JP2006024146A - Method and instrument for measuring moving object by image processing - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and an instrument for measuring a moving object by image processing reducing a measuring error. <P>SOLUTION: Physical quantities about representative points are measured using a substantial rear end of the moving object as the representative point, when the moving body is moved along a direction separated from a video camera 10 for imaging the moving body, and using a substantial front end of the moving object as the representative point, when the moving body is moved along a direction approaching to the video camera 10. A relative position with respect to an area of the moving object may be determined preliminarily as the representative point. A geometric gravity center of the moving object may be found as the representative point, when the moving body is determined not to be overlapped with another moving body, and a point corresponding to the geometric gravity center when determined not to be overlapped may be found as the representative point, on the basis of a moving vector of the moving body, when the moving body is detected to be overlapped with another moving body. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a moving object measuring method and apparatus for processing a time-series image and measuring a physical quantity such as a position, passing time, or speed of a moving object (movable object such as a car, a bicycle, or an animal) in the image.

  By capturing a moving object with a video camera, performing image processing, and tracking the moving object on the image, it is possible to automatically detect a traffic state or a traffic accident.

  As a method of tracking a moving object on an image, for example, a frame of 480 × 640 pixels is divided into 8 × 8 pixel blocks to form 60 × 80 block images, and each block image is an image of a block corresponding to the background image. Compared to the above, a method of detecting a moving object in units of blocks and obtaining an identification code (ID) and a motion vector (MV) of the moving objects in units of blocks based on the space-time correlation of the frame images at times t-1 and t. (For example, Patent Documents 1 to 3 below). Thereby, the same ID is given to the blocks recognized as the same moving object.

  However, since the moving object on the image is tracked, a specific error occurs when the position, passing time, speed, or the like of the moving object on the image is measured.

  For example, since a moving object is detected in units of blocks as described above, a moving object position measurement error having a maximum width of 1 block occurs.

  As shown in FIG. 3, since the distance from the video camera 10 varies depending on the position on the road surface 11, even if the block size is uniform on the image frame as shown in FIG. Block width) depends on the position in the frame. For example, the actual block widths Ka and Kb in FIG. 3 are the same width on the image, but the actual block width Kb is several times longer than the actual block width Ka. For this reason, the measurement error differs greatly depending on the position in the image frame.

  Furthermore, since the distance from the video camera differs depending on the height of the moving object even in the block at the same position, the actual block width is different and the measurement error is also different. In addition, since the frame rate is finite, the moving object moves discontinuously, and an error occurs in the measurement of the time required for the moving object to pass through the predetermined section.

  Therefore, when the average speed of the predetermined section of the moving object is measured by image processing, the error changes depending on the condition, and the reliability of the measurement value is lowered.

Also, as shown in FIG. 16, when the representative point of a cluster that is a block of blocks having the same ID is used as the geometric center of gravity of the cluster, if a part of the moving object is hidden behind another moving object, the representative point becomes Since the position is shifted from the position when the moving object is not hidden, the error fluctuates when the position, passing time, speed, or the like of the moving object is measured based on the representative point. FIG. 16 shows the same moving objects 131 to 133 at times t1 to t3 and the same moving objects 141 to 143 at times t1 to t3 superimposed on each other. RA1 to RA3 are moving objects 131 to 133, respectively. RB1 to RB3 are geometric centroids of the moving objects 141 to 143, respectively.
JP 2002-133421 A JP 2003-006655 A JP 2003-263626 A

  In view of the above problems, an object of the present invention is to provide a moving object measuring method and apparatus by image processing, which can further reduce a measurement error.

  Another object of the present invention is to provide a moving object measurement method and apparatus by image processing that can increase the reliability of measurement values.

  In the first aspect of the moving object measuring method by image processing according to the present invention, when the moving object moves in a direction away from the video camera that images the moving object, the substantially rear end of the moving object is used as the representative point, and the video camera When the moving object moves in a direction approaching, the approximate front end of the moving object is used as the representative point, and a physical quantity related to the representative point is measured.

  In the second aspect of the moving object measuring method based on image processing according to the present invention, a relative position with respect to the area of the moving object is determined in advance as a representative point of the moving object, and a physical quantity related to the representative point is measured.

  In the third aspect of the moving object measuring method by image processing according to the present invention, when it is determined that the moving object does not overlap another moving object, the geometric center of gravity of the moving object is obtained as a representative point, and the movement When it is detected that an object overlaps another moving object, a point corresponding to the geometric center of gravity when there is no overlap is obtained as a representative point based on the motion vector of the moving object. Then, a physical quantity related to the representative point is measured.

  According to the configuration of the first aspect, the representative point of the moving object is almost near the reference plane regardless of the height of the moving object, so that the measurement error of the physical quantity related to the position set on the reference plane can be reduced. it can.

  According to the configuration of the second aspect, since the relative position with respect to the area of the moving object is set, even if the shadow of the moving object is recognized as the moving object, the moving object is in the vicinity of the target position according to the physical quantity to be measured. Can be detected as a representative point, so that measurement errors can be reduced.

  According to the configuration of the third aspect, even if there is an overlap between moving objects on the image, the point corresponding to the geometric center of gravity when there is no overlapping of moving objects based on the motion vector of the moving object Is obtained as a representative point, so that the measurement error of the physical quantity obtained based on the representative point can be reduced.

  Other configurations, operations, and effects of the present invention will become apparent from the following description.

  Embodiments of the present invention will be described below with reference to the drawings. In the drawings, the same or similar elements are denoted by the same or similar reference numerals.

  FIG. 1 is a schematic functional block diagram of a moving object measuring apparatus 20 according to a first embodiment of the present invention that measures an average speed of a moving object by processing an image captured by a video camera (for example, an ITV camera) 10.

  The moving object measuring device 20 other than the storage unit can be configured by computer software, dedicated hardware, or a combination of computer software and dedicated hardware.

  The time-series images captured by the video camera 10 are stored in the image memory 21 at a rate of 30 frames / second, for example, and the oldest frame image is rewritten with a new frame image.

  In the present invention, not only when the image data stored in the image memory 21 is processed in real time by the components 23 to 26 but also when stored in an external storage device (not shown) and only necessary portions are processed later. It can also be applied to.

  Further, even if the image (original image) stored in the image memory 21 is directly processed by the constituent elements 23 to 26, it is converted into a spatial difference frame image by applying a filter such as a Laplacian filter or two-dimensional Fourier transform In addition, a configuration in which components (transformed images) obtained by performing inverse Fourier transform after emphasizing components having a frequency higher than a predetermined frequency are processed by the components 23 to 26 may be used. Hereinafter, “image” means an original image or a converted image.

  The background image generation unit 22 includes a storage unit and a processing unit. The processing unit accesses the image memory 21, and for example, the pixel values of the corresponding pixels of the frame image of all or thinned out objects in the past one minute. A histogram is created, and an image having the mode value (mode) as the pixel value of the pixel is generated as a background image in which no moving object exists, and this is stored in the storage unit. The background image is updated by performing this process periodically.

  In the ID generation / annihilation unit 23, data on the positions and sizes of the entrance slits A1 and B2 and the exit slits B1 and A2 arranged in the frame 15 as shown in FIG. 4 is set in advance.

  This setting is performed by a user by a setting means including a pointing device such as a mouse or an input device IN such as a keyboard, a display device OUT for confirming the setting, and an input processing program included in the ID generation / deletion unit 23. Based on.

  The mesh in FIG. 4 is a block boundary, and one block is, for example, 8 × 8 pixels. The width of the slits A1, B1, A2, and B2 in the moving object passing direction is equal to one block width in FIG. 4, but may be an integral multiple of one block width.

  The ID generation / annihilation unit 23 reads the image data in the entrance slits A1 and B2 from the image memory 21, and determines whether or not there is a moving object inside each block. Whether there is a moving object in a certain block is determined by whether the sum of absolute values of differences between the pixels in the block and the corresponding pixels in the background image is equal to or greater than a predetermined value.

  When the ID generation / annihilation unit 23 determines that there is a moving object in the block, the ID generation / annihilation unit 23 assigns a new moving object identification code (ID) to the block. If the ID generation / annihilation unit 23 determines that a moving object exists in a block adjacent to the ID-assigned block, the ID generation / annihilation unit 23 assigns the same ID as the assigned block to this adjacent block. This ID-added block includes a block adjacent to the slit. Hereinafter, a block of blocks to which an ID is assigned is referred to as a cluster.

  The ID is assigned to the corresponding block of the object map in the storage unit 24. The object map is for storing the moving object information of each block of 60 × 80 blocks in the above-described case, and the moving object information is provided with a flag indicating whether an ID is assigned and an ID. The ID and the motion vector (MV) of the block. Instead of using the flag, it may be determined that no ID is assigned when ID = 0. The most significant bit of the ID may be used as a flag. The MV of a block whose MV cannot be obtained by block matching is set equal to, for example, the average value of MVs of blocks having the same ID around it.

  The cluster that has passed through the entrance slit A1 or B2 is tracked by the moving object tracking unit 25 by a known method. For example, for high-speed processing, the approximate movement range of the cluster at time t is estimated based on the MV at time t-1 (frame number), and the above-mentioned moving object existence determination is performed for each block within this range. ID is assigned to the block on the moving direction side, and ID is erased on the block on the opposite direction to the movement, and the cluster at time t is determined. Further, the MV of the block to which the ID is assigned at time t is obtained by block matching between the frame images at time t−1 and t. For high-speed processing, the range in which block matching is performed is determined based on the MV at time t-1. The ID and MV of each block can be simultaneously determined using an evaluation function (for example, Patent Document 3 above). The ID and MV are updated on the object map in the storage unit 24.

  The tracking processing by the moving object tracking unit 25 is performed up to the inside of the exit slit for each cluster.

  The ID generation / annihilation unit 23 further checks whether an ID is assigned to the blocks in the exit slits B1 and A2 based on the object map in the storage unit 24, and if so, the cluster passes through the exit slit. Sometimes the ID disappears.

  FIG. 5 superimposes FIG. 4 on the object map of ID for easy understanding.

  The feature of the first embodiment lies in a partial configuration of the average speed and miscalculation calculation unit 26 and the ID generation / annihilation unit 23 related thereto, which will be described in detail below.

  FIG. 3 is an explanatory diagram for measuring the average speed of a moving object by image processing.

  The video camera 10 is disposed at a height H from the road surface 11 with the optical axis inclined obliquely downward. Generally, when the speed of a moving object is constant, the measurement accuracy increases as the measurement distance increases. Therefore, the average speed V of the distance L1 from the line Pa to Pb in the direction perpendicular to the paper surface is measured. That is, the distance L1 is actually measured, the lines Pa and Pb are set on the frame image, and the average is obtained by measuring the time from when the representative point of the moving object passes through the line Pa to the line Pb. The speed V is calculated.

  An image captured by the video camera 10 is a perspective projection of the subject on a plane S perpendicular to the optical axis of the video camera 10. FIG. 4 shows an image taken by the video camera 10 in which the above-described block boundary lines and slits are drawn.

  In FIG. 4, the lines Pa and Pb respectively coincide with the front ends of the entrance slit A1 and the exit slit B1, and the lines Pa and Pb are set by setting the slits A1 and B1 as described above. . Since the position is measured in units of blocks, the slits themselves may be regarded as lines.

  Here, assuming that the representative point of the moving object is, for example, its geometric center of gravity, the distance between the video camera 10 and the representative point varies depending on the height of the moving object, which causes an increase in the passage time measurement error of the lines Pa and Pb. It becomes. Therefore, when the moving object moves away from the video camera 10 as shown in FIG. 3, the rear end of the moving object is set as the representative point of the moving object. The “rear end” may be any of a plurality of moving object blocks corresponding to the side or one block among them.

  As a result, the representative point of the moving object is almost near the road surface 11 regardless of the height of the moving object, so that a physical quantity relating to the position of the lines Pa and Pb, for example, a measurement error of the position passing time or passing position can be reduced. . In addition, 12A and 12B in FIG. 3 respectively show the same moving object at the time when the trailing end of the moving object passes through the lines Pa and Pb.

  FIG. 2 is a schematic flowchart showing a line passage detection process of the ID generation / annihilation unit 23 of FIG. 1 for the entrance slit A1 and the exit slit B1 of FIG. 4 in one frame image.

  (S10) The above-mentioned moving object existence determination process is performed for each block in the entrance slit A1.

  (S11) If there is a new moving object in the entrance slit A1, the process proceeds to step S12, and if not, the process proceeds to step S13.

  (S12) An ID is assigned to a block in the entrance slit A1 that is determined to have a moving object. Next, the process proceeds to step S15.

  (S13) When the moving object passes through the entrance slit A1, that is, when the moving object block (block to which ID is assigned) existing in the entrance slit A1 in the previous frame is not present in the current frame. In step S14, it is determined that the moving object has passed the line Pa, and the process proceeds to step S14. Otherwise, the process proceeds to step S15.

  (S14) The average speed and error calculation unit 26 in FIG. 1 is notified of the ID of the moving object that has passed through the entrance slit A1 and the current frame number F (the entrance slit passage time).

  (S15) The above-mentioned moving object existence determination process is performed for each block in the exit slit B1.

  (S16) If there is a new moving object in the exit slit B1, the process proceeds to step S17, and if not, the process proceeds to step S18.

  (S17) The ID in the exit slit B1 is temporarily stored and the process is terminated.

  (S18) If the moving object passes through the exit slit B1, the process proceeds to step S19, and if not, the process ends.

  (S19) The average speed and miscalculation calculation unit 26 in FIG. 1 is notified of the ID (ID stored in step S17) and the current frame number F that have passed through the exit slit B1, and the ID stored in step S17 is erased. .

  Based on the ID and the frame number F received from the ID generation / annihilation unit 23 in steps S14 and S19, the average speed and miscalculation calculation unit 26 calculates an average speed V (m / s) by the following formula.

V = L1 / ((Fb−Fa) / f) (1)
Here, Fb and Fa are frame numbers for the same ID received in steps S14 and S19, respectively, and f is a frame rate.

  1 for the entrance slit B2 and the exit slit A2 in FIG. 5 in one frame image is the same as the line passage detection process for the entrance slit A1 and the exit slit B1 described above, respectively. . However, since the moving object 16A moves in a direction approaching the video camera 10, the front end of the moving object is set as a representative point of the moving object. 16B and 16A in FIGS. 4 and 5 respectively show the same moving object when the front end of the moving object passes through the lines Pb and Pa.

  Next, the measurement error δV of the average speed V will be described.

  The error δV mainly includes an error due to a frame rate and an error due to determination of the presence / absence of a moving object in units of blocks.

(1) Position measurement error D due to only frame rate f
For example, when f = 30 and the moving object speed is 60 km / h and 100 km / h, they move 0.56 m and 0.93 m in one frame time (1 / f second), respectively, and these become the maximum position measurement error D. .

(2) Position measurement error K by only block processing
Since the presence / absence of a moving object is determined in units of blocks, a position measurement error of an actual width (actual block width) K corresponding to one block width on the image occurs at the maximum.

  The actual block width K varies depending on the position of the block. In FIG. 3, when the viewing angle of one block width is η, and the actual block widths K of blocks BKa and BKb having lines Pa and Pb shown in FIG. 4 as one end are respectively Ka and Kb as shown in FIG. Is established.

tan (α−η) = H / (Ka + L2) (2)
tan (β−η) = H / (Kb + L1 + L2) (3)
Here, L2 is the distance from directly below the video camera 10 to the line Pa, and α and β are the angles formed by the line of sight from the video camera 10 to the lines Pa and Pb and the road surface 11, respectively.

  For α and β, the following equations hold:

tan (α) = H / L2 (4)
tan (β) = H / (L1 + L2) (5)
Assuming that L1 is divided into n blocks, the following equation is established.

α = β + n · η (6)
For example, when H = 10 m, L1 = 20 m, and L2 = 20 m, α = 0.46 rad and β = 0.25 rad from the above equations (4) and (5), and when n = 25, the above equation (6) Therefore, η = 0.0087 rad. If these values are substituted into the above equations (2) and (3), Ka = 0.45 m and Kb = 1.5 m, which are the maximum errors in the line Pa and Pb passing position measurement.

(3) Average speed measurement error δV
Generally, the right side denominator of the above equation (1) is expressed as T.
δV / V = −δT / T (7)
Is established. Here, δT is an error of the measurement time T.

Since δT and T are proportional to the section frame number measurement error X and the section frame number measurement value (Fb−Fa), respectively,
δV / V = −X / (Fb−Fa) (8)
It is expressed. The section frame number measurement error X is an average error X = ± (δFa + δFb) / 2, where δFa and δFb are frame number errors when the moving object passes through the lines Pa and Pa, respectively.
It is represented by

  Here, the number of frames ΔFa and ΔFb required for a moving object having an average speed V to travel the actual block widths Ka and Kb are expressed by the following equations, respectively.

ΔFa = f · Ka / V (10)
ΔFb = f · Kb / V (11)
The frame number error δFa is ΔFa when ΔFa> 1, that is, when K> D, the frame error is absorbed by the error due to the block processing, and becomes ΔFa at the maximum. The maximum is 1. Similarly, the maximum frame number error δFb is ΔFb when ΔFb> 1, and is maximum 1 when ΔFb ≦ 1.

  FIG. 6 shows a numerical calculation example of the speed measurement error δV when the average speed V is 40, 60, 80 and 100 km / h and the above-mentioned quantity related thereto.

  The average speed and miscalculation calculation unit 26 in FIG. 1 calculates the speed measurement error δV as described above according to the measurement value of the average speed V expressed by the above equation (1), and displays these V and δV as a display device. Display on OUT. As an application example of the first embodiment, for example, when the speed limit is Vlim, if V−δV> Vlim, it is determined that the speed is violated, and this is displayed on the display device OUT and an alarm (not shown) is sounded. The image is stored in an external storage device (not shown).

  According to the first embodiment, since the error δV is displayed according to the measurement value of the average speed V, the reliability of the average speed measurement value becomes high and practical even if there is a complicated error as described above. Can be realized. Further, since speed is measured by image processing without using radio waves and sound waves, there is an advantage that speed measurement is not hindered by speed measurement avoiding means such as jamming radio waves, disturbing sound waves, or scattering members.

  In FIG. 3, a surface Sh having an arbitrary predetermined height from the road surface 11 can be associated with the surface S. Therefore, the representative point for improving the measurement accuracy does not necessarily have to be near the road surface 11. The reference height Sh can be determined according to the physical quantity to be performed and the camera angle.

  FIG. 7 is a schematic functional block diagram of the moving object measuring apparatus 20A according to the second embodiment of the present invention that processes an image captured by the video camera 10 and measures the average speed of the moving object.

  The representative point tracking unit 27 tracks the representative point by obtaining the representative point of the moving object based on the object map in the storage unit 24 for each frame.

  The moving object measuring apparatus 20A needs to track a plurality of moving objects in the frame image in real time, and therefore needs to increase the processing speed. Therefore, when a monochrome image is used, it becomes difficult to distinguish a moving object and its shadow, and a representative point for minimizing measurement error cannot be determined. FIG. 9 shows a case where the shadows 12SA, 12SB, 16SA, and 16SB cannot be distinguished from the moving objects 12A, 12B, 16A, and 16B, respectively.

  The feature of the second embodiment is a method for determining a representative point of a moving object that solves such a problem.

  As shown in FIG. 9, when a moving object is detected in the entrance slit B2, the ID generation / annihilation unit 23A performs a moving object detection process on the block before entering the entrance slit B2. Furthermore, even if a moving object is detected in the exit slit B1, the moving object detection process is performed until the exit slit B1 is completely removed.

  FIG. 8 is a schematic flowchart showing the representative point determination process of the representative point tracking unit 27 for any one moving object cluster.

  (S20) A rectangle inscribed by the cluster of moving objects is obtained. FIG. 10 shows a configuration in which rectangles 17A, 17B, 18A and 18B in which four moving object clusters are inscribed are added to FIG.

(S21) The coordinates of the representative point R are (X0 + px · LY, Y0 + py · LY).
Asking. Here, (X0, Y0) are the coordinates of a rectangular reference point, for example, the lower left vertex of the rectangle. The ratios px and py are values set in advance before the moving object tracking process, satisfy 0 <px <1, 0 <px <1, and are set by the same means as the above-described setting means.

  The average speed and miscalculation calculation unit 26A in FIG. 7 calculates the average speed V of the moving object and its error δV in the same manner as in the first embodiment except for the representative points, and displays these on the display device OUT.

  According to the second embodiment, since the relative position with respect to the area of the moving object is set by the ratios px and py, even if the shadow of the moving object is recognized as the moving object, the moving object corresponding to the physical quantity to be measured is detected. The vicinity of the target position can be detected as a representative point, which greatly contributes to improvement in measurement accuracy.

  Note that the size of the shadow regarded as a part of the moving object varies depending on time and weather, and accordingly, the setting values of the ratios px and py need to be changed accordingly.

  However, for example, the average speed V is measured according to the first embodiment, and time series images suspected of speed violation are stored in an external storage device (not shown), and px and py are set appropriately by looking at the images later. By doing so, it becomes possible to determine speed violation more accurately.

  Further, the average speed and miscalculation calculation unit 26A obtains the length of the trajectory of the representative point as the travel distance L2 based on the actual block width of each block and the coordinates of the representative point obtained by the representative point tracking unit 27. In the same manner as in the first embodiment, the representative point average speed V of the moving object and its error δV may be calculated.

  In the above embodiment, the case where the video camera is arranged above the center of the road has been described. However, for example, when the video camera is arranged at an intersection, the angle of the moving object 30 with respect to the video camera is shown in FIG. Is different from the above case. In FIG. 12, 30S is a shadow of the moving object 30, which is recognized as a part of the moving object, and 31 is a rectangle including them. R is a representative point preferable for reducing the position measurement error of the moving object 30.

  Also, as shown in FIG. 13, when the moving object turns to the right, the direction of the moving object changes, making it difficult to set the representative point of the moving object near the road surface position. In FIG. 13, reference numerals 301 to 303 denote the same moving object at different times during movement, and 301S to 303S denote shadows of the moving objects 301 to 303 recognized as part of the moving object, respectively.

  In the third embodiment, the process shown in FIG. 11 is performed instead of the process of FIG. 8 so that the representative point can be set near the road surface position even in such a case.

  (S30) As in step S20 of FIG. 8, a rectangle inscribed by the cluster of moving objects is obtained. In FIG. 13, reference numerals 311 to 313 denote rectangles inscribed by the moving object clusters.

  (S31) The angle θ of the motion vector MVm of the moving object is obtained, and then the values of the ratio functions px (θ) and py (θ) are obtained. The motion vector MVm is an average vector of motion vectors of all blocks belonging to the same moving object. The ratios px (θ) and py (θ) are determined empirically in advance, satisfying 0 <px (θ) <1, 0 <px (θ) <1, and using θ as a parameter or a mathematical expression It is expressed as an approximation. θ is an angle on the frame image or an angle on the road surface. In the case of an angle on the road surface, an equation or a table for converting the angle on the frame image into an angle on the road surface according to the position on the frame image is used.

  Note that the motion vector MVm may be averaged except for those that deviate from the average by a predetermined value or more. The motion vector MVm may also be an average value of the MVs in or near the block including the geometric centroid.

(S32) Similar to step S21 in FIG. 8, the coordinates of the representative point R are set to (X0 + px (θ) · LY, Y0 + py (θ) · LY).
Asking.

  According to the third embodiment, since the ratios px and py are determined according to the angle of the motion vector of the moving object, the shadow of the moving object is recognized as the moving object, and the moving object changes its moving direction. However, the vicinity of the point of interest of the moving object corresponding to the physical quantity to be measured can be detected as a representative point, which greatly contributes to improvement in measurement accuracy.

  The fourth embodiment of the present invention is for solving the problem shown in FIG. 16 described in the background art section. In the fourth embodiment, the representative point tracking unit 27 in FIG. 7 performs the process shown in FIG. 14 on each moving object on the object map corresponding to each image frame.

  (S40) If there is no overlap between the moving objects on the image, the process proceeds to step S41, and if not, the process proceeds to step S43.

  (S41) The coordinates of the geometric gravity center G are obtained.

  (S42) This G is obtained as a representative point P (t) at time t. In FIG. 15, a cross indicates a representative point, and a geometric center G other than RX2 is a geometric point.

  (S43) The average value MVm of the motion vectors of each block of the moving object of interest is obtained. As described above, the motion vector MVm may be averaged except for those that deviate from the average by a predetermined value or more.

  (S44) A point obtained by moving the representative point P (t-1) of the corresponding moving object in the previous frame by MVm is obtained as the representative point P (t) at time t. RX2 in FIG. 15 is a representative point of the moving object 132 obtained in this way.

  According to the fourth embodiment, even if there is an overlap between moving objects on the image, a point corresponding to the geometric center of gravity when there is no overlapping of moving objects is representative based on the motion vector of the moving object. As can be seen from the comparison between FIG. 15 and FIG. 16, the representative point of the moving object and its trajectory can be obtained more accurately, and the measured values of other physical quantities obtained based on the representative point are also obtained. Become accurate.

1 is a schematic functional block diagram of a moving object measuring apparatus according to a first embodiment of the present invention that measures an average speed of a moving object by processing an image captured by a video camera. It is a schematic flowchart which shows the line passage detection process of the ID production | generation / extinction part 23 of FIG. 1 with respect to the entrance slit and exit slit in 1 frame image. It is a section average speed measurement explanatory view of a moving object by image processing. It is explanatory drawing which shows the image imaged with the video camera of FIG. 3 with a block boundary line, an entrance slit, and an exit slit. It is explanatory drawing which shows what overlapped FIG. 4 on the object map of ID. It is a table | surface which shows the numerical calculation example of speed measurement error (delta) V in case average speed V is 40, 60, 80, and 100 km / h, and the quantity related to this. It is a general | schematic functional block diagram of the moving object measuring device of Example 2 of this invention which processes the image imaged with the video camera, and measures the average speed of a moving object. It is a schematic flowchart which shows the representative point determination process of the representative point tracking part 27 with respect to arbitrary one moving object cluster. It is explanatory drawing which shows the case where the shadow of a moving object cannot be distinguished from a moving object. It is processing explanatory drawing of FIG. It is a schematic flowchart which shows the representative point determination process of the representative point tracking part 27 with respect to arbitrary one moving object cluster based on Example 3 of this invention. It is processing explanatory drawing of FIG. It is processing explanatory drawing of FIG. It is a schematic flowchart which shows the representative point determination process of the representative point tracking part 27 with respect to arbitrary one moving object cluster based on Example 4 of this invention. It is processing explanatory drawing of FIG. It is explanatory drawing of the problem of a prior art.

Explanation of symbols

10 Video camera 11 Road surface 12A, 12B, 131-133, 141-143, 16A, 16B, 30, 301-303 Moving object 12SA, 12SB, 16SA, 16SB, 30S, 301S, 303S Shadow 15 Frame 17A, 17B, 18A, 18B, 31, 311 to 313 Rectangle 20, 20A Moving object measurement device 21 Image memory 22 Background image generation unit 23 ID generation / extinction unit 24 Object map storage unit 25 Moving object tracking unit 26, 26A Average speed and miscalculation calculation unit 27 Representative Point tracking unit IN Input device OUT Display device Pa, Pb Line A1, B2 Entrance slit A2, B1 Exit slit K, Ka, Kb Actual block width F, Fa, Fb Frame number ΔFa, ΔFb Frame number δFa, δFb Frame number error V Average speed δV speed Measurement error RA1~RA3, RB1~RB3 geometric center of gravity

Claims (13)

In a moving object measurement method by image processing, which detects a moving object on a time-series image and measures a physical quantity of the moving object,
When the moving object moves in a direction away from the video camera that images the moving object, the substantially rear end of the moving object is used as the representative point, and when the moving object moves in a direction approaching the video camera, the moving object With the approximate front end as the representative point,
A moving object measuring method by image processing, characterized by measuring a physical quantity related to the representative point. In a moving object measurement method by image processing, which detects a moving object on a time-series image and measures a physical quantity of the moving object,
A relative position with respect to the area of the moving object is determined in advance as a representative point of the moving object,
A moving object measuring method by image processing, characterized by measuring a physical quantity related to the representative point.   3. The method of claim 2, wherein the area of the moving object is a rectangle inscribed by the moving object.   The relative position is substantially expressed by orthogonal coordinates (px · LX, py · LY) with one vertex of the rectangle as the origin, where LX and LY are the lengths of two adjacent sides of the rectangle. , Px and py are predetermined and 0 <px <1, 0 <px <1.   4. The method according to claim 3, wherein px and py are determined according to an angle of a motion vector of the moving object. In a moving object measurement method by image processing, which detects a moving object on a time-series image and measures a physical quantity of the moving object,
When it is determined that the moving object does not overlap with another moving object, the geometric center of gravity of the moving object is obtained as a representative point, and it is detected that the moving object overlaps with another moving object Is obtained based on the motion vector of the moving object as a representative point a point corresponding to the geometric center of gravity when there is no overlap,
A moving object measuring method by image processing, characterized by measuring a physical quantity related to the representative point. First and second lines are set in advance in the image frame, and an actual distance from the first line to the second line is measured in advance.
7. The time when the representative point moves from the first line to the second line is detected, and the average speed of the moving object is obtained as the physical quantity based on the time and the distance. The method as described in any one of these.   An error of the average speed is obtained and the error is output together with the average speed, and the error includes at least the one based on the frame rate and the one obtained by dividing the image frame into blocks and detecting the moving object in units of blocks. 8. The method of claim 7, wherein:   The error is substantially expressed as V · X / (Fb−Fa), where V is a measured value of the average velocity, and X corresponds to a time point when the representative point of the moving object passes through the first line. Average value of frame number error and frame number error corresponding to the time when the representative point passes the second line, (Fb−Fa) is the representative point moving from the first line to the second line The method according to claim 8, wherein the method is a measurement value of the number of frames required to perform the operation. In a moving object measuring apparatus by image processing that detects a moving object on a time-series image and measures a physical quantity of the moving object,
Setting means for setting a relative position of the moving object as a representative point of the moving object;
Processing means for processing the time-series image to obtain a physical quantity related to the representative point;
An output means for outputting the physical quantity;
A moving object measuring apparatus using image processing, characterized by comprising: The setting means is further for setting the first and second lines in the image frame and inputting an actual distance from the first line to the second line;
The apparatus according to claim 10, wherein the physical quantity is an average speed of the moving object obtained based on a time and a distance at which the representative point moves from the first line to the second line. The processing means divides the image frame into blocks, detects the moving object in units of blocks,
The processing means further determines an error of the average speed,
The output means outputs the error together with the average speed,
12. The apparatus according to claim 11, wherein the error includes a result of a frame rate and a result of detecting the moving object in units of blocks. In a moving object measuring apparatus by image processing that detects a moving object on a time-series image and measures a physical quantity of the moving object,
Determining means for determining whether or not the moving object overlaps with another moving object;
When the determination means makes a negative determination, the geometric center of gravity of the moving object is obtained as a representative point, and when the determination means makes an affirmative determination, there is no overlap based on the motion vector of the moving object. A processing means for obtaining a point corresponding to the geometric center of gravity as a representative point and obtaining a physical quantity related to the representative point;
Means for outputting the physical quantity;
A moving object measuring apparatus using image processing, characterized by comprising: