JP2017084019A - Object detection device, object detection method, and, object detection program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an object detection device capable of further improving safety by detecting an object that travels at a speed substantially equal to that of a vehicle, an object detection method and an object detection program.SOLUTION: An object detection device 1 comprises: moving position search means 22A for calculating an optical flow regarding a reference point of a reference image; ideal position calculation means 22B for calculating an ideal vector in which the reference point of the reference image moves towards a disappearance point; angle difference calculation means 22C for calculating an angle difference Δθ between an angle of the ideal vector and an angle of the optical flow; moving speed calculation means 22D for calculating a moving speed (k) from the reference point to an actual moving point; frequency distribution means 22E for calculating frequency distribution regarding the angle difference Δθ and the moving speed (k); and detection means 22F for detecting an object that travels in parallel with the vehicle or approaches the vehicle so as to overtake the vehicle, on the basis of the frequency distribution.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an object detection device, an object detection method, and an object detection program for detecting an object mounted on a vehicle and present in the vicinity of the host vehicle.

  In recent years, in order to improve safety when driving a vehicle, an object such as a vehicle or a person existing around the own vehicle is photographed by a camera, and the object is detected from the photographed image to notify the driver. Has been done. For example, Patent Document 1 tries to change the lane to a lane in which a parallel vehicle is traveling when there is a parallel vehicle that travels in a blind spot from a rearview mirror or side mirror. However, there is described a lane change support device that warns the user not to change the lane. Specifically, for example, an optical flow is calculated for an image obtained by shooting a lane that runs in parallel with the host vehicle, and a histogram is created by voting on the amount of movement between two frames used for the calculation. At this time, when there is a parallel running vehicle, a sharp peak appears in the histogram, so that the presence or absence of the parallel running vehicle can be detected with high accuracy.

JP 2007-182201 A

  However, the method described in Patent Document 1 can be detected when there is a relative speed between the host vehicle and the parallel running vehicle, and the host vehicle and the parallel running vehicle run in parallel with the relative speed ≈0. In such a case, there is a problem that a peak does not appear on the histogram and cannot be detected. The parallel running vehicle has no problem when it is running parallel to the host vehicle without a speed difference, but changes to the detection target at the moment when the speed of the host vehicle changes due to deceleration or the like. For example, when the host vehicle decelerates due to a left turn, the parallel running vehicle has a relative speed at that moment. Therefore, at the moment when the host vehicle decelerates, the algorithm can calculate the optical flow for the parallel running vehicle, but if the warning is issued after the deceleration, the response will not be in time, and the parallel running vehicle will be wound. There is a risk of getting injured.

  The present invention has been made based on such a problem, and can detect an object overtaking the own vehicle or an object traveling at the same speed as the own vehicle, thereby further improving safety. An object of the present invention is to provide an object detection device, an object detection method, and an object detection program.

  An object detection device of the present invention detects an object around the host vehicle, and includes an imaging unit that captures an image around the host vehicle, and an image that is captured in time series by the imaging unit. For a reference point of an arbitrary reference image, when a moving position search means for searching for an actual actual moving point in a detected image after the reference image and a reference point in the reference image are a stopped object The ideal position calculation means for calculating the ideal ideal movement point in the detected image of the reference point, and the angle difference between the movement direction from the reference point to the ideal movement point and the movement direction from the reference point to the actual movement point The angle difference calculating means, the moving speed calculating means for calculating the moving speed from the reference point to the actual moving point, the angle difference obtained by the angle difference calculating means, and the moving speed obtained by the moving speed calculating means. Find distribution That a frequency distribution means, based on the frequency distribution obtained by frequency distribution unit, in which a detection means for detecting an object approaching to or overtake running parallel with respect to the vehicle.

  An object detection method of the present invention detects an object around the host vehicle, and includes an imaging step for capturing an image around the host vehicle, and an image captured in time series by the imaging step. For a reference point of an arbitrary reference image, a moving position search step for finding an actual actual moving point in a detection image after the reference image, and a case where the reference point in the reference image is a stopped object The ideal position calculation step for calculating the ideal ideal moving point in the detected image of the reference point, and the angular difference between the moving direction from the reference point to the ideal moving point and the moving direction from the reference point to the actual moving point An angle difference calculating step, a moving speed calculating step for calculating a moving speed from the reference point to the actual moving point, an angle difference obtained by the angle calculating step, and a moving speed calculating step A frequency distribution step for obtaining a frequency distribution and a detection step for detecting an object that approaches the vehicle in parallel or overtakes the vehicle based on the frequency distribution obtained by the frequency distribution step. .

  The object detection program of the present invention detects an object around the host vehicle, and detects a reference point of an arbitrary reference image among images taken in time series after the reference image. The ideal ideal moving point in the detected image of the reference point is calculated when the moving position search means for searching for the actual actual moving point in the image and the reference point in the reference image are the stopped objects. Ideal position calculation means, angle difference calculation means for calculating the angle difference between the movement direction from the reference point to the ideal movement point and the movement direction from the reference point to the actual movement point, and movement from the reference point to the actual movement point A frequency distribution means for obtaining a frequency distribution, and a frequency distribution obtained by the frequency distribution means for the moving speed calculation means for calculating the speed, the angular difference obtained by the angle difference calculation means, and the moving speed obtained by the movement speed calculation means. Based on It is intended to function as detection means for detecting an object approaching to overtake, or running parallel with respect to the vehicle, as a computer.

  According to the present invention, the angle difference between the moving direction from the reference point to the ideal moving point and the moving direction from the reference point to the actual moving point is calculated, and the moving speed from the reference point to the actual moving point is calculated. Since the frequency distribution is obtained for them, for example, when there is an object that runs parallel to the own vehicle without a speed difference, the absolute value of the angle difference is small, and a large frequency appears in a class with a low moving speed, When there is an object approaching to overtake the host vehicle, a large frequency appears in a class having a large absolute value of the angle difference. Therefore, based on the frequency distribution, it is possible to detect an object that moves parallel to or overtakes the host vehicle. Accordingly, it is possible to find an object that runs parallel to the host vehicle at substantially the same speed while entering the blind spot, and to further improve safety. In addition, the algorithm can improve the robustness of the algorithm by detecting an object having a relative speed and detecting an object running in parallel in a place that may not be in time when warned.

It is a block diagram showing schematic structure of the target object detection apparatus which concerns on one embodiment of this invention. It is a figure explaining the optical flow in the detection image of the reference point of a reference image, and a reference point. It is a figure explaining the reference vector of a reference image, and the ideal vector of a reference point. It is a characteristic view showing the three-dimensional histogram regarding an angle difference and a moving speed. It is another characteristic view showing the three-dimensional histogram regarding the angle difference and the moving speed. It is a flowchart showing the step which detects the target object of the own vehicle by the target object detection apparatus shown in FIG.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 shows a schematic configuration of an object detection apparatus 1 according to an embodiment of the present invention. The object detection device 1 detects an object around the host vehicle. For example, the object detection device 1 detects an object approaching the host vehicle so as to run parallel or overtake the vehicle and turn the vehicle or turn the lane. This is to prevent collision with the object at the time of change.

  The object detection device 1 is used by being mounted on a host vehicle, for example, and includes an imaging unit 10 that captures an image around the host vehicle, a processing unit 20 including a memory 21, a CPU 22, and the like, and a warning unit 30 that issues a warning. It has. The imaging means 10 is constituted by a camera such as a CCD camera, a CMOS camera, or an ITV camera, for example. In the present embodiment, for example, the imaging unit 10 is attached to the side mirror of the host vehicle with the field of view facing rearward. Note that the imaging means 10 may be attached to any position as long as it can capture the side of the host vehicle or the rear from the roof. The imaging unit 10 is connected to the processing unit 20 via, for example, the A / D converter 11, and data of a plurality of frames of images captured in time series by the imaging unit 10 is continuously in time series. The digital signal is converted by the A / D converter 11 and sent to the processing unit 20.

  The memory 21 of the processing unit 20 stores, for example, an object detection program 21A. For example, the CPU 22 receives a command from the object detection program 21A stored in the memory 21, and receives the movement position search means 22A, the ideal position calculation means 22B, the angle difference calculation means 22C, the movement speed calculation means 22D, the frequency It functions as the distribution means 22E and the detection means 22F, and when it receives the data of the image picked up by the image pickup means 10, it performs image processing according to the instruction of the object detection program 21A and detects the object. Is set to

  For example, in the image captured in time series by the imaging unit 10, the movement position searching unit 22 </ b> A is an actual movement in which a reference point in any reference image has actually moved in a detection image after the reference image. Search for points. Specifically, for example, an optical flow indicating where the reference point of the reference image has moved in the detected image is calculated. The detected image is preferably an image of the next frame of the reference image. The moving position search means 22A calculates an optical flow for the reference point using one image captured by the image pickup means 10 as a reference image, and then uses the next frame image as the reference image and the optical for the reference point. It is preferable to calculate the optical flow of the reference point while sequentially changing the reference image to the image of the next frame so as to calculate the flow.

  For example, when the reference point in the reference image is a stopped object, the ideal position calculation unit 22B calculates an ideal ideal movement point in the detected image of the reference point. Specifically, for example, an ideal ideal vector in which the reference point of the reference image is directed to the vanishing point is calculated. This ideal vector can be calculated from the vanishing point, and the vanishing point can be estimated from the optical flow, or can be calculated from initial information such as the angle of view, the position of the imaging unit 10, and the moving speed of the imaging unit 10. it can. 2 and 3 show examples of reference images. FIG. 2 shows a case where there is a parallel object, and shows an optical flow in a reference point and a detected image of the reference point. FIG. 3 shows the reference point and the ideal vector of the reference point.

  The angle difference calculation means 22C is, for example, an angle between the movement direction from the reference point to the ideal movement point obtained by the ideal position calculation means 22B and the movement direction from the reference point to the actual movement point obtained by the movement position search means 22A. The difference Δθ is calculated. Specifically, for example, the angle difference Δθ between the angle of the ideal vector obtained by the ideal position calculating means 22B and the angle of the optical flow obtained by the moving position searching means 22A is calculated. For example, as shown in FIG. 2, when the target object is present in the reference image and the target object is running in parallel in the same direction as the host vehicle, the optical flow is substantially the same as the ideal vector. Since the direction is small or not at all, the angle difference Δθ between the angle of the ideal vector and the angle of the optical flow becomes zero or close to zero. When the object approaches to overtake the host vehicle, the optical flow is in the opposite direction to the ideal vector, so the angle difference Δθ between the ideal vector angle and the optical flow angle is close to 180 degrees in absolute value.

  The moving speed calculating means 22D calculates, for example, the moving speed k from the reference point to the actual moving point obtained by the moving position searching means 22A, that is, the relative speed of the reference point with respect to the host vehicle. Since the movement speed k of the reference point to the actual movement point is high, the movement amount of the reference point to the actual movement point is large, and therefore the optical flow of the reference point becomes long. Therefore, the moving speed k from the reference point to the actual moving point may be obtained, for example, by calculating in advance an ideal vector of speed and comparing it with the ideal vector. For example, as shown in FIG. 2, when the target object is present in the reference image and the target object is running in parallel in the same direction as the host vehicle at substantially the same speed, the actual moving point of the reference point The moving speed k to becomes smaller.

  The frequency distribution means 22E obtains the frequency distribution of the reference points with respect to, for example, the angle difference Δθ obtained by the angle difference calculation means 22C and the movement speed k obtained by the movement speed calculation means 22D. Specifically, for example, it is preferable to create a three-dimensional histogram for the angle difference Δθ obtained by the angle difference calculating unit 22C and the moving speed k obtained by the moving speed calculating unit 22D.

  4 and 5 show an example of the created histogram. FIG. 4 shows a case where there is an object that runs in parallel in the same direction as the host vehicle at almost the same speed, and FIG. 5 shows a case where there is an object approaching to pass the host vehicle. As shown in FIG. 4, when there is an object that runs in parallel in the same direction as the host vehicle, for example, the angle difference Δθ is smaller than 150 degrees in absolute value, and the moving speed k is A class having a frequency equal to or higher than a predetermined frequency appears in a range smaller than the speed of the road surface. In addition, as shown in FIG. 5, when there is an object approaching to overtake the host vehicle, for example, a class having a predetermined frequency or more in a range where the angle difference Δθ is 150 degrees or more in absolute value. appear. In FIG. 4, the road surface has a large frequency difference Δθ of around 0 degrees and a moving speed k of 25 km / h to 50 km / h. In FIG. 5, the angle difference Δθ is around 0 degrees. It is on the road surface that a large frequency appears from 20 km / h to 40 km / h.

  The detection unit 22F detects, for example, an object that approaches or overtakes the host vehicle based on the frequency distribution obtained by the frequency distribution unit 22E. For example, when there is a class having a frequency greater than or equal to a predetermined level in the frequency distribution obtained by the frequency distribution means 22E, depending on the class, there is an object approaching to run parallel to or overtake the host vehicle. It is preferable to detect.

  Specifically, for example, in the frequency distribution, the angle difference Δθ is an absolute value smaller than a predetermined angle difference setting value, the moving speed k is smaller than the road surface speed, and the difference between the moving speed k and the road surface speed is When there is a class having a frequency greater than or equal to a predetermined frequency difference within a predetermined speed difference set value, it is preferable to determine that there is an object that runs parallel to the host vehicle. Further, for example, in the frequency distribution, the angle difference Δθ is an absolute value that is equal to or larger than a predetermined angle difference setting value, and the difference between the moving speed k and the road surface speed is equal to or larger than a predetermined speed difference setting value. When there is a class having a frequency, it is preferable to determine that there is an object approaching to overtake the own vehicle.

  Here, it is preferable to set the angle difference setting value within a range of 130 degrees to 170 degrees, for example. The angle difference Δθ is close to zero when the object is running in parallel in the same direction as the host vehicle, and the angle difference Δθ is an absolute value when the object approaches to pass the host vehicle. This is because a good determination can be made if the angle difference set value is set and determined within the above-described range. The speed difference set value is preferably set within a range of 5 km / h to 20 km / h, for example. This is because when the difference between the moving speed k and the road surface speed is small, the speed of the host vehicle is low and the danger is low. It is assumed that there is an object that runs in parallel when the moving speed k is lower than the road speed, but it is more preferable if the moving speed is lower than 2/3 of the road speed.

  The detection means 22F also instructs the warning means 30 to issue a warning when, for example, an object approaching to run parallel to or overtakes the host vehicle is detected. For example, when an instruction is input from the detection unit 22F, the warning unit 30 displays a message such as “There is a parallel object” or “There is an overtaking object”, and the parallel object or overtaking object is displayed to the driver. It is preferable to notify the presence of an object.

  This object detection apparatus 1 is used as follows. First, as an advance preparation, the object detection device 1 is mounted on the own vehicle, and the position is adjusted so that the rear side is imaged by the imaging means 10 from the side. For example, as shown in FIGS. 2 and 3, in the image captured by the imaging unit 10, it is preferable that the boundary between the sky and the road surface be positioned near the center in the vertical direction.

  Next, an object around the host vehicle, specifically, an object that approaches the host vehicle so as to run side by side or overtake is detected as follows. FIG. 6 shows a step of detecting an object around the host vehicle by the object detection device 1. First, when the reference point in the reference image is a stopped object based on information known in advance by the ideal position calculation unit 22B of the processing unit 20, the position of the imaging unit 10, setting parameters, and the like, An ideal ideal movement point in the detected image of the reference point is calculated. Specifically, for example, as shown in FIG. 3, an ideal ideal vector in which the reference point of the reference image is directed to the vanishing point is calculated (step 101).

  Next, an image around the host vehicle is picked up by the image pickup means 10, the image data is converted into a digital signal by the A / D converter 11, and sent to the processing unit 20 (step S102). Subsequently, in the image captured in time series by the imaging unit 10 by the moving position searching unit 22A of the processing unit 20, a reference point in any of the reference images is a detected image (for example, after the reference image) In the image of the next frame of the reference image), the actual moving point that has actually moved is searched. Specifically, for example, as shown in FIG. 2, an optical flow indicating where the reference point of the reference image has moved in the detected image is calculated (step S103).

  Next, the angle difference calculation means 22C of the processing unit 20 moves from the reference point to the ideal movement point obtained by the ideal position calculation means 22B, and from the reference point to the actual movement point obtained by the movement position search means 22A. An angle difference Δθ with respect to the moving direction is calculated. Specifically, for example, the angle difference Δθ between the angle of the ideal vector obtained by the ideal position calculating means 22B and the angle of the optical flow obtained by the moving position searching means 22A is calculated (step S104). Further, the moving speed calculating means 22D of the processing unit 20 calculates the moving speed k from the reference point to the actual moving point obtained by the moving position searching means 22A, that is, the relative speed of the reference point with respect to the host vehicle (step S105).

  After that, the frequency distribution means 22E of the processing unit 20 obtains the frequency distribution of the reference points for the angle difference Δθ obtained by the angle difference calculation means 22C and the movement speed k obtained by the movement speed calculation means 22D. Specifically, for example, a three-dimensional histogram is created for the angle difference Δθ obtained by the angle difference calculating means 22C and the moving speed k obtained by the moving speed calculating means 22D (step S106). As a result, when there is an object that runs in parallel in the same direction as the host vehicle at substantially the same speed, for example, as shown in FIG. 4, the angle difference Δθ is an absolute value from a predetermined angle difference setting value. In the range where the moving speed k is smaller than the speed of the road surface, a class having a predetermined frequency or more appears. In addition, when there is an object approaching to overtake the host vehicle, for example, as shown in FIG. 5, the angle difference Δθ is an absolute value within a range equal to or larger than a predetermined angle difference setting value. A class with a appears.

  Next, the detection unit 22F of the processing unit 20 detects an object that is parallel to or overtakes the host vehicle based on the frequency distribution obtained by the frequency distribution unit 22E (step S107). Specifically, for example, in the frequency distribution, the angle difference Δθ is an absolute value smaller than a predetermined angle difference setting value of 130 degrees or more and 170 degrees or less, the moving speed k is smaller than the road surface speed, and the moving speed k is An object that runs parallel to the host vehicle when there is a class having a frequency greater than or equal to a predetermined speed difference within a range that is greater than or equal to a predetermined speed difference set value of 5 km / h or more and 20 km / h or less. Is determined to exist. Further, for example, in the frequency distribution, the angle difference Δθ is an absolute value greater than or equal to a predetermined angle difference setting value of 130 degrees or more and 170 degrees or less, and the difference between the moving speed k and the road surface speed is 5 km / h or more and 20 km / h. If there is a class having a frequency greater than or equal to a predetermined frequency difference within the following predetermined speed difference set value, it is determined that there is an object approaching to overtake the host vehicle.

  Subsequently, when it is determined by the detection means 22F that there is an object approaching in parallel to or overtaking the host vehicle (step S108; Y), the warning means 30 is instructed to issue a warning (step S109). When the instruction is input from the detection unit 22F, the warning unit 30 displays, for example, a warning and notifies the driver of the presence of the parallel running object or the overtaking object (step S110). Accordingly, for example, it is possible to notify whether or not there is an object approaching to run parallel to or overtake the host vehicle in the detection range up to about 20 m behind the host vehicle.

  As described above, according to the present embodiment, the angle difference Δθ between the movement direction from the reference point to the ideal movement point and the movement direction from the reference point to the actual movement point is calculated, and from the reference point to the actual movement point. For example, when there is an object that runs parallel to the host vehicle without a speed difference, the absolute value of the angle difference Δθ is small and the movement speed k is calculated. When a large frequency appears in a class with a small and there is an object approaching to overtake the host vehicle, a large frequency appears in a class with a large absolute value of the angle difference Δθ. Therefore, based on the frequency distribution, it is possible to detect an object that moves parallel to or overtakes the host vehicle. Accordingly, it is possible to find an object that runs parallel to the host vehicle at substantially the same speed while entering the blind spot, and to further improve safety. In addition, the algorithm can improve the robustness of the algorithm by detecting an object having a relative speed and detecting an object running in parallel in a place that may not be in time when warned.

  The present invention has been described with reference to the embodiment. However, the present invention is not limited to the above embodiment, and various modifications can be made. For example, in the above-described embodiment, each component has been specifically described. However, not all the components may be provided, and other components may be provided.

  DESCRIPTION OF SYMBOLS 1 ... Object detection apparatus, 10 ... Imaging means, 11 ... A / D converter, 20 ... Processing part, 21 ... Memory, 21A ... Object detection program, 22 ... CPU, 22A ... Moving position search means, 22B ... Ideal Position calculation means, 22C ... angular difference calculation means, 22D ... movement speed calculation means, 22E ... frequency distribution means, 22F ... detection means, 30 ... warning means

Claims (6)

An object detection device for detecting an object around the host vehicle,
Imaging means for capturing an image around the vehicle;
In an image captured in time series by the imaging means, for a reference point of any reference image, a moving position search means for searching for an actual actual moving point in a detection image after the reference image,
When the reference point in the reference image is a stationary object, ideal position calculation means for calculating an ideal ideal movement point in the detected image of the reference hand;
An angle difference calculating means for calculating an angle difference between a moving direction from the reference point to the ideal moving point and a moving direction from the reference point to the actual moving point;
A moving speed calculating means for calculating a moving speed from the reference point to the actual moving point;
Frequency distribution means for obtaining a frequency distribution for the angle difference obtained by the angle difference calculation means and the movement speed obtained by the movement speed calculation means;
An object detection apparatus comprising: a detection means for detecting an object that approaches or overtakes the vehicle in parallel based on the frequency distribution obtained by the frequency distribution means.   In the case where the frequency distribution obtained by the frequency distribution means includes a class having a frequency equal to or higher than a predetermined level, there is an object that approaches the vehicle in parallel or overtakes the vehicle according to the class. The object detection device according to claim 1, wherein the detection is performed.   In the frequency distribution obtained by the frequency distribution means, the detection means is an absolute value that is smaller than a predetermined angle difference set value in absolute value, the moving speed is lower than a road speed, and the moving speed and the road surface When there is a class having a frequency greater than a predetermined level within a range where the difference from the speed is equal to or greater than a predetermined speed difference set value, it is detected that there is an object running parallel to the host vehicle. The target object detection apparatus of Claim 1 or Claim 2.   In the frequency distribution obtained by the frequency distribution means, the detection means has an absolute value equal to or larger than a predetermined angle difference setting value in absolute value, and a difference between the moving speed and the road surface speed is a predetermined speed difference setting value. 4. The method according to claim 1, further comprising: detecting an object approaching to overtake the host vehicle when there is a class having a frequency equal to or greater than a predetermined value in the range of 2. The object detection apparatus described in 1. An object detection method for detecting an object around a host vehicle,
An imaging step of capturing an image of the periphery of the host vehicle;
In the image captured in time series by the imaging step, for a reference point of any reference image, a moving position search step for searching for an actual actual moving point in a detection image after the reference image,
When the reference point in the reference image is a stopped object, an ideal position calculating step for calculating an ideal ideal movement point in the detected image of the reference point;
An angle difference calculating step of calculating an angle difference between a movement direction from the reference point to the ideal movement point and a movement direction from the reference point to the actual movement point;
A moving speed calculating step for calculating a moving speed from the reference point to the actual moving point;
A frequency distribution step for obtaining a frequency distribution for the angle difference obtained by the angle calculation step and the movement speed obtained by the movement speed calculation step;
And a detection step of detecting an object that approaches the vehicle in parallel or overtakes based on the frequency distribution obtained in the frequency distribution step. An object detection program for detecting an object around the host vehicle,
For a reference point of an arbitrary reference image among images taken in time series, a moving position search means for searching for an actual actual moving point in a detection image after the reference image;
When the reference point in the reference image is a stopped object, ideal position calculation means for calculating an ideal ideal movement point in the detected image of the reference point;
An angle difference calculating means for calculating an angle difference between a moving direction from the reference point to the ideal moving point and a moving direction from the reference point to the actual moving point;
A moving speed calculating means for calculating a moving speed from the reference point to the actual moving point;
Frequency distribution means for obtaining a frequency distribution for the angle difference obtained by the angle difference calculation means and the movement speed obtained by the movement speed calculation means;
Based on the frequency distribution obtained by the frequency distribution means, detection means for detecting an object that approaches the vehicle in parallel or overtakes,
An object detection program for causing a computer to function as