KR101427181B1 - Calibration indicator used for calibration of onboard camera using variable ellipse pattern and calibration method of onboard camera using calibration indicator - Google Patents

Abstract

The present invention relates to a calibration indicator for a camera camera calibration in which a plurality of patterns are arranged, wherein a plurality of variable elliptic patterns, each of which is set so that each of the patterns appears as a circle in an image of the calibration indicator through a wide- .
According to the present invention, since the minutiae coordinates in the distorted image can be detected more accurately, the camera calibration performance is improved, and accurate image matching can be performed accordingly. Thus, visibility of the driver is improved by showing the vicinity of the vehicle accurately, Therefore, it is possible to accurately recognize and improve the safety during operation.

Description

Technical Field [0001] The present invention relates to a calibration indicator of a camera for a vehicle having a variable elliptic pattern, and a calibration method using the same. [0001] The present invention relates to a calibration indicator for a camera,

BACKGROUND OF THE INVENTION 1. Field of the Invention [0002] The present invention relates to a calibration indicator of a camera for a vehicle and a calibration method using the same. More particularly, the present invention relates to a calibration technique for accurately detecting a feature point located at an outer portion of an image.

Image sensor based cameras are widely used in vehicles. By turning on a portion of the environment surrounding the vehicle and displaying it on the monitor in front of the driver, such a system can be used to determine the position of the vehicle relative to other objects (such as other vehicles, pedestrians, cyclists, buildings, And better orientation of the vehicle, improve control of the overall traffic situation, and thereby improve vehicle drivability.

In order to allow the driver to observe all the surroundings of the vehicle at the same time, a system including a plurality of cameras is used. The camera is mounted to the vehicle such that any point within the peripheral region of the vehicle is within a field of view (FOV) of the at least one camera. As one of the most commonly used system structures described in many publications, such as Japanese Patent No. 4765649, for example, as shown in Fig. 1 (a), four cameras, i.e., a front view camera A rear view camera 1 of a rear view panel or a boot reed, and side view cameras 3 and 4 of left and right side mirror areas.

An image turned on by each camera is transmitted to an electronic control unit (ECU) including image processing means. The ECU is capable of converting images (eliminating "fish-eye" effects, distortion, and the like) and displaying the images in a single view to give the same feeling It combines. The ECU transmits the synthesized image to the display device (see Fig. 1 (b)). The advantage of this method is that the displayed image allows the driver to see the whole area around the vehicle at the same time and also allows the driver to observe the relative position of the vehicle to the parking markers and external objects in a very simple and natural form, It makes it easy and safe.

However, if such cameras are misaligned with respect to the vehicle and each other, the displayed image may not only be uncomfortable to the eyes of the driver but also cause confusion and misunderstanding of distances to external objects, which may ultimately hinder safety. Fig. 2 shows an exemplary result in the case of misalignment of the front camera 1. Fig.

In order to solve such a problem, correct recognition of the feature point of the image at the fixed position of the camera is required. If the minutiae can not be recognized correctly, the worker has to work manually. This operation is too labor-intensive to be used in a vehicle assembly line.

 In addition, when the work is manually performed, the working time is different according to the skill of the worker. The difference in the working time will have a cumulative effect on the sales because the product productivity will be different.

Various conventional methods for correcting a camera have been proposed. One of them is known in which a shape is known in advance, and a stereoscopic feature is placed on a three-dimensional space as an index for calibration, and the characteristic points of the three- A method of calculating camera parameters according to how they are displayed is known.

For example, a method of calculating a camera parameter of 11 degrees of freedom using a characteristic three-dimensional calibration index has been proposed. Here, in order to determine the camera parameters, a method of determining the camera parameters in a three-dimensional space by placing a characteristic three-dimensional object at a position where the three-dimensional object can be imaged by the camera, geometrically calculating at which point the three-dimensional object is displayed in the image coordinate system of the camera .

However, in the method of performing the calibration of the camera after attaching the camera to the vehicle as described above, when the object to be calibrated is a camera attached to the vehicle, calibration is performed in a vehicle assembly line such as a factory. In the production site, it is not efficient to place the three-dimensional object every time as a calibration index if a method of correcting a characteristic three-dimensional object is taken, and it is preferable to avoid placing a three- dimensional object around the object for correction in view of disaster prevention And the calibration index for performing the calibration of the camera is preferably an index that is problematic even if the operator steps on it, such as a floor pattern drawn on the floor of the factory.

In order to solve such a problem, as shown in FIG. 3, various floor pattern-type calibration indices for the calibration of a car camera have been developed.

These calibration indices are obtained by extracting the intersection points of the grid pattern using the chessboard pattern as in (a) or the checkerboard pattern grid as in (b) as feature points, and then calculating projection transformation parameters using the extracted feature points, (Top-down view).

Alternatively, as shown in (c), a method of extracting a center point of a point pattern as a feature point by using a calibration index of a plurality of dot pattern types arranged at isosceles intervals is also used.

FIG. 4 shows an example of an image in which a rear vision camera photographs a calibration index of a regular interval pattern. Referring to FIG. 4, when a calibration operation is performed using conventional calibration indices, the shape of the bottom pattern is disturbed at a portion far from the camera, so that the feature points are hard to be correctly recognized. In other words, since the outline part appears as a compressed image, the feature point interval of the grid pattern or the point pattern is narrow, so that accurate feature point extraction is difficult. As shown in FIG. 4, in the case of the point pattern, The circle appears as an ellipse due to the distortion of the wide-angle lens, so that the feature point extraction becomes more difficult.

SUMMARY OF THE INVENTION The object of the present invention is to provide a method of correcting a conventional circular point pattern calibration index by arranging ellipses at a boiling interval, , The shortening ratio and the direction are adjusted so that the ellipse is displayed in the form of a circle in the image so that the feature points located in the outer portion of the image can be accurately detected.

According to a preferred aspect of the present invention, there is provided a calibration index for calibration of a camera for a vehicle in which a plurality of patterns are arranged, wherein, in an image obtained by photographing the calibration index through a wide- There is provided a vehicle camera calibration index having a variable elliptic pattern set to appear in a circle shape.

Here, the variable elliptic pattern is a form in which at least one of a length of a major axis and a minor axis, a ratio of a length of a major axis of a major axis, and a direction of a major axis vary depending on a pattern position. As the position of the pattern is further away from the camera, The length or the ratio of the length of the minor axis to the major axis may be increased, or the direction of the major axis may be formed to face the camera, or both of these conditions may be satisfied.

In the case where the calibration indexes of the present invention are applied to an approach-view system in which a plurality of cameras having overlapping shooting regions mutually are installed in a vehicle, the variable elliptic patterns located at overlapping portions of cameras are rectangular The length of the long axis and the minor axis, the ratio of the length of the minor axis to the major axis, the ratio of the length of the minor axis to the major axis, and the direction of the major axis with respect to the center point, Wherein the variable elliptic pattern is disposed so that its major axis is oriented toward the center point, and the length of the major axis and minor axis of the ellipse and the ratio of the major axis of the minor axis to the major axis of the minor axis And the interval between the center points of the ellipse are increased.

In accordance with another aspect of the present invention, there is provided a variable elliptic pattern in which at least one of a length of a major axis and a minor axis, a ratio of a major axis of a minor axis to a major axis, A step of forming a plurality of circular patterns corresponding to the variable elliptic patterns from the photographed image, a step of detecting a plurality of circular patterns corresponding to the variable elliptical patterns from the photographed images, Extracting a center point of the circular pattern as a feature point, and performing the calibration of the camera using the extracted feature point. A method for calibrating a camera for a vehicle using the variable elliptic pattern is provided.

Here, the variable elliptic pattern is formed such that at least one of a length of the major axis and minor axis and a length ratio of the major axis of the minor axis increases as the position of the pattern is further away from the camera.

According to the present invention, since the minutiae coordinates in the distorted image can be detected more accurately, the camera calibration performance is improved, and accurate image matching can be performed accordingly. Thus, visibility of the driver is improved by showing the vicinity of the vehicle accurately, Therefore, it is possible to accurately recognize and improve the safety during operation.

In addition, there is also an effect that accuracy of minutiae coordinate detection can be increased while using the same algorithm.

FIG. 1 is a view showing a location where four cameras are installed in a vehicle and a top view image generated using the locations.
2 shows a top view image at the time of misalignment of the front camera.
Figure 3 shows calibration indices used for conventional vehicle camera calibration.
FIG. 4 shows an example of an image in which a rearview camera photographs a calibration index of a general check pattern and a dot pattern.
5 and 6 are views for explaining the principle of drawing a calibration index of a vehicle camera according to the present invention.
Fig. 7 shows a calibration index of a vehicle camera according to the present invention.
8 is an enlarged view of the right calibration index in Fig.
9 is an enlarged view of the calibration index of the overlapping point of the image.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings, which will be readily apparent to those skilled in the art to which the present invention pertains. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein.

In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Throughout the specification and claims, where a section includes a constituent, it does not exclude other elements unless specifically stated otherwise, but may include other elements.

5 and 6 are views for explaining the principle of drawing a calibration index of a vehicle camera according to the present invention.

FIG. 5 shows a circular mark displayed at a minutiae point (intersection) in an original image having a distortion obtained by photographing a grid pattern with a fish-eye lens camera, which is mainly used as a vehicle camera for an around view system. , And the projected image is transformed from the distorted image having the circular mark to the bottom surface.

Referring to FIG. 6, it can be seen that when the image of FIG. 5 is corrected, the circular mark of FIG. 5 is deformed into an ellipse in FIG. 6, and in particular, the shape of the ellipse has a constant pattern.

That is, although the circle mark nearest to the camera shows a normal circular shape, it can be seen that as the distance from the camera increases, the shape of the ellipse becomes longer. Further, it can be seen that the direction of the long axis of the circular mark is inclined to approximately the direction of the camera.

The applicant of the present invention has arrived at the present invention by focusing on the point that a normal circle in the distorted image is converted into an ellipse having a certain pattern in the distortion-corrected image according to the position (distance from the camera).

That is, when the calibration index for the vehicle image calibration is formed in the shape of a variable ellipse having a certain pattern on each feature point as shown in FIG. 6 and the image is taken with a fish-eye lens, It is based on the principle of inversion that it will appear in the form of a normal circle.

Fig. 7 shows a calibration index of a vehicle camera according to the present invention. As shown in FIG. 7, the calibration index of the camera for a vehicle according to the present invention forms a shape such that each pattern forms a deformed elliptic pattern deformed according to the position, as shown in FIG.

In Fig. 7, 1 denotes a front view camera, 2 denotes a rear view camera, 3 denotes a right view camera, and 4 denotes a left view camera, and the L line is a straight line extending virtually from the center of each camera.

Referring to FIG. 7, the pattern closest to the camera has a normal circular shape, and an ellipse (variable elliptic pattern) in which the longer axis becomes longer as the distance from the camera is longer. The longer axis is aligned in the camera direction It can be seen that a variable elliptic pattern is formed.

8 is an enlarged view of the right calibration index in Fig.

8, the ellipses disposed at the respective feature points of the right region are arranged such that the major axis thereof is the same as the direction forming the straight line with the right field of view camera 3, and the length of the major axis and minor axis of the ellipse, , So that the distance between the center points of the ellipses becomes larger as the distance from the camera increases in a radial fashion with respect to the camera, so that the pattern becomes a circle at the time of photographing with a wide angle camera. Preferably, the diameter of the pattern in the image photographed by the wide-angle camera is equal to or larger than 10 pixels, and the interval between the patterns is preferably equal to or greater than 6 pixels.

In FIG. 8, each line L, L ', L "is a straight line extending from a feature point pattern closest to the camera or camera, and it can be seen that the long axis direction of each feature point is in the same direction as the straight line. The direction of the major axis of the pattern may not necessarily be exactly coincident with the direction of the camera due to the influence of the detailed specifications or the surrounding environment. In this case, the image captured by the camera is checked by changing the fine direction and size of the patterns It is preferable to form a more patterned pattern.

9 is an enlarged view of the calibration index of the overlapping point of the image.

The elliptic patterns positioned at the overlapping areas of the cameras are arranged such that the major axis is located at the center point of a pattern formed at a position closest to the vehicle or a vertex closest to the vehicle among the overlapping areas set in a rectangle, The distance between the long and short axes of the ellipse, the ratio of the major axis of the minor axis to the major axis of the minor axis, and the distance between the center points of the ellipse become larger as the distance from the central point in the radial form around the center point increases. So that it can be maintained at 6 pixels or more.

As described above, by adjusting the size of the major axis and the minor axis, the shape of the ellipse appearing in the distorted image is displayed as the optimal circle shape. Then, when the center point of the circle is detected as the feature point, the erroneous detection is removed and the correction efficiency is maximized can do.

When the area is specified from the center of the pattern, the feature point of the bottom pattern is automatically detected to the designated area.

More specifically, first, a calibration index having the above-described bottom pattern is formed on a floor surface, a calibration index is photographed using a camera at a predetermined position, and then a plurality of circular patterns corresponding to the variable- The center point of the circular pattern is extracted, and the coordinates of the center point are selected as feature points, and the camera is calibrated.

1: Front view camera 2: Rear view camera
3: Right field of view camera 4: Left field of view camera

Claims (8)

1. A calibration index for a camera camera calibration in which a plurality of patterns are arranged,
A variable elliptic pattern in which each of the patterns is displayed in a circular shape in an image of the calibration index taken through a wide-angle lens,
Wherein the variable elliptic pattern has at least one of a length of a major axis and a minor axis, a ratio of a length of a major axis of the major axis, and a direction of the major axis depending on a pattern position.
delete The method according to claim 1,
Wherein the variable elliptic pattern has at least one of a length of a long axis and a length of a minor axis and a length ratio of a major axis of the minor axis increases as the position of the pattern is further away from the camera.
The method according to claim 1,
Wherein the variable elliptic pattern is formed such that the direction of the major axis is directed toward the camera.
The method according to claim 1,
In the case of an ambient view system in which a plurality of cameras having overlapping shooting regions are installed in a vehicle,
The variable elliptic patterns located at the overlapping areas of the cameras may be arranged such that a pattern formed at a vertex closest to the vehicle or a position closest to the vehicle among overlapping areas set in a rectangle is used as a center, Wherein at least one of the length of the major axis and the minor axis, the ratio of the length of the minor axis to the major axis, and the direction of the major axis are varied depending on the position.
6. The method of claim 5,
The variable elliptic pattern is arranged so that its major axis is directed to the center point, and the distance between the major axis and minor axis of the ellipse, the ratio of the minor axis to the major axis, and the distance between the center points of the ellipse are increased Wherein the camera calibration indicator has a variable elliptic pattern.
Forming a calibration index on the bottom surface in which a plurality of variable elliptic patterns in which at least one of the lengths of the major axis and the minor axis, the ratio of the length of the minor axis and the major axis, and the direction of the major axis are varied according to the position of the pattern;
Photographing the calibration indicator using a camera at a preset position;
Detecting a plurality of circular patterns corresponding to the variable elliptic patterns from the photographed image;
Extracting a center point of the circular pattern as a feature point;
And performing the calibration of the camera using the extracted minutiae points.
8. The method of claim 7,
Wherein the variable elliptic pattern has at least one of a length of a long axis and a length of a minor axis and a length ratio of a major axis of the major axis as the position of the pattern is further away from the camera.

Images