JP2016012264A - Article detection device, drive support apparatus, article detection method, and article detection program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an article detection device, an article detection method, and an article detection program, which enable more appropriate detection of a detection target article.SOLUTION: An article detection device includes: a vertical edge extraction part for extracting a vertical edge extending in a vertical direction in a pickup image picked up by an imaging part used for imaging the surroundings of a vehicle; an extended-area setup part for setting an extended area that has, as an edge, a vertical edge extracted by the vertical edge extraction part and a given width from the vertical edge toward a center direction of the pickup image; and a determination part for determining whether the vertical edge is consistent with a portion of a detection target article by using the extended area set by the extended-area setup part.

Description

  The present invention relates to an object detection device, a driving support device, an object detection method, and an object detection program.

  Conventionally, when the traveling state of the host vehicle is determined to be appropriate from at least one of the position in the vehicle width direction in the traveling lane of the host vehicle and the relative relationship of the preceding vehicle with the host vehicle, A collision possibility determination device for determining the possibility of collision with another vehicle excluding other vehicle detection means for detecting another vehicle existing ahead of the host vehicle and a preceding vehicle detected by the other vehicle detection means Relative position determination means for determining whether or not the relative positional relationship of the other vehicle to the own vehicle is the same for a predetermined time continuously, and based on the determination result of the relative position determination means, the own vehicle and the other vehicle There is known a collision possibility determination device including a collision possibility determination unit that determines the possibility of collision with (see, for example, Patent Document 1).

JP 2005-135025 A

However, in reality, there is a scene where it is necessary to monitor the other vehicle as a collision target even when the relative positional relationship of the other vehicle with respect to the host vehicle is not the same positional relationship for a predetermined time. Can occur. For this reason, in the conventional technique, there is a case where the detection of the collision target that should be detected originally occurs.
The present invention has been made in consideration of such circumstances, and an object of the present invention is to provide an object detection device, an object detection method, and an object detection program that can detect a detection target more appropriately. I will.

  According to the first aspect of the present invention, a vertical edge extraction unit (42) for extracting a vertical edge (VE) extending in the vertical direction in a captured image captured by the imaging unit (10) that captures the periphery of the vehicle (M). And an extended area setting unit (44) for setting an extended area (AVE) having a predetermined width from the vertical edge toward the center of the captured image, with the vertical edge extracted by the vertical edge extracting unit as one end And a determination unit (46) for determining whether or not the vertical edge corresponds to a part of the detection target using the extension region set by the extension region setting unit. .

  The invention according to claim 2 is the object detection device according to claim 1, wherein the determination unit is configured to perform the vertical edge based on a degree of overlap between the extended regions in a plurality of captured images captured in time series. Is determined to correspond to a part of the detection object.

  A third aspect of the present invention is the object detection device according to the second aspect, wherein the determination unit continues overlapping pixels that are included in the extended region and overlap each other in a plurality of captured images captured in time series. When the time (CT) is long, it is determined that the vertical edge corresponds to a part of the detection target.

  Invention of Claim 4 is an object detection apparatus of any one of Claim 1 to 3, Comprising: The said determination part is based on the speed of the horizontal direction in the said captured image of the said vertical edge, It is determined whether or not the vertical edge corresponds to a part of the detection target.

  The invention according to claim 5 is the object detection device according to claim 4, wherein the determination unit is configured such that the vertical edge is a detection target when the horizontal speed of the captured image of the vertical edge is small. It is determined that it corresponds to a part.

  A sixth aspect of the present invention is the object detection device according to the fourth or fifth aspect, wherein the determination unit includes pixels that are included in the extended region and overlap each other in a plurality of captured images captured in time series. An overlap duration is calculated, and an index value (V (VE)) using a difference between the overlap durations in the left and right end pixels of the extension region as a denominator is calculated, and the calculated index value is calculated as the vertical edge of the vertical edge. This is handled as the horizontal speed in the captured image.

  The invention according to claim 7 is the object detection device according to claim 6, wherein the determination unit has a small horizontal speed in the captured image of the vertical edge when the index value becomes an invalid value. Even in this case, it is not determined that the vertical edge corresponds to a part of the detection target.

  The invention according to claim 8 is the object detection device according to any one of claims 1 to 7, wherein the width of the extension region is set to be small when the distance between the vehicle and the detection target is long. When the distance between the vehicle and the detection object is short, the width of the extension area is set large.

  The invention according to claim 9 is the object detection device according to any one of 1 to 8, wherein when the horizontal coordinate in the captured image of the vertical edge is close to a central portion, the width of the extension region is set. When the horizontal coordinate in the captured image of the vertical edge is close to the left and right ends, the width of the extension area is set large.

  A tenth aspect of the present invention is the object detection device according to the eighth or ninth aspect, wherein the extension region setting unit and the determination unit perform processing in parallel on the widths of a plurality of extension regions, and perform the determination. A determination result selection unit that selects one determination result from the determination results of the widths of the plurality of extension regions by the unit.

  The invention according to claim 11 is a driving support device comprising: the object detection device according to any one of claims 1 to 10; and a driving support unit that performs driving support based on a detection result of the object detection device. It is.

  The invention according to claim 12 is to extract a vertical edge extending in a vertical direction from a captured image captured by an imaging unit that captures the periphery of the vehicle, and using the extracted vertical edge as one end, An object that sets an extension region having a predetermined width toward the center of the captured image, and determines whether or not the vertical edge corresponds to a part of the detection target using the set extension region. It is a detection method.

  The invention according to claim 13 causes a computer to extract a vertical edge extending in a vertical direction in a captured image captured by an imaging unit that captures the periphery of the vehicle, and uses the extracted vertical edge as one end. An extended area having a predetermined width is set from the edge toward the center of the captured image, and it is determined whether or not the vertical edge corresponds to a part of the detection object using the set extended area. This is an object detection program.

  According to the invention described in claims 1, 12, and 13, in the captured image captured by the imaging unit that captures the periphery of the vehicle, the vertical edge extending in the vertical direction is extracted, the vertical edge is one end, and from the vertical edge By setting an extended area with a predetermined width toward the center of the captured image and determining whether the vertical edge corresponds to a part of the detection target using the extended area, the detection target is more appropriately detected An object can be detected.

  According to the second and third aspects of the present invention, the vertical edge is determined based on the degree of overlap between the extended regions in the plurality of captured images captured in time series, that is, the degree of stagnation on the captured image of the object on which the vertical edge appears. Therefore, it is possible to detect the detection target more appropriately.

  According to the fourth to seventh aspects of the present invention, the vertical edge is detected based on the value indicating the horizontal speed in the captured image of the vertical edge, that is, the degree of stagnation on the captured image of the object on which the vertical edge appears. By determining whether or not it corresponds to a part of the detection object, it is possible to more appropriately detect the detection target.

  According to the eighth and ninth aspects of the invention, by dynamically changing the width of the extension region, it is possible to achieve both reduction in processing load and suppression of false detection and improvement in followability to an object. .

  According to the tenth aspect of the present invention, processing is performed in parallel for the widths of a plurality of extension areas, and one determination result is selected from the determination results for the widths of the plurality of extension areas. It is possible to prevent the determination result from being blurred.

  According to the eleventh aspect of the present invention, it is possible to perform appropriate driving assistance based on the detection target appropriately detected by the object detection device.

It is a figure which shows typically an example of a structure of the driving assistance apparatus 1 containing the object detection apparatus 5 which concerns on 1st Embodiment. It is a figure which shows the function structural example centering on the control apparatus 30 in 1st Embodiment. It is explanatory drawing for demonstrating the function of the object detection apparatus. It is a figure which shows a mode that the vertical edges VE1-VE3 are extracted in the captured image IM acquired with the camera 10. FIG. It is a figure which shows a mode that the expansion area | regions AVE1-AVE3 are set in the captured image IM. It is a figure which shows extension field AVE which overlaps with extension field AVE of the last image pick-up frame in a plurality of image pick-up frames. It is the figure which showed transition of count value CT and edge speed V (VE) when the horizontal direction coordinate of the vertical edge VE does not change. It is the figure which showed transition of count value CT and edge speed V (VE) when the horizontal direction coordinate of the vertical edge VE changes for every pixel. It is the figure which showed transition of count value CT and edge speed V (VE) when the horizontal direction coordinate of the vertical edge VE changes for every 2 pixels. It is the figure which showed transition of count value CT and edge velocity V (VE) when the horizontal direction coordinate of the vertical edge VE changes for every 3 pixels. It is the figure which showed transition of count value CT and edge speed V (VE) when the horizontal direction coordinate of the vertical edge VE changes for every 5 pixels from right to left. It is a flowchart which shows the flow of the process performed by the image process part 40 of 1st Embodiment. It is a figure which shows the function structural example of the image process part 40 which concerns on 2nd Embodiment. It is a figure which shows an example of the map used when the determination result selection part 48 selects the width | variety of the extended area AVE.

Hereinafter, embodiments of an object detection device, a driving support device, an object detection method, and an object detection program according to the present invention will be described with reference to the drawings.
<First Embodiment>
FIG. 1 is a diagram schematically illustrating an example of a configuration of a driving support device 1 including an object detection device 5 according to the first embodiment. The driving support device 1 is a device mounted on the vehicle M, for example, and includes a camera 10, a radar device 20, and a control device 30. Hereinafter, the vehicle on which the driving support device 1 is mounted is referred to as “own vehicle”.

  The camera 10 is attached to the upper part of a front windshield, the back surface of a room mirror, etc., for example. The camera 10 is a digital camera using a fixed imaging element such as a charge coupled device (CCD) or a complementary metal oxide semiconductor (CMOS), for example. For example, the camera 10 repeatedly captures an image of the front of the vehicle at a predetermined cycle and outputs data of the captured image to the control device 30.

  The radar device 20 is attached to, for example, the back side of the emblem of the vehicle M, the bumper, the periphery of the front grille, and the like. The radar apparatus 20 detects, for example, at least the position (distance and azimuth angle) of an object by emitting a millimeter wave in front of the vehicle M and receiving a reflected wave reflected by an object in front of the vehicle M. The radar apparatus 20 may be capable of detecting a relative speed with respect to an object. The radar apparatus 20 detects the position and speed of an object by, for example, FM-CW (Frequency-Modulated Continuous-Wave) method, and outputs the detection result to the control apparatus 30.

  The control device 30 includes, for example, a processor such as a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a HDD (Hard Disk Drive), an EEPROM (Electrically Erasable Programmable Read-Only Memory), a flash, and the like. A computer device in which a storage device such as a memory is connected via an internal bus.

  FIG. 2 is a diagram illustrating a functional configuration example centering on the control device 30 according to the first embodiment. The control device 30 includes, for example, an image processing unit 40, a sensor fusion processing unit 50, and a driving support unit 60. The image processing unit 40 includes, for example, a vertical edge extraction unit 42, an extended area setting unit 44, and a determination unit 46. For example, the camera 10, the radar device 20, and the control device 30 including the image processing unit 40 and the sensor fusion processing unit 50 correspond to the object detection device 5. Each functional unit of the control device 30 is, for example, a software functional unit that functions when the processor of the control device 30 executes a program. Note that some or all of these functional units may be hardware functional units such as LSI (Large Scale Integration) and ASIC (Application Specific Integrated Circuit). Further, the image processing unit 40, the sensor fusion processing unit 50, and the driving support unit 60 are not realized by the same computer device, but may be realized by separate computer devices. For example, the control device 30 may include a computer device that functions as the image processing unit 40 and the sensor fusion processing unit 50 and a separate computer device that functions as the driving support unit 60.

  Here, the function of the object detection device 5 will be outlined. FIG. 3 is an explanatory diagram for explaining the function of the object detection device 5. For example, the object detection device 5 first approaches an object OB1 such as a vehicle, a pedestrian, or a bicycle that approaches the host vehicle M from the side and has at least a part in the side area AS, and first analyzes an image captured by the camera 10. It is detected as a detection object candidate. Then, when the object OB1 that is a detection target candidate enters the center area AC, the object detection device 5 takes into account both the captured image analysis result of the camera 10 and the output of the radar device 20 to determine the object OB1. Processing such as detecting the position is performed (sensor fusion). Hereinafter, the detection process of the object OB1 in the side area AS will be mainly described.

  The vertical edge extraction unit 42 of the image processing unit 40 extracts a vertical edge extending in a substantially vertical direction in the captured image IM captured by the camera 10. The “vertical direction” is defined, for example, as a direction in which pixels are arranged in the vertical direction in the captured image IM (see FIG. 4). On the other hand, the “horizontal direction” is defined as a direction in which pixels are arranged in the horizontal direction in the captured image IM, for example. Further, the “vertical direction” may be defined as a direction having an inclination within about plus or minus 5 degrees with respect to a direction in which pixels are arranged in the vertical direction in the captured image IM. For example, in the captured image IM, the vertical edge extraction unit 42 is a feature line in which pixels having a luminance difference with a neighboring pixel in the horizontal direction are connected to a predetermined value or more, and the feature line extending in the substantially vertical direction is Extract as vertical edges. Not limited to this, the vertical edge extraction unit 42 may extract a vertical edge using a filter or the like for extracting a pixel having a particularly large luminance difference in the horizontal direction. FIG. 4 is a diagram illustrating how the vertical edges VE <b> 1 to VE <b> 3 are extracted from the captured image IM acquired by the camera 10. In the captured image IM, vertical edges are extracted from various objects such as the vending machine OB2, the utility pole OB3, and other guardrail pillars in addition to the object OB1 that is the detection target. The object detection device 5 performs a process of narrowing down to the vertical edge of the object OB1 that is a detection target candidate by the process described below.

  The extension region setting unit 44 uses the vertical edge VE extracted by the vertical edge extraction unit 42 as one end, and sets a predetermined width from the vertical edge VE to the center direction (as an example, toward the center axis C) in the horizontal direction of the captured image IM. 5 is a diagram showing a state in which the extended areas AVE1 to AVE3 are set in the captured image IM, even if the predetermined width is a fixed value such as 5 pixels, for example. Alternatively, it may be a variable value as in the second embodiment to be described later, as shown in the drawing, the extension region AVE is on the right side of the vertical edge VE in the portion on the left side of the center axis C of the captured image IM. And is set so as to have a width from the vertical edge VE toward the left side in the portion on the right side of the center axis C of the captured image IM. It is intended to include a vertical edge.

  The vertical edge extraction unit 42 and the extended region setting unit 44 repeat the above-described processing for all of the captured images IM repeatedly captured by the camera 10 (or for some captured images IM after the thinning process is performed). Run. Hereinafter, each of the plurality of captured images IM captured repeatedly in time series is referred to as an imaging frame as necessary.

  The determination unit 46 determines whether or not the vertical edge VE corresponds to a part of the detection target using the extension region AVE set by the extension region setting unit 44. Then, the vertical edge VE determined to correspond to a part of the detection target is selectively output to the sensor fusion processing unit 50.

  Hereinafter, a determination method by the determination unit 46 will be described. FIG. 6 is a diagram illustrating an extension area AVE that overlaps the extension area AVE of the immediately preceding imaging frame in a plurality of imaging frames. In the drawing, the extension area AVE (t) is an extension area in the imaging frame t, and the extension area AVE (t + 1) is an extension area in the imaging frame t + 1. The determination unit 46 counts a count value CT, which is an index value representing an overlap duration (or the number of overlap duration frames) between pixels, for pixels in the overlap region with the extended region AVE of the previous imaging frame. The total value (CT total) is calculated. Further, the determination unit 46 reflects the speed at which the vertical edge VE moves in the horizontal direction between temporally adjacent imaging frames (between temporally adjacent imaging frames after thinning if thinning processing is performed). An edge speed V (VE) that is an index value is calculated.

  7 to 11 exemplify changes in the horizontal coordinates of the vertical edge VE and the extended area AVE in the captured image IM, and are diagrams showing changes in the count value CT, the CT total, and the edge speed V (VE) for each. It is. Each drawing focuses on each pixel on a horizontal straight line (for example, LH in FIG. 6) having a certain vertical coordinate.

FIG. 7 is a diagram illustrating transition of the count value CT and the edge speed V (VE) when the horizontal coordinate of the vertical edge VE does not change. It is assumed that the vertical edge VE and the extension area AVE are set for the first time at the frame number t. In this case, the determination unit 46 sets the count value CT to “1” for the horizontal direction coordinates k to k + 4. Moreover, the determination part 46 calculates edge speed V (VE) based on Formula (1), for example. When the tracking of the extended area AVE is started for the first time, the edge velocity V (VE) becomes an invalid value. Instead of this method, the difference in the horizontal coordinate of the vertical edge VE may be simply obtained. Further, the determination unit 46 calculates the CT total as “5”.
V (VE) = | (width of extension area AVE) / (difference in count value CT at the left and right ends of extension area AVE) | (1)

  The determination unit 46 counts up the count value CT to “2” for the horizontal coordinates k to k + 4 at the next frame number t + 1, and calculates the edge velocity V (VE) as an invalid value (the left and right ends of the extension region AVE). Because the count value CT of both is “2” and the denominator is zero). Further, the determination unit 46 calculates the CT total as “10”. The determination unit 46 counts up the count value CT to “3” for the horizontal coordinates k to k + 4 at the next frame number t + 3, and calculates the edge velocity V (VE) as an invalid value (same as above). Further, the determination unit 46 calculates the CT total as “15”. The determination unit 46 counts up the count value CT to “4” for the horizontal coordinates k to k + 4 at the next frame number t + 3, and calculates the edge velocity V (VE) as an invalid value (same as above). In addition, the determination unit 46 calculates the CT total as “20”.

  The determination unit 46 calculates the count value CT and the sum thereof, and the edge velocity V (VE) in this way. (1) When the CT sum becomes equal to or greater than a first set value (for example, 10), or (2) When the edge speed V (VE) is equal to or lower than a second set value (for example, 2.5), the vertical edge VE that is the origin of the extended area AVE is selected as a part of the detection target candidates. It is determined that it is present, and is output to the sensor fusion processing unit 50. An object that satisfies these conditions (1) and (2) is an object whose horizontal coordinates in the captured image IM do not change much. By setting the conditions in this way, the left and right ends of the captured image IM are like objects (vehicles, etc.) that pass in front of the host vehicle M at a relatively high speed and fixed objects (vending machines, etc.) at the road ends. A flowing object can be appropriately excluded with one logic. These excluded objects have a low risk of colliding with the host vehicle, and have a low necessity for performing driving assistance described later. Therefore, unnecessary driving assistance can be suppressed by the processing of the determination unit 46.

  In the example of FIG. 7, since the CT total is 10 or more at frame number t + 1, it is determined that the vertical edge VE is part of the detection target candidate at the time of frame number t + 1.

  When the extension area AVE no longer overlaps with the extension area AVE of the previous imaging frame, the determination unit 46 clears the count value CT, the CT total, and the edge speed V (VE), and creates a new extension area. As AVE, calculation of the count value CT, the CT total, and the edge speed V (VE) is started.

  FIG. 8 is a diagram showing transition of the count value CT and the edge speed V (VE) when the horizontal coordinate of the vertical edge VE changes by one pixel. In the example of FIG. 8, since the CT total is 10 or more and the edge speed V (VE) is 2.5 or less at the frame number t + 2, the vertical edge VE is detected at the time of the frame number t + 2. It is determined that it is a part of the candidate object.

  FIG. 9 is a diagram illustrating changes in the count value CT and the edge speed V (VE) when the horizontal coordinate of the vertical edge VE changes by two pixels. In addition, illustration of a horizontal direction coordinate is abbreviate | omitted after this figure. In the example of FIG. 9, since the edge speed V (VE) is 2.5 or less at the frame number t + 2, it is determined that the vertical edge VE is a part of the detection target candidate at the time of the frame number t + 2. Is done.

  FIG. 10 is a diagram illustrating transition of the count value CT and the edge speed V (VE) when the horizontal coordinate of the vertical edge VE changes by 3 pixels. For example, it is assumed that an object (such as a vehicle) that passes in front of the host vehicle M at a relatively high speed moves the vertical edge VE with such a tendency. In the example of FIG. 10, since neither of the conditions (1) and (2) is satisfied, the vertical edge VE is excluded.

  FIG. 11 is a diagram showing the transition of the count value CT and the edge speed V (VE) when the horizontal coordinate of the vertical edge VE changes from right to left by five pixels. The fixed edge existing on the left side of the road often moves the vertical edge VE with such a tendency. In the example of FIG. 11, since neither of the conditions (1) and (2) is satisfied, the vertical edge VE is excluded.

  7 to 11 exemplify that the vertical edge VE moves on the captured image IM at a constant speed, but the method of the present embodiment does not lose its effectiveness even when the speed of the vertical edge VE changes. .

  FIG. 12 is a flowchart illustrating a flow of processing executed by the image processing unit 40 according to the first embodiment. The processing of this flowchart is repeatedly executed at a predetermined cycle, for example. First, the image processing unit 40 acquires an imaging frame from the camera 10 (step S100). Next, the vertical edge extraction unit 42 extracts the vertical edge VE from the imaging frame (step S102). If the vertical edge VE is not extracted, the process returns to step S100.

  When the vertical edge VE is extracted, the extension area setting unit 44 sets the extension edge AVE (step S106). Then, the determination unit 46 calculates the count value CT, the CT total, and the edge speed V (VE) (step S108).

  Next, the determination unit 46 determines whether or not the CT total is greater than or equal to the first set value (step S110). When the CT total is less than the first set value, the determination unit 46 determines whether or not the edge speed V (VE) is equal to or lower than the second set value (step S112). When the edge speed V (VE) is equal to or lower than the second set value, the determination unit 46 determines whether or not the edge speed V (VE) is an invalid value (step S114).

  When the CT total is equal to or higher than the first set value, or when the edge speed V (VE) is equal to or lower than the second set value and is not an invalid value, the determination unit 46 sets the vertical edge VE as a detection target candidate. (Step S116), the count value CT is cleared, and one routine of this flowchart is terminated. Otherwise, the determination unit 46 determines that the vertical edge VE is not a detection target candidate (step S120), and ends one routine of this flowchart.

  When the vertical edge VE determined as the detection target candidate enters the center area AC in FIG. 3, the sensor fusion processing unit 50 takes into account both the captured image analysis result of the camera 10 and the output of the radar apparatus 20. Processing such as detecting the position of the object is performed. The sensor fusion processing unit 50 may perform clustering processing based on both the captured image analysis result of the camera 10 and the output of the radar apparatus 20 to perform object identification processing based on the target.

  Then, the driving support unit 60 controls the speaker 70, the electronically controlled brake device 80, the power steering device 90, and the like based on the position of the object detected by the sensor fusion processing unit 50. Specifically, the driving support unit 60 instructs the electronically controlled brake device 80 and the power steering device 90 to output a sound representing a warning or warning to the speaker 70 or to perform automatic brake control or automatic steering control. To do.

  According to the object detection device 5 of the first embodiment described above, the vertical edge VE extending in the vertical direction is extracted from the captured image captured by the camera 10 that captures the periphery of the vehicle, and the vertical edge is one end. By setting an extension area AVE having a predetermined width from the vertical edge toward the center of the captured image, and determining whether the vertical edge VE corresponds to a part of the detection object using the extension area AVE The detection object can be detected more appropriately.

  Further, according to the object detection device 5 of the first embodiment, the vertical edge VE is detected based on the CT total representing the degree of stagnation on the captured image of the object on which the vertical edge appears and the edge speed V (VE). Therefore, it is possible to detect the detection target more appropriately.

  Moreover, according to the driving assistance apparatus 1 using the object detection apparatus 5 of the first embodiment, driving assistance can be appropriately performed based on the detection target appropriately detected by the object detection apparatus 5.

Second Embodiment
Hereinafter, a second embodiment of the present invention will be described. The object detection device 5 of the second embodiment differs from the first embodiment in the processing content of the image processing unit 40. Hereinafter, the difference will be mainly described.

  FIG. 13 is a diagram illustrating a functional configuration example of the image processing unit 40 according to the second embodiment. The image processing unit 40 according to the second embodiment includes a vertical edge extraction unit 42, extended region setting units 44-1 to 44-n (n is a natural number of 2 or more), determination units 46-1 to 46-n, The determination result selection unit 48 is provided. Since the function of the vertical edge extraction unit 42 is the same as that of the first embodiment, description thereof is omitted.

  The image processing unit 40 according to the second embodiment prepares a plurality of widths of the extension region AVE by the extension region setting units 44-1 to 44-n, and expands the width of the extension region AVE according to the relationship between the host vehicle and the object. Whether to adopt the area AVE is dynamically changed. The extension area setting unit 44-1 sets the extension area AVE with the smallest width (for example, two pixels), and the extension area setting unit 44-n sets the extension area AVE with the largest width (for example, n + 1 pixels). Set. Then, the determination units 46-1 to 46-n perform the same processing as that of the first embodiment on the extension areas AVE set by the corresponding extension area setting units 44-1 to 44-n, respectively.

  Here, increasing the width of the extended area AVE improves the followability to the object on the captured image, but has a property of increasing the processing load and increasing the false detection. For this reason, it is preferable to dynamically change the width of the extension area AVE according to the behavior of the object in the captured image.

  More specifically, when the distance between the host vehicle M and the object is long, or when the object is present in the central portion in the horizontal direction of the captured image, the lateral speed (≈edge speed) in the captured image of the object. V (VE)) tends to be slow. In this case, by reducing the width of the extension area AVE, the processing load can be reduced and erroneous detection can be suppressed. On the other hand, when the distance between the host vehicle M and the object is short, or when the object is present at the end in the horizontal direction of the captured image, the lateral speed of the captured image of the object tends to increase. . In this case, the followability to the object can be improved by increasing the width of the extension area AVE.

  Based on the determination results made by the determination units 46-1 to 46-n based on the distance between the host vehicle and the object and the horizontal coordinate of the vertical edge VE in the captured image, for example, The determination result is selected (that is, the width of the extension area AVE used for processing is selected). The determination result selection unit 48 selects the width of the extension area AVE using, for example, the map shown in FIG. FIG. 14 is a diagram illustrating an example of a map used when the determination result selection unit 48 selects the width of the extension area AVE. The distance between the host vehicle and the object may be derived from the position on the captured image of the vertical edge VE, or may be grasped using the output of the radar device 20.

  Here, it is conceivable that the determination unit 46 performs the processing after selecting the width of the extension area first, instead of performing the same calculation as that of the first embodiment for each of the extension areas AVE having a plurality of widths. However, in this case, since the processing of the certain imaging frame is processed with k as the extension area width and then the processing of the next imaging frame is performed with the width of the expansion area as k + 1, the arithmetic processing is switched. There is a possibility that the judgment result may be blurred due to the switching of the. On the other hand, as in this embodiment, the same calculation as in the first embodiment is performed for each of a plurality of width extension areas AVE, and only the determination result is switched, so that the determination caused by the switching of the calculation processing is performed. It is possible to prevent blurring of the result. Note that the determination unit 46 may perform the processing after selecting the width of the extended region first as long as the processing mode does not cause such a blurring of the determination result.

  According to the object detection device 5 of the second embodiment described above, in addition to the effects of the first embodiment, by dynamically changing the width of the extension area AVE, the processing load can be reduced and erroneous detection can be suppressed. It is possible to achieve both improvement in the followability to the object.

  Further, according to the object detection device 5 of the second embodiment, processing is performed in parallel for the widths of the plurality of extension areas AVE, and one determination result is selected from the determination results for the widths of the plurality of extension areas AVE. Thus, it is possible to prevent the determination result from being blurred due to the switching of the arithmetic processing.

  As mentioned above, although the form for implementing this invention was demonstrated using embodiment, this invention is not limited to such embodiment at all, In the range which does not deviate from the summary of this invention, various deformation | transformation and substitution Can be added.

  For example, the object detection device 5 may not include the radar device 20. In this case, when the object OB1 that is a detection target candidate enters the center area AC, the position of the object OB1 may be finally detected by the captured image analysis of the camera 10. In this case, a functional unit such as a “second image processing unit” may be provided instead of the “sensor fusion processing unit 50”. There is no particular limitation on the function of the second image processing unit, and the second image processing unit detects the position of the object OB1 by performing, for example, pattern matching processing or the like.

  When the radar device 20 is not provided, the object detection device 5 has a position on the image of the vertical edge VE determined by the determination unit 46 as being a part of the detection target (particularly on the image at the lower end of the vertical edge VE). The position of the object OB1 may be directly detected from

DESCRIPTION OF SYMBOLS 1 Driving assistance apparatus 5 Object detection apparatus 10 Camera (imaging part)
DESCRIPTION OF SYMBOLS 20 Radar apparatus 30 Control apparatus 40 Image processing part 42 Vertical edge extraction part 44 Expansion area setting part 46 Determination part 50 Sensor fusion processing part 60 Driving support part M Vehicle

Claims (13)

A vertical edge extraction unit that extracts a vertical edge extending in a vertical direction in a captured image captured by an imaging unit that captures the periphery of the vehicle;
An extension region setting unit that sets the extension region having a predetermined width from the vertical edge toward the center of the captured image, with the vertical edge extracted by the vertical edge extraction unit as one end;
A determination unit that determines whether or not the vertical edge corresponds to a part of the detection object using the extension region set by the extension region setting unit;
An object detection apparatus comprising: The determination unit determines whether or not the vertical edge corresponds to a part of the detection target based on the overlapping state of the extension regions in a plurality of captured images captured in time series.
The object detection apparatus according to claim 1. The determination unit determines that the vertical edge corresponds to a part of the detection target when a plurality of captured images captured in time series includes a long overlap duration of pixels included in the extension region and overlapping each other. To
The object detection apparatus according to claim 2. The determination unit determines whether or not the vertical edge corresponds to a part of the detection target based on a horizontal speed of the captured image of the vertical edge.
The object detection device according to claim 1. The determination unit determines that the vertical edge corresponds to a part of the detection target when a horizontal speed in the captured image of the vertical edge is small.
The object detection apparatus according to claim 4. The determination unit calculates overlap durations of pixels included in the extension region and overlapping each other in a plurality of captured images taken in time series, and the overlap durations of pixels at the left and right ends of the extension region are calculated. Calculating an index value having a difference as a denominator, and treating the calculated index value as a horizontal speed in the captured image of the vertical edge;
The object detection apparatus according to claim 4 or 5. The determination unit determines that the vertical edge corresponds to a part of the detection target even when the horizontal value of the vertical edge in the captured image is small when the index value becomes an invalid value. do not do,
The object detection apparatus according to claim 6. When the distance between the vehicle and the detection object is long, the width of the expansion area is set small, and when the distance between the vehicle and the detection object is short, the width of the expansion area is set large.
The object detection device according to claim 1. When the horizontal coordinate in the captured image of the vertical edge is close to the center portion, the width of the extension region is set small, and when the horizontal coordinate in the captured image of the vertical edge is close to the left and right end portions, the extension region is set. Set the width of
The object detection device according to claim 1. The extension area setting unit and the determination unit perform processing in parallel for the width of a plurality of extension areas,
A determination result selection unit for selecting one determination result from the determination results for the widths of the plurality of extension regions by the determination unit;
The object detection apparatus according to claim 8 or 9. The object detection device according to any one of claims 1 to 10,
A driving support unit that performs driving support based on the detection result of the object detection device;
A driving support apparatus comprising: In the captured image captured by the imaging unit that captures the periphery of the vehicle, a vertical edge extending in the vertical direction is extracted,
With the extracted vertical edge as one end, an extended region having a predetermined width from the vertical edge toward the center of the captured image is set,
Determining whether the vertical edge corresponds to a part of the detection object using the set extension region;
Object detection method. On the computer,
In the captured image captured by the image capturing unit that captures the periphery of the vehicle, the vertical edge extending in the vertical direction is extracted,
With the extracted vertical edge as one end, an extension region having a predetermined width is set from the vertical edge toward the center of the captured image,
Using the set extension region, it is determined whether the vertical edge corresponds to a part of the detection object,
Object detection program.

Images