JP2988383B2 - Moving object detection system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving object detection system for picking up an image of a moving object such as a vehicle by an image pickup device such as a camera and selectively detecting an image of a specific portion on a screen thereof. The present invention relates to a moving object detection system capable of detecting an on-going vehicle with flexibility within a predetermined range and correcting a license plate or the like captured from an oblique angle into an image viewed straight from the front.

[0002]

2. Description of the Related Art Conventionally, there is an automatic vehicle detection technology based on images necessary for monitoring traffic violations on roads, measuring traffic volume, and collecting tolls on toll roads and parking lots. For this purpose, the right and left symmetry when the vehicle is viewed from the front and rear is used as an optimal clue, and there is also an automatic license plate recognition technology for identifying a plurality of vehicles.

However, in practice, it is difficult to obtain an image in which the license plate in front of and behind the vehicle is viewed straight from the front, and some consideration is required. At that time, it is important to install the camera on the side of the road where traffic is not obstructed and to correct the image taken at an oblique angle to the traveling direction by image processing technology. For this purpose, various detection systems for moving objects have been proposed. Have been.

FIG. 7 is a configuration diagram illustrating a first conventional example using a scale. In the first conventional example, a work 102 on which an L-shaped scale 101 with graduations at equal intervals is arranged is imaged by a CCD camera 103, and the obtained image information is corrected by an image processing unit 104 and used for inspection of a body surface or the like. Is an optical shape recognition device. Further, since the surface of the work 102 is imaged in an oblique direction, the image on the scale 101 is referred to by the image processing unit 104 to correct the distorted screen, and the same image information as when viewed straight from above vertically is formed. .

Therefore, by arranging the known scale 101 in the screen in advance, it becomes equivalent to the surface to be imaged being already measured, and the camera parameters are unambiguously determined by the distortion amount obtained only from the obtained screen. Has been decided. In other words, by giving a known scale to the surface of the work 102, the surface dimension, which is a determinant of the camera parameter, is set in advance on the screen to be imaged and the camera parameter is determined, and the distortion amount is corrected by the camera parameter. .

FIG. 8 is a configuration diagram illustrating a second conventional example using an optical sensor. In the second conventional example, a moment when an optical axis is blocked by a vehicle 122 entering on a road 121 is determined by an optical sensor 1.
23, and the front of the vehicle 122 is
4 is a license plate reading device which is used for detecting a license plate by taking an image with an image pickup device such as 4 and correcting the obtained image information by an image processing device 125. Therefore, the determinant of the camera parameter is the positional relationship between the camera 124 and the front of the vehicle 122, and is set in advance as the distance to the optical axis of the optical sensor 123.

[0007]

However, there are the following problems when trying to detect a moving vehicle at an arbitrary position using a conventional moving object detection system. In order to use a scale with a scale as in the first conventional example, it is necessary to arrange this scale on each vehicle. However, it is not realistic when imaging unspecified vehicles one after another. Furthermore, it is extremely difficult to fix it to a moving vehicle.

The use of the optical axis of the optical sensor as in the second conventional example can determine the positional relationship with an unspecified vehicle, but the positional relationship between the camera and the vehicle sometimes changes as the vehicle moves. If it is assumed that it changes every moment, even if the installation position and imaging direction of the camera with respect to the road can be set in advance,
Since this positional relationship cannot be specified, camera parameters prepared in advance cannot be used, and extra equipment for installing an optical sensor is required.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and has an advantage that a moving vehicle can be flexibly within a predetermined range when detecting a specific portion from a screen imaged by the camera. It is an object of the present invention to provide a moving object detection system that can detect and correct a license plate or the like captured from an oblique angle into an image viewed straight from the front.

[0010]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, a moving object detection system of the present invention captures an image of a moving object in an oblique direction by an image pickup device, stores an oblique image pickup screen, and stores the oblique image pickup screen. in the moving body detection system for detecting emmetropia image of the moving body from the rectangular imaging plane, and the distance value setting means for designating to generate distance values within a predetermined range to the mobile from the imaging device, the Quadratic in distance value
Generate camera parameters to uniquely determine the original coordinates
On the oblique imaging screen with this camera parameter.
That the image information of the movable body to process, and en-face image generating means for forming an en-face image with corrected surface of the moving object in the viewing direction of the imaging device, image storage means for storing an en-face images formed And a target area extracting means for extracting a symmetrical area in the stored stereoscopic image.

According to this detection system, a distance value within a predetermined range from the image pickup device to the moving body is designated by the distance value setting means, and the camera parameter at this distance value is set in the stereoscopic image generating means to set the distance value. The image detection is started by the instruction of the means, the image information of the moving body on the oblique imaging screen is read and processed, the surface of the moving body is formed into a stereoscopic image, stored in the image storage unit, and the A region having symmetry is extracted by the target region extracting means.

[0012] According to a second aspect of the present invention, there is provided a moving object detection system.
The distance value setting unit sets a plurality of distance values within a predetermined range, the stereoscopic image generation unit forms a stereoscopic image for each distance value, the stereoscopic image storage unit, each stereoscopic image, It is configured to store for each distance value.

According to this detection system, a plurality of distance values within a predetermined range are set by the distance value setting means, and a stereoscopic image is formed for each distance value by the stereoscopic image generating means. Each of the stereoscopic images is stored for each distance value by the standard image storage means.

According to a third aspect of the present invention, there is provided a moving object detection system.
Mobile traveling imaged obliquely stores oblique imaging screen by the imaging device, the moving body detection system for detecting emmetropia image of the moving body from the oblique imaging screen, the mobile from the imaging device Two-dimensional plane setting means for generating a distance value within a predetermined range up to and generating and specifying a camera parameter for uniquely determining two-dimensional coordinates at the distance value; and a two-dimensional plane setting means. By
On the oblique imaging screen using the camera parameters generated by
Processing means for processing image information of a moving object in the three-dimensional coordinate system, correcting a surface of the moving object in a line-of-sight direction of the image pickup apparatus , and forming a partial normal-view image generating means for forming a normal-view plane in a two-dimensional coordinate system; A partial emmetropic image storage means for storing the obtained emmetropic plane and a symmetry calculating means for calculating the symmetry in the stored emmetropic plane are provided.

According to this detection system, a distance value within a predetermined range from the image pickup device to the moving object is generated by the two-dimensional plane setting means, a camera parameter of two-dimensional coordinates at this distance value is designated, and a two-dimensional plane is specified. In accordance with an instruction from the setting unit, the camera parameters are set in the partial emmetropic image generating unit, image detection is started, image information of the moving body on the oblique imaging screen is read and processed, and the surface of the moving body is three-dimensionally read. In the coordinate system, the correction is made in the direction of the line of sight of the image pickup device, a two-dimensional coordinate system emmetropic plane is formed and stored in the partial emmetropic image storage means.

According to a fourth aspect of the present invention, there is provided a moving object detection system.
The two-dimensional plane setting unit sets a plurality of distance values within a predetermined range to generate a camera parameter group at each distance value, and the partial stereoscopic image generation unit sets a standard viewing plane group at each camera parameter. The partial emmetropic image storage means stores each emmetropic plane for each distance value, and selects a symmetric emmetropic plane by the symmetry calculating means.

According to this detection system, a plurality of distance values within a predetermined range are set in the two-dimensional plane setting means to generate a camera parameter group, and a normal viewing plane group in each camera parameter is formed in the partial normal viewing image generating means. And
Each emmetropic plane is stored in the partial emmetropic image storage means for each distance value, and a symmetric emmetropic plane is selected.

[0018]

Embodiments of the present invention will be described below with reference to the drawings. In addition, each part denoted by the same reference numeral as the conventional example is a part having the same function, and the detailed description is omitted. FIG. 1 is a configuration diagram illustrating a first embodiment according to the present invention. This embodiment includes a screen input device 1 for inputting a screen obtained by the camera 124 or the like in FIG. 8, a first storage device 2 for storing the input screen, and the like, and a screen in the first storage device 2. This is a moving object detection system including a first image processing device 3 for processing image data and obtaining necessary image information, and an output device 4 such as a display for displaying image information by the first image processing device 3.

The first storage device 2 stores an oblique imaging screen storage unit 21 for storing an oblique imaging screen introduced from the screen input device 1 and a group of normal-view images 34 introduced from the image processing device 3. And a normal-view image group storage unit 22 for performing the operation. The standard-view image group storage unit 22 stores the standard-view image group 34 for each distance value. Note that the oblique imaging screen is a screen in which the moving object is imaged from an oblique direction, and the group of stereoscopic images 34 is an image group in which the moving object is directly viewed from the front by correction.

The first image processing device 3 includes a distance value setting unit 31 for setting a distance value, a standard image generation unit 32 for generating a standard image 34, and a symmetric region for extracting a symmetric region. The target area extracting means 33 is provided. The distance value setting means 31 sets a plurality of distances from the imaging device to the moving object within a predetermined range .

In the stereoscopic image generating means 32, a distance value setting
The image information of the moving object on the oblique imaging screen is processed using the camera parameters formed in the step 31 to form a stereoscopic image 34 in which the surface of the moving object is corrected in the line-of-sight direction of the imaging device for each distance value. I do. The target area extracting means 33 extracts a symmetric area for each of the stereoscopic images 34.

Next, the operation of the first embodiment will be described. FIG. 2 is a diagram illustrating the relationship between two three-dimensional coordinate systems in FIG. The imaging device installation position θ is a three-dimensional fixed coordinate system θ provided on the imaging surface in the camera 124 in FIG.
The first origin of XYZ, the X axis is parallel to the road surface facing the road 121, the Y axis is orthogonal to the extension of the road surface of the road 121, and the Z axis is opposite to the vehicle 122 in FIG.
Are provided in parallel with.

Further, the three-dimensional coordinate system θ-XYZ is rotated by a tilt angle α around the Y axis, and rotated by a tilt angle β around the X axis which is the X ′ axis, and the first origin θ is set to its own origin. And a three-dimensional rotation coordinate system θ-X′Y′Z ′ directed toward the vehicle 122 with the line of sight of the camera 124, that is, the optical axis of the lens as the Z ′ axis.

Normally, a vehicle 122 traveling in the direction of a camera 124 along a road 121 has a three-dimensional fixed coordinate system θ since its front license plate is parallel to the XY axis plane.
It can be said that -XYZ is a space in which the license plate as the stereoscopic image 34 can be most easily digitized. On the other hand,
Since the X'Y 'axis plane is parallel to the imaging plane of the camera 124, the three-dimensional rotational coordinate system θ-X'Y'Z' is optimal for digitizing an oblique imaging screen with a license plate in the field of view. Space.

In the above, the oblique imaging screen of the vehicle 122 is input from the image input device 1 to the oblique imaging screen storage means 21, and a plurality of distances are set by the distance value setting means 31.
The plurality of distances are obtained in a predetermined range from the camera 124 to the moving object in correspondence with the three-dimensional fixed coordinate system θ-XYZ, and n distance values Z generated at predetermined intervals Dz
i (i = 1, 2,..., n).

For each distance value Zi, the minimum value is Zmin.
Assuming that the maximum value is Zmax, there is a relationship represented by the following expression 1, and the distance value Zi is introduced into the stereoscopic image generating means 32 and subjected to correction. Zmin ≦ Z1 <Z2 <... <Zn <Zmax...

The stereoscopic image generation means 32 introduces the oblique imaging screen from the oblique imaging screen storage means 21, corrects it, reverse-converts it into a stereoscopic image 34, and stores it in the stereoscopic image group storage means 22.

Here, as camera parameters, 2a is the horizontal angle of view of the camera 124 (the angle formed by the visual field along the IX axis), and 2b is the vertical angle of view (the angle formed by the visual field along the IY axis). , Vertical isx (pixels) × horizontal isy (pixels) is defined as the effective size of one screen of the original image in the digital image of the camera 124. The pixel positions of the digital image in the two-dimensional display coordinate system θ ″ -IXIY are defined as variables IX and IY.
Then, the oblique imaging screen of the two-dimensional display coordinate system θ ″ -IXIY can be inversely transformed into the stereoscopic image 34 of the three-dimensional fixed coordinate system θ-XYZ by the following Expression 2.

IX = ｛1+ (X cos α−Z sin α) / (X sin α cos β−Y sin β + Z cos α cos β) tana｝ isx / 2 IY = ｛1+ (X sin α sin β + Y cos β + Z cos α sin β) / (X sin α cos β-Y sinβ + Z cosα cosβ) tanb｝ isy / 2 Expression 2 The process of mapping from the three-dimensional coordinate system to the two-dimensional coordinate system will be described later in detail.

Of the parameters in Equation 2, the angles of view a and b and the effective size of the original image one screen arex and isy are specific to the optical system of the camera 124 and the photoelectric conversion circuit.
Also, the inclination angles α and β of the line of sight are fixed values depending on the installation state of the camera 124 as described above, and are all constants determined at the time of imaging.

Accordingly, if the distance value Zi is set as the variable Z, the perspective distortion of the oblique image pickup screen is corrected, and the variables X and Y, which are the pixel positions in the normal-view image 34, can be obtained. In other words, by giving a plurality of variables Z within a predetermined range, the surface of the license plate in the three-dimensional fixed coordinate system θ-XYZ is uniquely determined from the two-dimensional display coordinate system θ ″ -IX · IY according to Expression 2. The image is displayed as the stereoscopic image 34.

Here, in order to facilitate understanding of the above operation, a process of mapping from a three-dimensional coordinate system to a two-dimensional coordinate system will be described. The camera parameters required for will be described specifically. First, the three-dimensional fixed coordinate system θ-X
When the actual license plate in YZ is mapped on the oblique imaging screen in the three-dimensional rotation coordinate system θ-X′Y′Z ′, the following Expression 3 using the inclination angles α and β of the camera 124 is used. Coordinate transformation can be performed.

[0033]

(Equation 1)

FIG. 3 is a diagram for explaining the relationship between the three-dimensional rotation coordinate system and the two-dimensional display coordinate system in FIG. The three-dimensional rotating coordinate system θ-X′Y′Z ′ is translated along the Z ′ axis, and is moved to the second origin θ ′ which is separated from the first origin θ by a distance f.
A dimensional movement coordinate system θ′-x′y′z ′ is provided.

That is, the conversion to the three-dimensional moving coordinate system θ'-x'y'z 'according to the following equation 4 is performed by connecting the image of the vehicle 122 that has reached the camera 124 to the image pickup plane at the focal length f. Are equivalent. [X ′, y ′, z ′] = [X ′ × (f / Z ′), Y ′ × (f / Z ′), 0] Equation 4

Further, the three-dimensional moving coordinate system θ'-x'y '
the two-dimensional display coordinate system θ ″ − on the x′y ′ axis plane of z ′.
IX · IY, and the two-dimensional display coordinate system θ ″ −IX ·
IY has a third origin θ ″ in the third quadrant of the x′y ′ axis plane. Here, if the angles of view 2a and 2b and the effective size of one original image isx and isy are used, the resolution of the pixel (g
_x , g _y ) is the following Expression 5, and Expression 6 is obtained. g _x = 2ftan a / isx g _y = 2ftan b / isy Equation 5

[0037]

(Equation 2)

Further, when the above equations 4 and 5 are substituted into equation 6, the following equation 7 is obtained. IX = (1 + X ′ / Z′tan a) isx / 2 IY = (1 + Y ′ / Z′tan b) isy / 2 Equation 7

From the above, the mapping from the three-dimensional fixed coordinate system θ-XYZ to the two-dimensional display coordinate system θ ″ -IX · IY is expressed by the following equation (3).
Is substituted into Equation 7 to obtain Equation 2. In addition,
The coordinates (isx / 2, isy / 2) are in the three-dimensional moving coordinate system θ'-x'y '
It can be said that this is an offset value of the third origin θ ″ at z ′.

FIG. 4 is a diagram for explaining the operation of the target area extracting means in FIG. In the target area extraction means 33,
Read the emmetropic image 34 from the emmetropic image group storage means 22,
On this emmetropic image 34, a window 35 shown in FIG. At the center of window 35 is the axis of symmetry t
Is selected, all the left and right pixel pairs equidistant from the symmetry axis t are selected, and the luminance difference between these pixel pairs is calculated and compared with a predetermined reference value δT. In addition, the number of pixel pairs having a luminance difference smaller than the reference value δT is calculated to obtain Npair.
And verify the left-right symmetry.

However, the vehicle 1 can be obtained by simply _{obtaining Npair.}
If it is a left-right symmetric area such as the background of the road 22 or the road 121, the image is included in the image of the vehicle 122. It is necessary to extract by narrowing down to a region that includes a large amount and prevent _Npair from becoming too large.

Therefore, Npair is obtained, and the edge image 34 is subjected to edge detection and binarization to form a binary edge image 36, which has a luminance value of "1" existing in the same window 35. The number of pixels is N _edge , and the following equation 8
To determine the degree of symmetry S for measuring the degree of symmetry. S = N _pair × N _edge Equation 8

Subsequently, the position of the symmetry axis t and the window 35
Are varied, and all the degrees of symmetry S are obtained for the plurality of obtained stereoscopic images 34 as described above, and each of the calculated degrees of symmetry [S] is compared with each other. The region in the stereoscopic image 34 corresponding to the window 35 having the largest degree of symmetry S is supplied to the output device 4 as a detection result.

FIG. 5 is a configuration diagram illustrating a second embodiment according to the present invention. The second embodiment corresponds to the first embodiment in FIG.
The second embodiment is the same as the first embodiment except that the storage device 2 is replaced with the second storage device 5 and the first image processing device 3 is replaced with the second image processing device 6.

The second storage device 5 comprises the above-described oblique imaging screen storage means 21 and a partial stereoscopic image group storage means 52 for storing the processed partial stereoscopic image group 34. In the partial emmetropic image group storage means 52, the formed partially emmetropic image 3
Four groups are stored for each distance value.

The second image processing device 6 includes a two-dimensional plane setting means 6 for generating camera parameters of a two-dimensional coordinate system.
1, a partial stereoscopic image generating means 62 for forming a standard viewing plane in a two-dimensional coordinate system, and a symmetry calculating means 63 for calculating symmetry in the standard viewing plane.

The two-dimensional plane setting means 61 generates respective camera parameters for a plurality of distance values within a predetermined range .

The partial emmetropic image group generating means 62 includes a two-dimensional flat
The camera parameters generated by the plane setting means 61 are
The image information of the moving object on the oblique imaging screen is corrected using the image data, and a specific plane of the moving object is rotated in three-dimensional coordinates in the direction of the line of sight of the imaging device to form a normal viewing plane in a two-dimensional coordinate system.

FIG. 6 is a diagram for explaining the operation of the two-dimensional plane setting means in FIG. 5. FIG. 6 (a) shows a normal view plane, and FIG. 6 (b) shows FIG. Viewed from above along. The emmetropic plane has a lattice shape having a width of 2B (half width B) in the X-axis direction and a height of C in the Z-axis direction, and is provided in parallel with the XY-axis plane of the three-dimensional fixed coordinate system θ-XYZ. Each intersection of the grid is a pixel position, and has a center axis A perpendicular to the XZ-axis plane on a line segment at the center of the grid. Usually, as in the first embodiment, the license plate of the vehicle 122 is parallel to the normal viewing plane.

In the two-dimensional plane setting means 61, the emmetropic plane 6
Camera parameters for numerically determining 4 are set in advance. That is, the coordinates of the center axis A are (Xa, Za) on the XZ axis plane, and the coordinates (Xa, Za), the half width B, and the height C are within a predetermined range including the maximum value and the minimum value. It is changed little by little at intervals, and predetermined as a plurality of camera parameter sets. Subsequently, a plurality of camera parameter sets are generated and transferred to the partial stereoscopic image generation means 62, and the activation of the partial stereoscopic image generation means 62 is instructed for each transfer.

The partial emmetropic image generation means 62 uses the given camera parameter set [XaZaBC], sets the variable Z in the above equation 2 to Z = Za,
In the same manner as in the embodiment, the two-dimensional display coordinate system θ ″ -IX · IY is inversely converted to the three-dimensional fixed coordinate system θ-XYZ to determine the respective XY coordinates. Form a standard viewing plane consisting of a group of pixels to be positioned,
A plurality of emmetropic plane groups are stored in the partial emmetropic image group storage means 52 for each camera parameter set.

The symmetry calculating means 63 reads out each emmetropic plane from the partial emmetropic image group storage means 52, and sets the central axis A as the symmetry axis t, the emmetropic plane as the window 35, and the symmetric region extracting means 33 of the first embodiment. In the same manner as in the above, the degree of symmetry S group [S] is calculated. Further, the calculated [S] is compared with each other, and the emmetropic plane having the maximum degree of symmetry S is supplied to the output device 4 as a detection result.

The present invention is not limited to the above-described embodiment. For example, the present invention may be applied not only to the front or front license plate of the vehicle 122 but also to the rear or rear license plate. Good.

The two-dimensional coordinate system and the emmetropic plane are not necessarily orthogonal or perpendicular to the road 121, but may be orthogonal to the running direction of the vehicle 122 or parallel to the license plate surface. Of course, various changes can be made without departing from the gist.

[0055]

As described above, the moving object detection system according to the present invention has the following effects. First, in order to generate a plurality of distances within a predetermined range, or to generate a plurality of camera parameters corresponding to these distances and prepare them in advance, equipment such as an optical sensor that indicates the moment of imaging is used. No equipment for measuring distance such as a range sensor is required beside the road.

Secondly, since a mobile object is detected by forming a stereoscopic image or a stereoscopic plane at a plurality of distances from the camera to the mobile object, it is not necessary to attach a scale for dimension display to the mobile object itself. In the case where an unspecified vehicle is successively imaged, it can be said that even if the vehicle is in progress, it can be easily detected and is realistic.

Third, in order to extract a region having the greatest symmetry from a plurality of stereoscopic images obtained from a plurality of distances or camera parameters, an unexpected subtle displacement of the moving position of the moving object is detected. Also allows flexible detection.

Therefore, when detecting a specific portion from a screen imaged by a camera, a moving vehicle is detected with flexibility within a predetermined range, and a license plate or the like captured from an oblique angle is detected in front of the vehicle. This makes it possible to provide a moving object detection system that can correct an image viewed from the front.

[Brief description of the drawings]

FIG. 1 is a configuration diagram illustrating a first embodiment according to the present invention.

FIG. 2 is a diagram illustrating a relationship between two three-dimensional coordinate systems in FIG.

FIG. 3 is a diagram illustrating a relationship between a three-dimensional rotation coordinate system and a two-dimensional display coordinate system in FIG. 2;

FIG. 4 is a diagram illustrating an operation of a target area extracting unit in FIG. 1;

FIG. 5 is a configuration diagram illustrating a second embodiment according to the present invention.

FIG. 6 is a diagram illustrating the operation of a two-dimensional plane setting unit in FIG. 2;

FIG. 7 is a configuration diagram illustrating a first conventional example using a scale.

FIG. 8 is a configuration diagram illustrating a second conventional example using an optical sensor.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 screen input device 2 first storage device 3 first image processing device 4 output device 21 oblique imaging screen storage means 22 normal view image group storage means 31 distance value setting means 32 normal view image generation means 33 target area detection means 5 second storage Apparatus 6 Second image processing apparatus 52 Partial emmetropic image group storage means 61 Two-dimensional plane setting means 62 Partial emmetropic image generating means 63 Symmetry calculating means

Claims (4)

(57) [Claims] 1. A moving object detection system for imaging a moving object that travels in an oblique direction by an image pickup device and storing an oblique imaging screen, and detecting a stereoscopic image of the moving object from the oblique imaging screen. a distance value setting means for designating to generate distance values within a predetermined range to the mobile from the imaging device, to uniquely determine the two-dimensional coordinates in the distance value
Generate camera parameters for this camera parameter
Processing the image information of the moving object on the oblique imaging screen,
A stereoscopic image generating means for forming a stereoscopic image in which the surface of the moving object is corrected in the direction of the line of sight of the imaging device ; an image storage means for storing the formed stereoscopic image; and a symmetry in the stored stereoscopic image. And a target area extracting means for extracting a certain area. 2. The distance value setting means sets a plurality of distance values within a predetermined range, the stereoscopic image generation means forms a stereoscopic image for each distance value, and the stereoscopic image storage means The moving object detection system according to claim 1, wherein each emmetropic image is stored for each distance value. 3. A moving object detection system for capturing an oblique imaging screen by imaging an advancing moving object from an oblique direction with an imager, and detecting a stereoscopic image of the moving object from the oblique imaging screen. A distance value for generating a distance value within a predetermined range from the image pickup device to the moving body and generating and specifying a camera parameter for uniquely determining two-dimensional coordinates at this distance value.
Two-dimensional plane setting means , and camera parameters generated by the two-dimensional plane setting means .
Image information of the moving object on the oblique imaging screen
Processing , correcting the surface of the moving body in the direction of the line of sight of the image pickup device in the three-dimensional coordinate system, and forming a partial stereoscopic image generating means for forming a normal viewing plane in the two-dimensional coordinate system; and storing the formed normal viewing plane. A moving object detection system, comprising: a partial emmetropic image storage unit for calculating a symmetry in a stored emmetropic plane; 4. A plurality of distance values within a predetermined range are set by the two-dimensional plane setting means, a camera parameter group at each distance value is designated, and each camera parameter is set by the partial stereoscopic image generation means. 4. The moving object detection system according to claim 3, wherein the emmetropic plane group is formed, and the partial emmetropic image storage means stores each emmetropic plane, and the symmetry calculating means selects a symmetric emmetropic plane. .