JP4661829B2 - Image data conversion device and camera device provided with the same - Google Patents

Description

The present invention relates to an image data conversion apparatus and a camera apparatus including the same, and more specifically, when an image including spherical distortion imaged through a fisheye lens is input to correct distortion and viewed with human eyes. The present invention relates to image correction that is converted into an image with a little discomfort and output.

2. Description of the Related Art Conventionally, a configuration shown in a block diagram of FIG. 11 is generally used as a camera device that captures an image through a fisheye lens or an image data conversion device that inputs an image captured by the camera device and corrects and outputs the image. In FIG. 11, the camera device 80 sequentially selects a fish-eye lens 81, an image sensor 82 that images light input through the fish-eye lens 81 and converts it into an electrical signal, and sequentially specifies pixel readout addresses for the image sensor 82. An image processing unit 83 that takes in pixel signals and outputs them as digital image signals.

The image data conversion device 90 also converts a digital input terminal 91 for inputting a digital image signal output from the camera device 80, an analog input terminal 92 for inputting an analog image signal, and the analog image signal into a digital signal. An A / D converter 93, an image memory unit 94 for storing the converted digital signal or image data input from the digital input terminal 91 for each frame, and a coordinate position of pixel data of the stored image data Is converted by a predetermined method, and image conversion means 96 for instructing pixel interpolation of the converted image data;
An image interpolation unit 95 that interpolates pixels of the converted image output from the image memory unit 94 in accordance with an instruction of the image conversion unit 96; a D / A converter 97 that converts the image data subjected to the pixel interpolation into an analog signal; And an analog video output terminal 98 for outputting the analog signal.

Each input terminal is connected to an analog or digital video signal captured through a fisheye lens. Any one of these input signals is input to the image memory unit 94 to perform image data conversion.

The conversion methods performed by the image conversion means 96 are roughly classified into two methods. FIG. 6 is an image data conversion model diagram for explaining one conversion method, and FIG.
Is a perspective view showing a relationship between an input image and an output image, and FIG. 6B is a top view thereof.

With reference to FIG. 6A, the positional relationship of the coordinates to be converted will be described first. The coordinate system is set as follows. First, as a space to indicate the position of the subject, the position of the fisheye lens is the origin,
Consider an (x, y, z) coordinate space in which the direction of the optical axis is the z-axis. Further, azimuth angle (ψ) and zenith angle (θ) are considered as parameters indicating the subject position viewed from the origin.

It is assumed that a subject photographed with a fisheye lens is positioned on a hemispherical surface having a radius of 1 because the position on the image sensor is determined by the angle viewed from the fisheye lens. This hemisphere is called a virtual object plane. And for the planar image you want to find as a result,
FIG. 6A shows the (u, v) coordinate system, and the center (origin) of the (u, v) coordinate system is (
x, y, z) It is assumed that it is located at a distance of 1 from the origin of the coordinate system and is in contact with the above-described hemispherical surface which is a virtual object plane. The plane represented by the (u, v) coordinate system is associated with the pixels of the display image.

The image plane (fisheye image) obtained through the fisheye lens is (p, q) as shown in FIG.
It shall be expressed in a coordinate system. The (p, q) coordinate system is in a positional relationship parallel to the (x, y) plane, and its origin is on the z-axis. Since the image circle diameter varies depending on the size of the image sensor and the focal length of the fisheye lens at the position on the imaging surface, the radius of the image on which the subject at a position 90 degrees (z = 0) from the front of the lens is projected. Assume that a fisheye image is reflected in a circle of 1.

FIG. 10 shows an example of this fisheye image. This image shows an outdoor straight road, electric wires parallel to the road, street lamps built vertically on the side of the road, houses along the road, and the like. The image is distorted from the center of the optical axis, that is, from the center of the image to the periphery, and the top, bottom, left and right of the screen are curved.

The image conversion is obtained by calculating which position on the imaging surface (p, q coordinate) the point (u, v coordinate) on the plane image is calculated and referring to the luminance information of the point. The data of the plane image desired to be obtained can be generated (see, for example, Patent Document 1).

Note that information on the viewpoint (ψ, θ, α) and the magnification (zoom magnification) that is the size of the planar image is
It is set in advance as data for calculation.

This method is characterized by relatively little image distortion. In this manner, FIG.
FIG. 7 shows an image obtained as a result of converting the above image.

On the other hand, FIG. 9 is an image data conversion model diagram for explaining the other conversion method. FIG. 9A is a perspective view showing a relationship between an input image and an output image, and FIG. 9B is a top view thereof. Show.

In this conversion method, image data on a circular surface is converted to a cylindrical surface. FIG. 9A is a diagram showing a relationship between an arbitrary point P0 on the hemispherical surface O virtually imagining the entire circumferential surface photographed by a fisheye lens and a point P2 on the cylindrical surface C which is a surface to be converted. . In the xyz coordinate system, the hemispherical surface O has a bottom circle (a circle constituting the edge of the hemispherical surface) located on the xy plane with the origin at the center.
The z axis coincides with the optical axis direction of the fisheye lens.

The cylindrical surface C is a cylinder having a radius R with the z axis as the central axis. That is, the cylindrical surface C can be expressed as a curved surface in an rψz cylindrical coordinate system with R as a constant. Here, a point on the cylindrical surface C is represented by (R, ψ, z) as a point on the rψz cylindrical coordinate system. However, ψ is an angle formed by a line segment formed by the intersection point of the perpendicular line drawn on the xy plane and the xy plane and the origin and the x axis. A point where a straight line obtained by extending a line connecting an arbitrary point P0 on the hemisphere O from the origin intersects the cylindrical surface C is P2. Note that the angle formed by the line connecting the origin to the point P0 and the z axis is the angle of view θ related to the point P0.
It becomes.

The position of a certain pixel P1 in the image picked up by the fisheye lens is a position at an image height h from the origin, and an angle formed by a line segment connecting the origin and P1 and the X axis is ψ. h is expressed by the above equation 1) and is determined by the angle of view θ.

Therefore, the image conversion in this system is to convert P1 to P2. The specific calculation will be described below. First, consider the case where the parameters are θ and ψ. In this case, the coordinates of P1 on the XY plane are (h · cos ψ, h · sin ψ), and if h is eliminated from the above equation (1), (2f · tan (θ / 2) · cos ψ, 2f · tan (θ / 2) ) ・ Sin
ψ).

On the other hand, the coordinates of P2 are such that R is a constant and z has a relationship of z = R / tan .theta.
(ψ, R / tan θ). Therefore, by moving the parameters θ and ψ, P1 on the circular image on the XY plane can be specified, and a point on the rψz cylindrical coordinate system, that is, the cylindrical surface C
Since the upper point can also be specified, the image data can be converted in accordance with the expressions indicating P1 and P2 (see, for example, Patent Document 2).
. However, since the z coordinate of P2 becomes infinite when θ = 0, in reality, the value of θ is greater than a certain value.

FIG. 8 shows an image obtained as a result of image conversion of the image of FIG. 10 by this method. However, FIG.
Since 0 is an angle of view of 180 degrees, only the portion corresponding to this angle of view is also converted in FIG. The feature of this method is that the left and right angle of view is wide, and a maximum of 360 degrees can be displayed.

However, the method described with reference to FIG. 6 can be converted into a relatively easy-to-view image, but has a practical limitation on the angle of view because it is linearly converted from the optical axis origin to P (u, v) plane coordinates. Images near the top, bottom, left, and right edges on a plane cannot be displayed. In addition, an infinite plane is required to display all of them. Theoretically, it cannot handle an angle of view of 180 degrees or more.

Furthermore, as shown in FIG. 7, an object that is long in the Z direction (the depth direction of the image), for example, the ends of a house standing on the left and right sides of the road is extremely enlarged, and the human eyes feel uncomfortable.

On the other hand, the method of FIG. 9 has the same angle of view restriction in the vertical direction as in the method of FIG. 6, but there is no restriction in the left-right direction, and a complete range can be displayed.

However, as shown in FIG. 8, the straight line in the X direction (left and right direction of the screen) and the straight line in the Z direction are all curved, and the image is very difficult to see.

Further, as described above, each method has advantages and disadvantages, and it is necessary to select an image data conversion device according to any image data conversion method according to the application of the camera device. For this reason, when the image data conversion device is built in the camera device, two types of camera devices corresponding to the two image data conversion methods are produced, and management such as parts procurement, production management, inventory management, and sales management is performed. There was a problem that the business cost increased and the management cost increased as compared with the case of one model.
Japanese Patent No. 3126955 (page 3, FIG. 1) Japanese Patent No. 3025255 (page 4, FIG. 2)

The present invention solves the above-described problems, converts an image captured through a fisheye lens to a wide angle as much as possible, and eliminates straight line distortion between the Y-axis and Z-axis, and straight line distortion in the X-axis direction. An object of the present invention is to provide an image data conversion device that converts images into images that are easy to see for humans and to reduce the management cost of a camera device incorporating the image data conversion device.

In order to solve the above-described problems, the present invention inputs an image of a subject including spherical distortion captured using a fisheye lens as an input image signal, and performs image conversion so as to remove the spherical distortion from the input image signal. In the image data conversion device provided with the image conversion means for outputting the output image as a signal,
The image conversion means converts the entire input image so that the linear object directed toward the optical axis of the fisheye lens and the vertical object are substantially linear in the output image,
In addition, the subject that is straight in the horizontal direction is substantially linear in the central portion of the output image, and the left and right portions that are connected to the central portion and the left and right, and the central portion and the left portion of the output image. And the entire input image is converted so as to be bent near the boundary of the right portion.

Further, the image conversion means is configured based on a virtual image conversion model,
The image conversion model includes the input image formed of a circular plane, a hemispherical virtual object surface on which a subject is displayed corresponding to the fisheye lens, a correction surface for correcting the spherical distortion, and a display surface for image conversion. The virtual optical axis corresponding to the optical axis of the fisheye lens is inserted through the center of these surfaces at a right angle from the optical axis origin located at the center of the input image, and the correction surface is disposed at the center. The center plane is composed of a left side surface and a right side surface respectively arranged on the left and right sides of the center plane, and the left side surface and the right side surface are bent toward the optical axis origin,
The image on the virtual object surface projected onto the correction surface in the virtual optical axis direction is projected onto the display surface by an orthogonal projection method, and the projected image on the display surface is used as the output image. .

The image conversion means is configured based on a virtual conversion model,
The conversion model includes the input image composed of a circular plane, a hemispherical virtual object surface on which a subject is projected corresponding to the fisheye lens, a correction surface for correcting the spherical distortion, and a display surface for image conversion. The virtual optical axis corresponding to the optical axis of the fisheye lens is inserted at a right angle from the origin of the optical axis located at the center of the input image, and the correction surface is
Consists of a central surface disposed in the center and left and right side surfaces disposed on the left and right of the central surface, the left and right side surfaces are bent toward the optical axis origin,
The image on the virtual object surface projected onto the correction surface in the virtual optical axis direction is projected onto the display surface by an orthogonal projection method, and the projected image on the display surface is used as the output image. .

Further, the conversion model forming the correction surface is such that the horizontal and vertical coordinates of the display surface are x, y, and the three-dimensional coordinates of the correction surface are x, y, z. z is

Calculate with the following formula.

Furthermore, the image data conversion device is provided inside the camera device, and the captured image is converted by the image data conversion device and output.

By using the above means, according to the image data conversion apparatus of the present invention,
The invention according to claim 1 can convert an image captured by a vehicle camera into an image that is easy for the driver to see. Specifically, the displayed angle of view is wide and a straight line of an actual landscape is displayed linearly when displayed through the image data converter.

In the present invention, there is no bending distortion in the vertical direction and the depth direction of the optical axis, and there is almost no horizontal bending distortion at the center of the driver's field of view, that is, around the optical axis and right and left of the screen. The image is less uncomfortable for the person.

The invention according to claim 2
In the image conversion model, the correction surface is composed of a center surface and left and right side surfaces, and is arranged so as to surround the virtual object surface, so that the displayed angle of view is wide, and the correction surface is composed of a plurality of surfaces, Since the horizontal direction on the correction surface has a configuration in which linear portions on each surface are connected, the horizontal distortion can be reduced and the image can be converted into an easily viewable image.

The invention according to claim 3 is:
In the image conversion model, the expression representing the correction surface can be easily configured, and the image conversion means can be easily configured.

According to a fourth aspect of the present invention, an image data conversion device is provided in the camera device so that one type of image data conversion can be performed without producing two types of camera devices incorporating different types of image data conversion devices. Since only one type of camera device incorporating the device needs to be produced, the management cost of the camera device can be reduced.

  DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail as examples based on the attached drawings.

1A is a perspective view of an image data conversion model showing a virtual relationship between an input image and an output image, and FIG. 1B is a top view thereof. In the conversion method according to the present invention, the input image 1 which is an image on a circular surface is once converted onto a correction surface 3 (three-dimensional) having a predetermined shape,
The image on the correction surface 3 is projected as a target output image on the rectangular display surface 4 by an orthogonal projection method. Therefore, arbitrary image conversion is performed by changing the shape of the correction surface 3 and actually the conversion formula.

As described in the background art, it is assumed that the subject photographed with the fisheye lens is positioned on the hemispherical surface 2 having the radius 1 because the position on the image sensor is determined by the angle viewed from the fisheye lens. It is assumed that the input image 1 that is an image of the subject is formed in a circle having a radius of 1, specifically on an imaging surface of an imaging element (not shown). Note that this hemispherical surface is referred to as a virtual object plane 2. The center of the circular input image 1 is the optical axis origin: O corresponding to the origin of the lens used for imaging, and an imaginary line extending perpendicularly from this point is the optical axis 5.

As shown in FIG. 1A, the correction surface 3 includes a central surface 3a positioned in front of the optical axis 5 and an optical axis 5.
The right side surface 3b disposed on the right side in the traveling direction of the optical axis 5 and the left side surface 3c disposed on the left side in the traveling direction of the optical axis 5.
It consists of and. The center surface 3 a of the correction surface 3 and the display surface 4 are disposed perpendicular to the optical axis 5, and the correction surface 3 is disposed so as to wrap around the virtual object surface 2. Further, the center point of the correction surface 3 and the center point of the display surface 4 are positions penetrated by the optical axis from the optical axis origin. Further, the correction surface 3 and the display surface 4 are linear in the vertical direction, and the display surface 4 is also linear in the horizontal direction.

Therefore, the vertical (Y) direction of the image of the correction surface 3 on which the image from the virtual object plane 2 is copied is horizontal (X
) There is no distortion that the vertical straight line is curved no matter where the direction is. Further, on the horizontal (X) direction line passing through the center of the correction surface, there is no distortion such that the horizontal straight line is curved.

In the lateral direction of the correction surface, distortion occurs where the lateral straight line is bent at a place other than on the lateral line passing through the center of the correction surface.

FIG. 2 is an image showing an example in which the image of FIG. 10 taken through a fisheye lens is corrected according to the present invention. In the entire range of FIG. 2, the vertical straight line in the actual landscape is converted as a straight line, such as the center line of the road toward the center of the image (back of the optical axis), the side line, the electric wire, the ridge line of the building, etc. There is no distortion, and it is straight according to the actual scenery. Also, the vicinity of the center in the left-right direction of the image is linear as in the actual landscape. Further, although the horizontal straight lines near the left and right sides of the image are inclined, they are corrected to be substantially linear. It should be noted that bending distortion occurs at a portion where the vicinity of the center in the left-right direction of the image and the horizontal straight line near the left-right direction are gently bent and connected.

This is because, as described above, the correction surface 3 is composed of the center surface 3a, the right side surface 3b, and the left side surface 3c, which are smoothly connected. Looking at each surface itself, it is a flat shape that gently curves in the optical axis direction, and the connecting portions of these surfaces are bent. Therefore, the corrected image is also converted with this characteristic.

The manner in which the horizontal straight line is distorted will be described with reference to the image of FIG. 3 which has been subjected to image conversion according to the present invention.
As in the other images, roads, electric wires, and ridge lines of buildings are linear in the depth direction of the optical axis (direction toward the center of the image).

In FIG. 3, a plurality of curves composed of white lines connected from the left end to the right end represent the linear distortion in the horizontal direction. That is, when a plurality of straight lines exist in the left-right direction in an actual scene, it shows how the straight lines are converted by the present invention. In FIG. 3, it is almost a straight line near the center of the image, and it is almost a straight line although it is inclined at the left and right ends of the image.

Although not shown in the drawings, the image data conversion device according to the present invention is used when, for example, an image captured by a front or rear camera provided with a fisheye lens is displayed on a display device such as a car navigation system. . In this case, what is important is not to reproduce the image accurately but to make the image easy to see for the driver.

An image that is easy to see for the driver is an image that has a wide angle of view and that accurately reproduces the perspective, that is, a straight line of an actual landscape is straight even when it is displayed through the camera. Is to be expressed. Therefore, it can be said that image conversion with as little bending distortion as possible is suitable for such applications. In the present invention, as described with reference to FIGS. 2 and 3, there is no bending distortion in the vertical direction and the depth direction of the optical axis, and the center of the driver's field of view, that is, the periphery of the optical axis and the left and right sides of the screen. In the vicinity of the edge, although the horizontal straight line of the landscape is inclined and converted, there is almost no horizontal bending distortion except in two places. For this reason, the image is less uncomfortable for the driver.

FIG. 4 is an explanatory diagram for explaining a specific example of the image data conversion model according to the present invention, and is basically the same as the diagram described in FIG. For reference, the image pickup surface 6 of the image pickup device is shown by a dotted line. In general, a circular imaging device is not used in a camera, and an imaging device having an aspect ratio of 3: 4 is used, and an image output signal output from the camera is also an image of this ratio.

The image data converter itself converts the input image and outputs it. As described with reference to FIG. 11 of the background art, the output image, that is, the pixel position P (x0, y0) on the display surface 4 and The relationship with the pixel position I (x3, y3) of the input image 1 is previously tabulated, and the input I (x3, y3) pixel is output to the position of P (x0, y0) in an actual apparatus. I am doing so.

Accordingly, in FIG. 4, the pixel position P (x0, y0) on the display surface 4 is temporarily set to the pixel position H on the correction surface 3.
(X1, y1, z1), and the coordinates are converted to polar coordinates K (
r2, φ2, θ2), and finally converted to the pixel position I (x3, y3) of the input image 1. This conversion is calculated at all pixel positions P (x0, y0) on the display surface 4, and the calculated result is stored in a storage element such as a ROM (hereinafter referred to as conversion ROM). The conversion ROM inputs x0 and y0 to the address, and is stored in the location indicated by the address value. The contents of the data output by accessing this address are x3 and y.
It is configured to have a value of 3.

Therefore, for example, in the image data conversion apparatus of FIG.
All pixels are stored while designating the pixel position I (x3, y3). Then, in the image conversion means 96 having the conversion ROM for performing the image data conversion according to the present invention described above, the output data (address value) of the conversion ROM is converted into the image memory while sequentially specifying x0 and y0 as the addresses of the conversion ROM. The pixel data of the image memory obtained as a result is used as the pixel data of the pixel position P (x0, y0) on the display surface 4.

In this conversion, x3 and y3 calculated for x0 and y0 are real numbers including decimal numbers, and cannot be used as they are as addresses of the image memory of the image memory unit 94. Accordingly, using the calculated address of the integer part of x3, y3, the image data of the image memory unit 94 of this address and its peripheral address is read, and these image data are calculated in the decimal part of the calculated x3, y3. Image correction such as addition and distribution is performed by the image interpolation unit 95 in correspondence with the address. Note that the image interpolation method is not limited to this, and any method may be used, and only the image data indicated by the addresses of the calculated integer parts of x3 and y3 may be used without performing image interpolation.

Next, the pixel position P (x0, y0) on the display surface 4 and the input image 1 are obtained by using FIG. 4 and Equations 1 to 4 in FIG. A detailed description will be given of the relationship with the pixel position I (x3, y3).

In FIG. 4, the coordinate position of the optical axis origin of the display surface 4 is set as the display origin P (0, 0), the right end in the x direction passing through this origin is P (1, 0), and the left end is P (-1, 0). ). Also, let the upper end in the y direction passing through the origin be P (0, −0.75) and the lower end be P (0, 0.75). Note that −0.75 to 0.75 in the y direction represents a ratio in the vertical direction when the x direction is set to −1 to 1, and in this case, the screen size is horizontal 4: vertical 3 (horizontal 1). : Vertical 0.75). Therefore, as shown in Equation 1 in FIG. 5, the pixel position of the display coordinate P (x0, y0) in the plane orthogonal coordinate system is defined by x0, y0.

In the same way, the correction surface 3 is represented by H (x1, y1, z1) of three-dimensional orthogonal system coordinates.
X1 and y1 are the same values as x0 and y0. Z1 indicates the position in the optical axis direction, and is calculated as a function of x0 using the equation shown in Equation 2.

Next, using the calculated x1, y1, and z1, polar coordinates K (r2, φ2, θ2) of the pixels shown on the virtual object plane 2 are calculated using Equation 3. As shown in FIG. 4, here, the size of the logical radius of the circular input image is calculated with r2 = 1. Φ2 represents an angle (azimuth angle) counterclockwise from the x-axis of the input image, and θ2 represents an angle (zenith angle) between the radius line at φ2 and the optical axis z. .

Next, using Equation 3 for inputting the calculated φ2 and θ2, the position of the pixel of the subject shown on the virtual object plane 2 is converted into input image coordinates I (x3, y3) that are plane orthogonal coordinates. To do.
This is the position of the imaged pixel. Actually, however, the input image is made to fit in the aspect ratio of the image sensor, for example, horizontal 4: vertical 3 (horizontal 1: vertical 0.75). This is the imaging coordinates I ′ (x4, y
4) and is input to the image data converter as an image signal including a range where light does not reach during imaging.

As described above, in the present invention, there is no bending distortion in the vertical direction and the depth direction of the optical axis, and the center of the driver's visual field, that is, the periphery of the optical axis center and the vicinity of the left and right sides of the screen are in the lateral direction. Since there is almost no curvature distortion, the image is less uncomfortable for the driver.

Furthermore, in the image conversion model according to the present invention, the correction surface is composed of the center surface and the left and right side surfaces, and is arranged so as to surround the virtual object surface, so that the displayed angle of view is wide and there are a plurality of correction surfaces. The horizontal direction on the correction surface is a configuration in which linear portions on each surface are connected to each other, so that the horizontal distortion can be reduced and the image can be converted into an easy-to-see image. In addition, the expression representing the correction surface can be easily configured, and the image conversion means can be easily configured.

Further, by providing the image data conversion device described in the embodiment in the camera device 80 of FIG. 11, for example, a wide angle can be obtained without producing two types of camera devices incorporating different types of image data conversion devices. It is only necessary to produce one type of camera device incorporating the image data conversion device according to the present invention, which has the characteristic that the image distortion is small, and the management cost of the camera device can be reduced.

In this embodiment, the correction surface is described as a combination of three surfaces. However, the present invention is not limited to this. For example, one surface is arranged in front of the optical axis and two surfaces are arranged on the left and right sides, for a total of five surfaces. A plurality of surfaces may be combined, for example, correction may be performed with. In this case, since the virtual object plane can be further wrapped, it is possible to deal with a wider-angle input image. Further, the area ratio of the plurality of surfaces constituting the correction surface may be arbitrarily changed according to the application of the image data conversion apparatus.

In this embodiment, the image conversion means is configured using the conversion ROM. However, the present invention is not limited to this, and the conversion table is stored in the RAM and used, or the calculation using the image conversion model described above is performed. It may be performed at high speed and the input image may be converted into an output image in real time.

Furthermore, in this embodiment, the screen size is described as an aspect ratio of 3: 4. However, the present invention is not limited to this. For example, when the aspect ratio is 9:16, a value of y: −0.5625 to 0.5625 is used. Use as

Further, in the image conversion model according to the present invention, the conversion from the virtual object plane to the input image is performed using Equation 4
This is because the present embodiment is premised on an equidistant projection lens, and when a lens based on stereoscopic projection is premised, the following equation 5 or g (θ) indicating the lens characteristics is used: Equation 6
Etc. may be used. In the present invention, the present invention can be applied even when a lens having a viewing angle exceeding 180 degrees is used. In particular, when the viewing angle exceeds 150 degrees, a greater effect than the method described in FIG. 6 can be obtained. Can do.

g (θ) = 2f · tan (θ / 2)... Equation 5 where f represents the focal length.

g (θ) = 3f · tan (θ / 3)... Equation 6 where f represents the focal length.

Specifically, instead of Equation 4,

  Equation 7 is used.

Further, in the image conversion model according to the present invention, the coordinate calculation formula in the z direction of the correction surface is shown in Equation 2, but this shows only one type of correction surface. For example, when Expression 2 is used, the limit angle of view that can be projected from the optical axis origin to the correction surface is about 213 °, and x1 at this time is about ± 1.55. Therefore, when Expression 2 is used, the angle of view that can be captured by the fisheye lens is limited to within about 213 °.

If a wider angle of view is required, Equation 2 can be changed to cope with it. As described above, the present invention is not limited to the expression 2, and the advantageous polynomials and numerical sequences that are effective for the present invention and approximate to this expression are also within the scope of the present invention.

FIG. 2A is a perspective view and FIG. 2B is a top view showing an image data conversion model according to the present invention. It is the image after performing the image data conversion by this invention. It is an image explaining the characteristic of the horizontal direction of the image data conversion by this invention. It is a perspective view for demonstrating the conversion procedure using the image data conversion model by this invention. 4 is a conversion formula used in the image data conversion model according to the present invention. A conventional image data conversion model is shown, (A) is a perspective view, and (B) is a top view thereof. It is the image after image conversion using the model of the conventional image data conversion. It is an image after image conversion using another conventional image data conversion model. FIG. 5A is a perspective view and FIG. 5B is a top view showing another conventional image data conversion model. It is the image (before image data conversion) imaged via the fisheye lens. It is a block diagram which shows the conventional image data converter and camera apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Input image 2 Virtual object surface 2 Hemisphere 3 Correction surface 3a Center surface 3b Right side surface 3c Left side surface 4 Display surface 5 Optical axis 6 Imaging surface 80 Camera apparatus 81 Fisheye lens 82 Imaging element 83 Image processing part 90 Image data converter 91 Digital Input terminal 92 Analog input terminal 93 A / D converter 94 Image memory unit 95 Image interpolation unit 96 Image conversion means 97 D / A converter 98 Analog video output terminal

Claims (4)

Image conversion means for inputting an image of a subject including spherical distortion imaged using a fisheye lens as an input image signal, performing image conversion so as to remove the spherical distortion from the input image signal, and outputting an output image as a signal In an image data conversion device comprising:
The image conversion means converts the entire input image so that the linear object directed toward the optical axis of the fisheye lens and the vertical object are substantially linear in the output image,
In addition, the subject that is straight in the horizontal direction is substantially linear in the central portion of the output image, and the left and right portions that are connected to the central portion and the left and right, and the central portion and the left portion of the output image. And an image data conversion device obtained by image-converting the entire input image so as to be bent near the boundary of the right portion. The image conversion means is configured based on a virtual image conversion model,
The image conversion model includes the input image formed of a circular plane, a hemispherical virtual object plane on which the subject is projected corresponding to the fisheye lens, a correction surface for correcting the spherical distortion, and a display subjected to image conversion. The virtual optical axis corresponding to the optical axis of the fisheye lens is inserted in a right angle from the optical axis origin located at the center of the input image, and the correction surface is
Consists of a central surface arranged in the center, a left side surface and a right side surface respectively arranged on the left and right of the central surface, the left side surface and the right side surface are bent toward the optical axis origin,
The image on the virtual object surface projected onto the correction surface in the virtual optical axis direction is projected onto the display surface by an orthogonal projection method, and the projected image on the display surface is used as the output image. 2. The image data conversion apparatus according to claim 1, wherein: In the conversion model for forming the correction surface, when the horizontal and vertical coordinates of the display surface are x, y and the three-dimensional coordinates of the correction surface are x, y, z, z in the virtual optical axis direction is ,









The image data conversion apparatus according to claim 2, wherein the image data conversion apparatus is calculated by the following formula. A camera device comprising the image data conversion device according to any one of claims 1 to 3, wherein a captured image is converted by the image data conversion device and output.