JP5533323B2 - Movement amount estimation apparatus and movement amount estimation method - Google Patents

Description

  The present invention relates to a movement amount estimation apparatus and a movement amount estimation method.

  Conventionally, an object detection device that detects the amount of movement of an object is known. This object detection apparatus calculates an optical flow that is a movement vector of feature points on an image based on two images at predetermined time intervals. Then, the object detection device detects the amount of movement of the object based on the generated optical flow (see Patent Document 1).

JP 2008-97126 A

  However, the conventional object detection device has room for improvement in the estimation accuracy of the movement amount. For example, in a conventional object detection apparatus, an optical flow is calculated by determining feature points from the front or rear surface of a vehicle light or bumper. However, the front or rear surface of the vehicle is a small area from the whole vehicle, and the amount of information is small, which causes a decrease in estimation accuracy. For this reason, when trying to determine the feature points from the entire front and rear side of the vehicle including the windshield and rear glass, there is a difference in depth between the bumper part and the glass part. It will decline.

  The present invention has been made to solve such a conventional problem, and an object of the present invention is to provide a movement amount estimation apparatus and a movement amount estimation method capable of improving the estimation accuracy of the movement amount. There is to do.

The movement amount estimation apparatus according to the present invention includes: an imaging unit that images a predetermined region; an optical flow calculation unit that calculates an optical flow from an image obtained by imaging by the imaging unit; and an assumed vehicle side surface and road surface. A ground line segment estimation means for estimating a ground line segment that is a line segment to be intersected, and an optical flow calculation means in a plane extending vertically upward including the ground line segment estimated by the ground line segment estimation means Assuming that the calculated optical flow exists, a vector calculation means for calculating a vector in real space from the optical flow, a clustering means for clustering the vectors calculated by the vector calculation means, and a result of clustering by the clustering means From the vector belonging to the largest cluster with the largest number of vectors Characterized in that it comprises a movement amount estimating means for estimating the amount of movement of the vehicle, a.

  According to the present invention, it is assumed that an optical flow exists in a plane extending vertically upward including a ground line segment that intersects the assumed vehicle side surface and the road surface, and a vector in the real space of the optical flow is calculated. To do. In this way, since the vector in the real space is calculated on the assumption that the optical flow exists in the plane including the vehicle side surface, the movement amount can be estimated from the vector in the real space, and the accuracy of the movement amount estimation is improved. Can be improved. In particular, since it is a vehicle side vector in real space, the amount of movement in the lateral direction of the vehicle is small when the vehicle is traveling. Therefore, by estimating the amount of movement from the vector of the vehicle side surface, Can easily reduce the estimation error. Accordingly, the estimation accuracy of the movement amount can be improved.

It is a schematic block diagram of the movement amount estimation apparatus which concerns on this embodiment. It is a top view which shows the driving state of the vehicle shown in FIG. It is a block diagram which shows the detail of the computer shown in FIG. It is a figure which shows an example of the virtual surface set by the area | region setting part shown in FIG. It is a figure which shows the mode of calculation of the vector of real space by the vector calculation part shown in FIG. 3, Comprising: (a) shows the captured image, (b) has shown the vector of real space. It is a flowchart which shows the movement amount estimation method which concerns on this embodiment, Comprising: The process until the setting of a virtual surface is shown. It is a flowchart which shows the movement amount estimation method which concerns on this embodiment, Comprising: The process until estimation of a movement amount is shown. It is a schematic block diagram of the movement amount estimation apparatus which concerns on 2nd Embodiment. It is a top view explaining the estimation method of the ground line segment by the movement amount estimation apparatus according to the second embodiment. It is a flowchart which shows the movement amount estimation method which concerns on 2nd Embodiment, Comprising: The process until the setting of a virtual surface is shown.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments of the invention will be described with reference to the drawings. FIG. 1 is a schematic configuration diagram of a movement amount estimation apparatus 1 according to the present embodiment. As shown in FIG. 1, the movement amount estimation device 1 is mounted on a vehicle V <b> 1 and includes a camera (imaging unit) 10, a three-dimensional object detection radar (three-dimensional object detection unit) 20, and a calculator 30.

  The camera 10 shown in FIG. 1 is mounted so that the optical axis is at an angle θ from the horizontal to the downward direction at the position of the height h at the rear of the vehicle V1. The camera 10 captures a predetermined area from this position. The three-dimensional object detection radar 20 irradiates a laser beam behind the vehicle V and collects information corresponding to the position and distance of the three-dimensional object from the reflected light. The laser beam irradiation range of the three-dimensional object detection radar 20 can be irradiated so as to cover at least the imaging range of the camera 10. The computer 30 estimates the movement amount for the three-dimensional object on the vehicle rear side. In this embodiment, as described above, the three-dimensional object detection radar 20 is described as a laser radar. However, other types of radar devices such as a millimeter wave radar may be used.

  FIG. 2 is a top view showing a traveling state of the vehicle V1 shown in FIG. As shown in FIG. 2, the camera 10 images the vehicle rear side at a predetermined angle of view a. At this time, in addition to the lane in which the host vehicle V1 travels, the left and right lanes can be imaged within the angle of view of the camera 10.

  FIG. 3 is a block diagram showing details of the computer 30 shown in FIG. In FIG. 3, the camera 10 and the vehicle speed sensor 20 are also illustrated in order to clarify the connection relationship.

  As shown in FIG. 3, the computer 30 includes an area setting unit (ground line segment estimation unit, three-dimensional object detection unit) 31, a vector calculation unit (optical flow calculation unit, vector calculation unit) 32, and a movement amount estimation unit ( Movement amount estimation means, clustering means, vehicle judgment means) 33.

  The region setting unit 31 estimates a ground line segment. The ground line segment is a line segment that is an intersection line between the assumed vehicle side surface and the road surface. In obtaining the ground line segment, the region setting unit 31 detects the external position of the three-dimensional object around the host vehicle based on the information from the three-dimensional object detection radar 20.

  More specifically, the region setting unit 31 can specify the position of the surrounding three-dimensional object with respect to the host vehicle by the reflected light from the three-dimensional object detection radar 20. As shown in FIG. 2, since the three-dimensional object detection radar 20 irradiates the rear of the vehicle with laser light, the region setting unit 31 detects the outer positions P1 and P2 of the other vehicle V2 located behind the host vehicle V1. It is possible.

  The area setting unit 31 linearly approximates the outer position P1 along the lane among the outer positions P1 and P2 detected as described above, and estimates the ground line segment L from the perpendicular of the approximate line. In this way, the region setting unit 31 presumably specifies the vehicle side surface and estimates the ground line segment L that is an intersection line between the side surface and the road surface.

The area setting unit 31 sets the virtual plane after calculating the ground line segment L. FIG. 4 is a diagram illustrating an example of a virtual plane set by the area setting unit 31 illustrated in FIG. 3. As shown in FIG. 4, the region setting unit 31 sets a surface that includes the ground line segment L and extends vertically upward as a virtual surface IS. The height h _IS virtual surface IS is limited to a predetermined height. The predetermined height is determined in advance so that the calculation cost described later does not become large and a general vehicle height is included. Thereby, it can avoid including an unnecessary area | region, including a vehicle side part. In addition, the region setting unit 31 transmits information on the virtual plane IS to the vector calculation unit 32.

なお、ｍは正の整数である。また、座標系は図４に示すように画像縦方向がＸ軸とされ、画像横方向がＹ軸とされる。 The vector calculation unit 32 inputs captured image data obtained by imaging with the camera 10 and calculates an optical flow from the captured image data. The optical flow can be calculated as described in, for example, JP-A-2008-97126. An optical flow is obtained by the vector calculation unit 32. If the start point coordinates of the optical flow are (xs, ys) and the end point coordinates are (xe, ye), the obtained results are as follows.

Note that m is a positive integer. In the coordinate system, as shown in FIG. 4, the vertical direction of the image is the X axis, and the horizontal direction of the image is the Y axis.

  The vector calculation unit 32 calculates a real space vector from the optical flow on the assumption that the calculated optical flow exists in the virtual plane IS. 5A and 5B are diagrams illustrating how the real space vector is calculated by the vector calculation unit 32 illustrated in FIG. 3. FIG. 5A illustrates a captured image, and FIG. 5B illustrates a real space vector. .

For example, the optical flow is calculated as shown in FIG. The coordinates of the optical flow OP can be specified on the image, but since the position (particularly the depth) of the three-dimensional object showing the optical flow is unknown on the image, the vector in the real space is unknown. Therefore, the vector calculation unit 32 calculates the optical flow OP as a vector in the real space on the assumption that the optical flow OP exists in the virtual plane IS as shown in FIG. At this time, the vector calculation unit 32 performs various calculation processes to calculate a vector in the real space from the optical flow OP. As for the background and the like, when the host vehicle V1 is traveling, an optical flow OP is generated toward the rear of the vehicle.

ここで、Ｘ０〜Ｚ０はカメラ１０の位置を示し、ＤＸ〜ＤＺが画像上の座標（ｘ，ｙ）から決定される値である。また、ｔは変数である。 Specifically, the vector calculation unit 32 executes the following processing. First, assuming that the depth of a point on the captured image is unknown, the following expression is satisfied.

Here, X0 to Z0 indicate the position of the camera 10, and DX to DZ are values determined from the coordinates (x, y) on the image. T is a variable.

The virtual surface IS can be expressed by the following equation.

The vector calculation unit 32 substitutes equation (1) into equation (2). Thereby, the variable t shown in Formula (1) is obtained. As a result, three-dimensional coordinates can be obtained, and the start point and end point of the optical flow OP are converted into real space coordinates as shown below.

Moreover, the vector calculation part 32 calculates a vector from the following formula | equation (3), after calculating | requiring the coordinate of real space.

なお、（Ｌ０，Ｌ１，Ｌ２）及び（Ｌ３，Ｌ４，Ｌ５）は仮想面ＩＳ内の互いに独立なベクトルであるとする。 Since the vector shown in Expression (3) is a vector in the virtual surface IS, it can be separated into two independent vectors in the virtual surface IS. Therefore, the vector calculation part 32 isolate | separates a vector as shown in the following formula | equation (4).

It is assumed that (L0, L1, L2) and (L3, L4, L5) are mutually independent vectors in the virtual plane IS.

Further, v1, v2 and vm can be expressed by vector components xx and yy, and are expressed as follows.

FIG. 3 will be referred to again. The movement amount estimation unit 33 estimates the movement amount of the other vehicle V2 based on the vectors v1 to vm calculated by the vector calculation unit 32. In estimating the movement amount, the movement amount estimation unit 33 first clusters the vectors v1 to vm, and estimates the movement amount of the other vehicle V2 from the vector belonging to the largest cluster having the largest number of vectors as a result of the clustering. is there. For clustering, existing methods such as Nearest Neighbor method and K-Means method may be adopted, or these approximation methods may be adopted. The vectors belonging to the largest cluster are assumed to be vectors vb1 to vbn (n is a positive integer less than or equal to m).

  By this clustering, the optical flow OP directed toward the rear of the vehicle shown in FIG. 5B is removed. That is, the length of the optical flow OP varies depending on the distance from the camera 10 even if the host vehicle V1 is traveling at a constant speed. For this reason, the optical flow OP such as the background does not belong to the maximum cluster, and the optical flow OP directed toward the rear of the vehicle is removed.

In more detail, the movement amount estimation unit 33 estimates the movement amount v of the other vehicle V2 based on the average value of the vectors vb1 to vbn belonging to the maximum cluster. That is, the movement amount estimation unit 33 estimates the movement amount v of the other vehicle V2 from the following equation (5).

  As described above, the movement amount estimation apparatus 1 according to the present embodiment calculates the vectors v1 to vm in the real space on the assumption that the optical flow OP exists in the virtual plane IS including the assumed vehicle side surface. The amount of movement can be estimated from the real space vectors v1 to vm, and the accuracy of the amount of movement estimation can be improved. In particular, since the vehicle side vectors v1 to vm in real space, the amount of movement in the lateral direction of the vehicle is small when the vehicle is traveling, so that the amount of movement is estimated from the vector to estimate from the front and rear surfaces of the vehicle. This makes it easier to reduce the estimation error.

  Next, the movement amount estimation method according to the present embodiment will be described. FIG. 6 is a flowchart showing the movement amount estimation method according to this embodiment, and shows processing up to the setting of the virtual plane IS. As shown in FIG. 6, first, the region setting unit 31 detects the external positions P1, P2 of the three-dimensional object based on information from the three-dimensional object detection radar 20 (S1).

  Next, the area setting unit 31 extracts the outline positions P1, P2 detected in step S1 along the lane (S2). Next, the region setting unit 31 linearly approximates the outer position P1 along the lane, and estimates the ground line segment L from the approximate straight line (S3). After the estimation, the region setting unit 31 determines whether or not the number of detected external positions P1 and P2 detected in step S1 is equal to or greater than a predetermined value (S4). Note that the predetermined value is determined in advance by experimentally determining the number of external positions P1, P2 that should be detected in a normal vehicle.

  If it is determined that the number of detected outer positions P1 and P2 is not equal to or greater than the predetermined value (S4: NO), the area setting unit 31 determines that there is no other vehicle V2 (S5), and then the process shown in FIG. To do. That is, when the other vehicle V2 exists, a certain number of detections can be obtained. Therefore, when the number of detections is less than the predetermined value, the region setting unit 31 determines that the other vehicle V2 does not exist.

  On the other hand, when it is determined that the number of detected external positions P1, P2 is equal to or greater than a predetermined value (S4: YES), the area setting unit 31 determines that another vehicle V2 exists (S6), and sets the virtual plane IS. (S7). Thereafter, the process shown in FIG. 6 ends.

  FIG. 7 is a flowchart showing the movement amount estimation method according to the present embodiment, and shows processing up to the estimation of the movement amount. As shown in FIG. 7, first, the vector calculation unit 32 calculates an optical flow OP from the captured image data (S11). Next, the vector calculation unit 32 converts the optical flow OP from the above-described equations into vectors v1 to vm in the real space (S12). At this time, the vector calculation unit 32 calculates the vectors v1 to vm on the assumption that the optical flow OP exists in the virtual plane IS as described above.

Thereafter, the movement amount estimation unit 33 clusters the vectors v1 to vm converted in step S12 (S13). Then, the movement amount estimation unit 33 determines whether or not the number of vectors belonging to the maximum cluster is equal to or greater than a predetermined number (S14). When it is determined that the number is not equal to or greater than the predetermined number (S14: NO), the movement amount estimation unit 33 determines that the target from which the optical flow OP is obtained is not the other vehicle V2 (S15). That is, it is determined that the other vehicle V2 does not exist behind the host vehicle V1. Here, when the other vehicle V2 exists, a certain number of vectors can be obtained. As shown in FIG. 5B, the vectors of the other vehicle V2 are basically the same. Therefore, a certain number of vectors belonging to the maximum cluster are secured. Therefore, when it is determined that the number of vectors belonging to the maximum cluster is not equal to or greater than the predetermined number (S14: NO), the movement amount estimation unit 33 determines that the other vehicle V2 does not exist behind the host vehicle V1. Then, the process shown in FIG. 7 ends. The predetermined number of vectors is a value obtained in advance by experiments or the like, in which the number of vectors that should be detected in the case of a vehicle is set in advance.

  On the other hand, when it is determined that the number of vectors belonging to the maximum cluster is greater than or equal to the predetermined number (S14: YES), the movement amount estimation unit 33 determines that the target from which the optical flow OP is obtained is the other vehicle V2 (S16). ). That is, it is determined that the other vehicle V2 exists behind the host vehicle V1. Then, the movement amount estimation unit 33 estimates the movement amount of the other vehicle V2 (S17). As for the calculation of the movement amount, averaging or the like may be performed as described above. Thereafter, the process shown in FIG. 7 ends.

  Thus, according to the movement amount estimation apparatus 1 and the movement amount estimation method according to the present embodiment, the virtual plane IS extending vertically upward including the ground line segment L that is the intersection line between the assumed vehicle side surface and the road surface. Assuming that an optical flow OP exists in the image, a vector in the real space of the optical flow OP is calculated. As described above, assuming that the optical flow OP exists in the virtual plane IS including the vehicle side surface, the vectors v1 to vm in the real space are calculated. Therefore, the movement amount can be estimated from the vectors v1 to vm in the real space. And the accuracy of movement amount estimation can be improved. In particular, since the vectors v1 to vm are vectors on the side surface of the vehicle in real space, the amount of movement in the lateral direction of the vehicle is small when the vehicle is traveling. Therefore, by estimating the amount of movement from the vectors v1 to vm, In addition, the estimation error can be reduced more easily than when estimating from the rear surface. Accordingly, the estimation accuracy of the movement amount can be improved.

  Further, the calculated vectors v1 to vm are clustered, and the movement amount of the vehicle is estimated from the vectors vb1 to vbn belonging to the largest cluster having the largest number of vectors as a result of clustering. Basically, the vector corresponding to the vehicle in the virtual plane IS extending vertically upward including the ground line segment is the same. Further, a portion corresponding to the background in the virtual plane IS is a vector different from the above vector. Therefore, as a result of clustering, the movement amount can be estimated by estimating the movement amount of the vehicle from the vectors vb1 to vbn belonging to the largest cluster having the largest number of vectors. In particular, since clustering is performed, even if noise or the like is included in the vector in the vehicle portion, it can be removed. Therefore, the estimation accuracy of the movement amount can be further improved.

  It is determined that the other vehicle V2 exists when the number of vectors of the maximum cluster is equal to or greater than the predetermined number, and it is determined that the other vehicle V2 does not exist when the maximum number of vectors is less than the predetermined number. Here, when the other vehicle V2 exists, each vector of the other vehicle V2 itself is basically the same vector, and therefore, the number of vectors of the maximum cluster is a predetermined number or more. On the other hand, when there is no vehicle, various vectors such as backgrounds are included and the number of vectors of the maximum cluster does not exceed a predetermined number. Therefore, the presence or absence of the other vehicle V2 can be determined from the number of vectors of the maximum cluster.

  Further, since the movement amount of the other vehicle V2 is estimated based on the average value of the vectors vb1 to vbn belonging to the largest cluster, even if noise is included in the vectors vb1 to vbn belonging to the largest cluster, it can be averaged and moved further. The amount estimation accuracy can be improved.

  In addition, since the virtual surface IS that is assumed to have the optical flow OP is limited to a predetermined height from the road surface, it is only necessary to perform an operation on the virtual surface IS within the predetermined height range, thereby reducing the calculation cost. it can.

  Further, of the detected external positions P1 and P2 of the three-dimensional object, the external position P1 along the lane is linearly approximated, and the ground line segment L is estimated from the perpendicular of the approximate straight line. Since the ground line segment L is estimated, the estimation accuracy of the ground line segment L can be improved. Thereby, the virtual surface IS can be made to coincide with the vehicle side surface with high accuracy.

  Next, a second embodiment of the present invention will be described. The movement amount estimation apparatus 1 and the movement amount estimation method according to the second embodiment are the same as those of the first embodiment, but a part of the configuration and processing contents are different. Hereinafter, differences will be described.

  FIG. 8 is a schematic configuration diagram of the movement amount estimation apparatus 2 according to the second embodiment. As illustrated in FIG. 8, the movement amount estimation apparatus 2 according to the second embodiment does not include the three-dimensional object detection radar 20. For this reason, the method of estimating the ground line segment L by the region setting unit 31 is different from that of the first embodiment.

  FIG. 9 is a top view for explaining a method of estimating the ground line segment L by the movement amount estimating apparatus 2 according to the second embodiment. As shown in FIG. 9, in the second embodiment, the area setting unit 31 recognizes the lane marking L 'by white line recognition or the like, and grounds a position away from the lane marking L' by a predetermined distance d. The line segment L is estimated. Thus, in the second embodiment, since the ground line segment L is estimated from the lane line L ′, the calculation cost is reduced. Note that the predetermined distance d is a mode obtained by sampling a plurality of travel positions (distances from the lane line L ′) of a vehicle that actually travels in the adjacent lane, and calculating an average value of the plurality of sampled distances or statistical processing. A value or the like is set in advance.

  In the second embodiment, the setting accuracy of the virtual surface IS may be lower than that in the first embodiment. However, if the predetermined distance d is set appropriately, the virtual surface IS is the same as the virtual surface IS that should be originally set. The difference does not become too large, and there is no problem in practical use. Instead of the above, the region setting unit 31 may estimate the ground line segment L from the relative position from the camera 10 (the host vehicle V1).

  FIG. 10 is a flowchart showing the movement amount estimation method according to the second embodiment, and shows processing up to the setting of the virtual plane IS. As shown in FIG. 10, first, the region setting unit 31 estimates the ground line segment L (S21). As described above, the estimation method sets the position separated from the lane line L ′ by a predetermined distance d as the ground line segment L. Thereafter, the area setting unit 31 sets the virtual surface IS (S22). The setting method of the virtual plane IS is the same as that in step S7 shown in FIG.

  Thus, according to the movement amount estimation apparatus 2 and the movement amount estimation method according to the second embodiment, the estimation accuracy of the movement amount can be improved as in the first embodiment, and whether or not the other vehicle V2 exists. Can be determined, and the calculation cost can be reduced. In addition, the estimation accuracy of the ground line segment L can be improved.

  Furthermore, according to the second embodiment, since the position separated from the lane line L ′ by a predetermined distance d is estimated as the ground line segment L, the calculation cost can be suppressed in estimating the ground line segment L, and the apparatus configuration can be reduced. Simplification can be achieved.

  As described above, the present invention has been described based on the embodiments, but the present invention is not limited to the above-described embodiments, and modifications may be made without departing from the spirit of the present invention.

  For example, although the movement amount estimation apparatus 1 according to the first embodiment includes the three-dimensional object detection radar 20, the present invention is not limited to this, and any device that can detect the position of the other vehicle V2 may be used. Good.

DESCRIPTION OF SYMBOLS 1 ... Movement amount estimation apparatus 10 ... Camera (imaging means)
20 ... Solid object detection radar (three-dimensional object detection means)
30 ... Calculator 31 ... Area setting unit (ground line segment estimation means, solid object detection means)
32. Vector calculation unit (optical flow calculation means, vector calculation means)
33 ... Movement amount estimation unit (movement amount estimation means, clustering means, vehicle determination means)
a ... angle of view d ... predetermined distance IS ... virtual plane L ... ground line segment L '... lane marking OP ... optical flow P1, P2 ... detection position V1 ... own vehicle V2 ... other vehicle

Claims (8)

Imaging means for imaging a predetermined area;
An optical flow calculation means for calculating an optical flow from an image obtained by imaging by the imaging means;
A ground line segment estimation means for estimating a ground line segment that is a line segment that intersects the assumed vehicle side surface and the road surface;
Assuming that the optical flow calculated by the optical flow calculation means exists in a plane extending vertically upward including the ground line segment estimated by the ground line segment estimation means, a vector in real space is calculated from the optical flow. Vector calculating means for
Clustering means for clustering the vectors calculated by the vector calculation means;
As a result of clustering by the clustering means, a movement amount estimation means for estimating a movement amount of the vehicle from a vector belonging to the largest cluster having the largest number of vectors;
A movement amount estimation apparatus comprising: Wherein it is determined that the vehicle is present when the maximum number of vectors in the cluster is equal to or greater than a predetermined number of predetermined claim 1, further comprising a vehicle determining means for determining that the vehicle when less than the predetermined number is not present The movement amount estimation apparatus according to 1. The movement amount estimating means, on the basis of the average value of the maximum cluster belongs vector, the movement amount estimating apparatus according to claim 1 or claim 2, characterized in that to estimate the amount of movement of the vehicle. Imaging means for imaging a predetermined area;
  An optical flow calculation means for calculating an optical flow from an image obtained by imaging by the imaging means;
  A ground line segment estimation means for estimating a ground line segment that is a line segment that intersects the assumed vehicle side surface and the road surface;
  Assuming that the optical flow calculated by the optical flow calculation means exists in a plane extending vertically upward from the ground line segment of the vehicle side estimated by the ground line segment estimation means, a vector in real space is calculated from the optical flow. Vector calculating means for calculating;
  A movement amount estimation means for estimating a movement amount of the vehicle based on the vector calculated by the vector calculation means;
A movement amount estimation apparatus comprising: The movement according to any one of claims 1 to 4 , wherein the ground line segment estimation means estimates a position separated from a lane line by a predetermined distance as a ground line segment. Quantity estimation device. Imaging means for imaging a predetermined area;
An optical flow calculation means for calculating an optical flow from an image obtained by imaging by the imaging means;
A ground line segment estimation means for estimating a ground line segment that is a line segment that intersects the assumed vehicle side surface and the road surface;
Assuming that the optical flow calculated by the optical flow calculation means exists in a plane extending vertically upward including the ground line segment estimated by the ground line segment estimation means, a vector in real space is calculated from the optical flow. Vector calculating means for
A movement amount estimation means for estimating a movement amount of the vehicle based on the vector calculated by the vector calculation means;
And a solid object detecting means for detecting the contour position of the three-dimensional object around the vehicle,
The ground line segment estimation means linearly approximates the external position along the lane among the external positions of the solid object detected by the solid object detection means, and estimates the ground line segment from the perpendicular of the approximate straight line. momentum estimator transfer characterized. The movement amount estimation apparatus according to any one of claims 1 to 6, wherein a plane on which an optical flow is assumed is limited to a predetermined height predetermined from a road surface. An optical flow calculation step of calculating an optical flow from an image obtained by imaging by an imaging means for imaging a predetermined area;
A ground line segment estimation step for estimating a ground line segment that is a line segment that intersects the vehicle side surface and the road surface;
Assuming that the optical flow calculated in the optical flow calculation step exists in a plane extending vertically upward including the ground line segment estimated in the ground line segment estimation step, a vector in the real space of the optical flow is calculated. A vector calculation step,
A clustering step of clustering the vectors calculated in the vector calculation step;
As a result of clustering in the clustering step, a movement amount estimation step for estimating a movement amount of the vehicle from a vector belonging to the largest cluster having the largest number of vectors;
A moving amount estimation method characterized by the above.

Images