detection apparatus which concerns on one Embodiment of this invention. It is a figure which shows the 1st example of the image which extracted the 1st edge from the edge image of the level | step difference detection apparatus which concerns on one Embodiment of this invention. It is a figure which shows the 1st example of the image which extracted the 2nd edge from the edge image of the level | step difference detection apparatus which concerns on one Embodiment of this invention. It is a figure which shows the 2nd example of the edge image extracted by the surrounding environment imaging part of the level | step difference detection apparatus which concerns on one Embodiment of this invention, and the edge image which extracted the edge. It is a figure which shows the 2nd example of the image which extracted the 1st edge from the edge image of the level | step difference detection apparatus which concerns on one Embodiment of this invention. It is a figure which shows the 2nd example of the image which extracted the 2nd edge from the edge image of the level | step difference detection apparatus which concerns on one Embodiment of this invention. It is a figure which shows the edge image of N-1 step and the edge image of N step of the level | step difference detection apparatus which concerns on one Embodiment of this invention. It is a figure which shows the image which took AND of the edge image of N-1 step shown in FIG. 9, and the edge image of N step. It is a figure which shows the own vehicle shadow assumed from the position of the sun of the level | step difference detection apparatus which concerns on one Embodiment of this invention, and an actual own vehicle shadow. It is the figure which took the exclusive OR (XOR) of the own vehicle shadow assumed from the position of the sun shown in FIG. 11, and an actual own vehicle shadow.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a level difference detection device and peripheral devices thereof according to the first embodiment of the present invention.

  As shown in FIG. 1, the level difference detection device 100 includes a surrounding environment imaging unit 10, an edge detection unit 20, a level difference determination unit 30, and a map data storage unit 40, and a vehicle control unit that controls the traveling of the vehicle. 50.

  The ambient environment imaging unit 10 captures the surrounding environment of the host vehicle with a camera mounted on the host vehicle, and outputs the captured ambient environment image to the edge detection unit 20.

  The edge detection unit 20 detects an edge in the ambient environment image captured by the ambient environment imaging unit 10. Furthermore, the process which extracts the 1st edge along the driving | running route of the own vehicle from the detected some edge, and extracts the some 2nd edge which cross | intersects the 1st edge from the detected several edge I do. Then, the detected edge data and the data of the first edge and the second edge are output to the step determination unit 30. That is, the edge detection unit 20 has functions as an edge detection unit, a first edge extraction unit, and a second edge extraction unit.

  Based on the edge data detected by the edge detection unit 20, the step determination unit 30 determines whether or not the edge is an edge due to a step existing on the traveling road. Moreover, the level | step difference determination part 30 is provided with the memory 30a which memorize | stores the past edge information.

  The map data storage unit 40 stores map data of an area where the vehicle travels, and also stores information indicating whether or not there is a step on the road.

  The vehicle control unit 50 includes a host vehicle position estimation unit 50a and a steering control unit 50b. Based on the step edge data determined by the step determination unit 30 and the map data stored in the map data storage unit 40, the host vehicle position estimation unit 50a determines the position and azimuth of the host vehicle (the direction in which the vehicle is traveling). ). Further, the steering control unit 50b determines that the vehicle is in the lane (right and left) based on the vehicle position and azimuth angle estimated by the vehicle position estimation unit 50a and the map data stored in the map data storage unit 40. Steering control is performed so as to maintain the area between the white lines.

  Next, the operation of the level difference detection apparatus 100 according to the present embodiment configured as described above will be described with reference to the flowchart shown in FIG.

  First, in step S110, the ambient environment imaging unit 30 captures an image in front of the host vehicle with a camera, and acquires the surrounding image of the host vehicle. As a result, for example, a surrounding image M1 as shown in FIG. 3A is obtained as a first example. As a second example, for example, a surrounding image M2 as shown in FIG. 6A is obtained.

  In step S120, the edge detection unit 20 detects the edges of the surrounding images M1 and M2 acquired in the process of step S110. As a result, in the first example, an edge image including a plurality of edges as shown in FIG. 3B is obtained. In the second example, an edge image including a plurality of edges as shown in FIG. 6B is obtained. Here, the edge detection can use a known method such as the Canny method.

  In step S130, the edge detection unit 20 executes a first edge extraction process. In this process, based on the position and azimuth of the host vehicle (for example, the angle of the traveling direction of the host vehicle when the east direction is set to 0 °) and the map data stored in the map data storage unit 40, a step is performed. An edge along the traveling path of the host vehicle is extracted from the edge image detected in the process of S120, and this is defined as a first edge. Specifically, the edge group detected in the process of step S120 is converted into map data from the vehicle position and the azimuth angle, and is approximately parallel to information (for example, white line, curbstone, etc.) registered as a travel path in the map data. An edge (for example, an angle formed by a straight line between the detected edge and the traveling road direction is 2 ° or less) is extracted as the first edge. As a result, in the first example, edges L1, L2, and L3 indicated by thick lines in FIG. 4 are extracted as the first edges. In the second example, the edges L11, L12, and L13 indicated by the thick lines in FIG. 7 are extracted as the first edges.

  Next, in step S140, the edge detection unit 20 executes a second edge extraction process. In this process, a plurality of edges that intersect with the first edge extracted in the process of step S130 are extracted as second edges. As a result, in the first example, the edges L4a and L4b indicated by the thick lines in FIG. 5 are extracted as the second edges. That is, since the edges L4a and L4b intersect the first edge L1, they are extracted as the second edge. In the second example, the edges L14a, L14b, and L14c indicated by the thick lines in FIG. 8 are extracted as the second edges.

  In step S150, the level difference determination unit 30 executes a first level difference determination process. The first step determination process determines whether or not the first edge extracted in step S130 is a step based on whether or not the second edge extracted in step S140 described above is discontinuous. As a specific process, combinations of edges that are substantially on the extension line are extracted from the second edges. On the approximate extension line, for example, it is assumed that the difference between the slopes of the straight lines is not more than a predetermined value (for example, ± 2 °) and the distance between end points close to each other is not more than a predetermined value (for example, 1 m). When a combination of edges on the approximate extension line is extracted, the first edge having two end points close to each other is determined as an edge that generates a step.

  In the example shown in FIG. 5, the second edges L4a and L4b are discontinuous at the point where they intersect with the first edge L1, and each of the second edges L4a and L4b is on a generally extended line (one straight line). 1) and the distance between the end points of each of the second edges L4a and L4b is short (below a predetermined value), so the first edge L1 intersecting with each of the second edges L4a and L4b is an edge due to a step. It is judged that it is. Here, the first step only when the second edge intersects the first edge and there is a step in the area corresponding to the first edge in the map data stored in the map data storage unit 40. Detection may be performed. In this case, unnecessary step detection can be omitted.

  In step S160, the level difference determination unit 30 executes a second level difference determination process. The second step determination process determines whether or not the plurality of second edges extracted in the process of step S140 described above exist within a predetermined distance range (for example, within 50 m) with respect to the traveling direction of the vehicle. The step is determined based on whether or not the plurality of second edges are parallel to each other.

  For example, in the example shown in FIG. 8, three second edges L14a, L14b, and L14c are extracted, and each second edge L14a, L14b, and L14c exists within a predetermined distance range (for example, within 50 m), and These are substantially parallel to each other, and have a polygonal line shape in the same direction at a point intersecting the first edge L11. That is, there are a plurality of second edges L14a, L14b, and L14c parallel to each other, and each second edge has discontinuity in the same direction. In this case, since each of the second edges L14a, 14b, and 14c can be assumed to be an edge caused by the shadow of an object caused by sunlight, the first edge L11 is an edge that forms a step due to the discontinuity of the shadow line. Judge that there is. Here, also in the second step determination process, as in the first step determination process described above, the second edge intersects the first edge, and further, the second step determination process includes the first step in the map data stored in the map data storage unit 40. The second step detection may be performed only when there is a step in a region corresponding to one edge.

  In step S170, the level difference determination unit 30 executes a third level difference determination process. In the third step determination process, it is determined whether or not the first edge extracted in the process of step S130 is a step using the shadow of the own vehicle. When past edge information is stored in the memory 30a provided in the level difference determining unit 30, the edge information of the previous step (N-1 step) shown in FIG. The edge information as shown in FIG. 10 is obtained by taking “AND” (logical sum) of the edge information of the present step (N step) shown. And the edge of the position close to the own vehicle in this edge information is extracted as a shadow edge of the own vehicle. That is, in the example shown in FIG. 10, the edge L21 is extracted as being a shadow edge of the own vehicle.

  Then, based on the extracted shadow edge of the own vehicle, the relative direction of the sun with respect to the own vehicle is calculated, and the shape of the shadow of the own vehicle generated on the plane from the current sun direction is calculated. Alternatively, the shadow shape data of the own vehicle generated on the plane for each sun direction may be prepared in advance, and the corresponding shadow shape data of the own vehicle may be selected.

  Calculated (or selected) shadow shape (the edge of the vehicle shadow detected when the vehicle acquired in the past is traveling on a plane; see edge L22 in FIG. 11A) and extraction When the actual shadow edge of the own vehicle (see edge L23 in FIG. 11B) is obtained, the exclusive OR (XOR) of these is calculated to extract the distorted portion of the shadow of the own vehicle. . As a result, the edge L31 shown in FIG. 12 is extracted. Since the edge L31 has a polygonal line at the portion where the extracted edge L31 and the first edge L2 intersect, it is determined that the first edge L2 is one of the edges that generate a step. To do.

  Here, in order to calculate the relative direction of the sun with respect to the own vehicle from the extracted shadow of the own vehicle, for example, in the case of FIG. By converting the coordinates of the “real space”, the coordinates R of the shadow at the center of the vehicle front end in the real space can be obtained, and the angle formed by the straight line connecting the actual vehicle front end center and the coordinates R becomes the shadow direction θ. Since the sun is positioned 180 degrees opposite to the shadow direction θ, the sun direction can be recognized.

  In step S180, the vehicle position estimation unit 50a gives the attribute of “step edge” to the edge determined as a step in steps S150 to S170 among the edge information detected in the process of step S120. And the own vehicle position P and the azimuth angle (phi) of the own vehicle are calculated based on edge information and map information. Specifically, after converting the edge information into the real space, the vehicle position P and the azimuth angle φ are estimated so as to match the map information. When performing matching, the edge to which the attribute of “step edge” is assigned is matched with map data having step information (stored in the map data storage unit 40), and other edge information is Matches with map data without step information.

  In step S190, the steering control unit 50b performs steering control so that the host vehicle follows the lane based on the host vehicle position P, the azimuth angle φ estimated in the process of step S180, and the map information in the map data storage unit 40. I do.

  In step S200, it is determined whether the ignition is off. If the ignition is off, the process is terminated. If the ignition is on, the process returns to step S110.

  In this way, by comparing the first edge image and the second edge image, a step (curbstone, etc.) existing on the traveling road ahead of the vehicle can be detected with high accuracy and distinguished from the white line drawn on the road. Can be reliably performed.

  Further, in the present embodiment, a process is performed in which the edge detected in the image is coordinate-converted on the real space, or the map information is coordinate-converted on the image space. These conversions can be performed as follows.

(A) Conversion processing from the real space to the image space is executed by the following equation (1).

I = C2C1R (1)
Where I is a coordinate in the image space, R is a coordinate in the real space, C1 is a coordinate transformation matrix from the vehicle coordinate system to the camera coordinate system, and C2 is a coordinate transformation matrix from the camera coordinate system to the image space.

(B) The conversion process from the image space to the real space is executed by the following equation (2).

R = C4C3I (2)
Where R is a coordinate in the real space, I is a coordinate in the image space, C3 is a coordinate transformation matrix from the image space to the camera coordinate system, and C4 is a coordinate transformation matrix from the camera coordinate system to the vehicle coordinate system.

  Thus, in the step detection device according to the present embodiment, an edge that faces the traveling direction of the vehicle is extracted as a first edge, and a plurality of edges that intersect the first edge are extracted as second edges, Based on these relationships, it is determined whether or not the edge detected as the first edge is a step. Specifically, when two second edges located on one straight line are interrupted at a portion intersecting the first edge, and the distance between each second edge is less than a predetermined distance, It is determined that the first edge (L1 in FIG. 5) is an edge due to a step.

  In addition, when there are a plurality of second edges parallel to each other and each second edge has the same discontinuity at a portion where it intersects with the first edge, the first edge (L11 in FIG. 8). ) Is determined to be an edge due to a step. Accordingly, it is possible to detect a step such as a curb on the traveling road with high accuracy. As a result, it is possible to perform processing for estimating the position and azimuth of the host vehicle with high accuracy in comparison with map data.

  Further, when the two second edges are interrupted at a portion intersecting with the first edge and the distance between the end points of each second edge is equal to or less than a predetermined value, the two second edges are the same. Since it is assumed that the edge is due to an object, it is determined that the first edge is an edge due to a step. On the other hand, when the distance between the end points of the second edge exceeds a predetermined distance, there is a high possibility that the two second edges are not edges due to the same object, so the first edge is not determined to be an edge due to a step. Therefore, the accuracy of the step determination can be improved.

  Whether or not the first edge is a stepped edge based on the discontinuity of the second edge when a plurality of second edges parallel to each other exist within a predetermined distance (for example, within 50 m). Therefore, the accuracy of the step detection can be improved.

  Furthermore, since the level difference determination is performed only when level difference information exists in the map data, it is possible to prevent erroneous recognition that a level difference exists in a portion where no level difference exists, and to improve the accuracy of level difference detection.

  Further, in the third step determination process, the step is determined by comparing the shadow generated when the host vehicle is traveling on the plane with the shadow of the host vehicle included in the photographed image. Even when there is no second edge that intersects the first edge to be formed, the step can be determined. As a result, the vehicle position can be estimated with high accuracy.

  The step detecting device of the present invention has been described based on the illustrated embodiment. However, the present invention is not limited to this, and the configuration of each part is replaced with an arbitrary configuration having the same function. Can do.

  For example, in the embodiment described above, an example in which shadow edges (L4a, L4b in FIG. 5 and L14a, L14b, L14c in FIG. 8) existing in front of the host vehicle are extracted and step detection is performed using these edges will be described. However, even when a laser sensor or the like is used instead of the shadow, the step can be detected by the same method.

  INDUSTRIAL APPLICABILITY The present invention can be used for detecting a step such as a curbstone existing in the host vehicle with high accuracy.

DESCRIPTION OF SYMBOLS 10 Ambient environment imaging part 20 Edge detection part 30 Ambient environment imaging part 30 Level | step difference determination part 30a Memory 40 Map data storage part 50 Vehicle control part 50a Own vehicle position estimation part 50b Steering control part 100 Level | step difference detection apparatus

Claims (5)

Imaging means for capturing an image around the vehicle;
Edge detection means for detecting edges in the image captured by the imaging means;
First edge extraction means for extracting a first edge along the traveling path of the vehicle from a plurality of edges detected by the edge detection means;
Second edge extraction means for extracting a plurality of second edges that intersect the first edge extracted by the first edge extraction means from the plurality of edges detected by the edge detection means;
Two second edges located on one straight line are interrupted at a portion intersecting the first edge, or there are a plurality of second edges parallel to each other, and each second edge is in the same direction. Step determination means for determining that the first edge is an edge of a step when having discontinuity,
A step detecting device comprising:   In the step determining means, the two second edges located on the one straight line are interrupted at a portion intersecting the first edge, and the distance between the two second edges is a predetermined distance or less. 2, the step detection device according to claim 1, wherein the first edge is determined to be an edge of a step.   The step determining means determines that the first edge is an edge of a step when there are a plurality of second edges parallel to each other and each second edge is within a predetermined range. The level | step difference detection apparatus of Claim 1 characterized by the above-mentioned.   A map data storage unit for storing map data including information on a step existing on the road is further provided, wherein the step determination unit detects a step only at a place where the step exists in the map data. The level | step difference detection apparatus of any one of Claims 1-3. The edge detection means detects an edge of the own vehicle shadow based on edges acquired in time series,
The step determining means is an exclusive logic of an edge of the own vehicle shadow detected when the own vehicle acquired in the past is traveling on a plane and an edge of the own vehicle shadow acquired in the current shooting. The step detection device according to any one of claims 1 to 4, wherein the step is determined based on a shape in which a summed edge intersects the first edge.

Images