JP3302078B2 - Inspection method of windshield of automobile - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION This invention is, wind Sea of motor vehicles
For simulating optical distortion of a perspective image (hereinafter referred to as “perspective distortion”) when an object is observed through a screen (that is, a windshield) and inspecting the perspective distortion to inspect a windshield of an automobile. About
Is what you do.

[0002]

2. Description of the Related Art Recently, the design of the outer shape of an automobile has been to adopt a complicated three-dimensional curved surface in consideration of aerodynamic characteristics. For this reason, the shape of a glass such as a windshield provided in the automobile is required. Are also being curved. However, due to the limitations of glass forming technology, it is difficult to keep the glass surface smooth in response to the curved surface, so when the driver looks at the object through the windshield, the object may appear distorted. is there. This phenomenon is referred to as a perspective distortion phenomenon, and the magnitude of the perspective distortion is generally defined as a lateral distortion angle, which is a lateral distortion angle, and a vertical distortion angle, for each point on a plate-like body such as a windshield. It can be represented by two physical quantities, a certain longitudinal distortion angle. Then, as shown in FIG. 9, the transverse distortion angle is the angle α formed between the horizontal line segment AB observed without passing through the plate-shaped body and the line segment A′B ′ observed through the plate-shaped body, and The distortion angle is defined by an angle β formed between a vertical line segment AC observed without passing through the plate-shaped body and a line segment A′C ′ observed through the plate-shaped body.

Makiguchi et al. Have been studying how to determine the threshold distortion angle at which humans begin to see through distortion by combining the actually detected distortion angle and the sensory evaluation of the windshield of an automobile. ("Analysis of Perspective Distortion of Windshield Glass for Automobile", 15th Symposium on Multivariate Analysis by Nikkagiren, November 1991
Month). According to this, it was found that the threshold distortion angle was different due to the difference in the so-called observation zones G1 to G4 (see FIG. 10) when the windshield 1 was observed from the eye point of the vehicle defined by the JIS standard. When the horizontal distortion angle is set on the horizontal axis and the vertical distortion angle is set on the vertical axis in each of the observation zones G1 to G4, the threshold distortion angles are approximately represented by a straight line (discrimination function) as shown in FIG. It became clear that it was expressed. Therefore, by using this discriminant function, the perspective distortion that can be sensed by humans can be reduced by detecting the vertical and horizontal distortion angles through each of the plurality of small regions that constitute the substantially entire region of the windshield 1. It is possible to evaluate whether or not the windshield 1 is generated.

[0004]

[Problems to be solved by the present invention]
In detecting the perspective distortion of a plate-like body such as a windshield, as shown in FIG. 12, a camera is held at an eye point EP of the automobile 2 with the windshield actually mounted, and The distortion angle is detected by taking an image of the panel 4 having the orthogonal lattice-shaped linear pattern.

The inventor of the present invention has jointly invented a system capable of detecting the perspective distortion of a plate-like body with high accuracy in a short time with another person and has applied for a patent (Japanese Patent Application No. 4-141001). .
In this system, as shown in FIGS. 13 and 14,
The windshield 1 is attached to the mounting table 5 which can be rotated in the horizontal direction, and while the windshield 1 is rotated in the horizontal direction, the three LEDs 6 at right angles in the LED plate 7 in which a large number of LEDs 6 are vertically arranged are arranged. This target is sequentially imaged by the swingable CCD camera 8 held at the eye point EP, thereby detecting perspective distortion. In this case, as shown in FIG. 14A, a large number of LEDs 6 are arranged in two rows vertically and a large number of rows horizontally at appropriate lattice intervals. In the first imaging, three LEDs (n, 1), (n, 2), and (3) in FIG.
Only (n−1, 1) emits light and these three LEDs are imaged. In the second imaging, the three LEDs (n−1, 1), (n−1) in FIG. 1,2), (n-
Only 2 and 1) emit light, and these three LEDs are imaged. Then, such imaging is sequentially performed over substantially the entire surface of the windshield 2. Further, by detecting the coordinate positions of the three LEDs imaged in this way, the windshield 1 corresponding to the target composed of these three LEDs can be changed in the horizontal distortion angle α and the vertical distortion angle β shown in FIG. Is calculated for each small area.

However, in this system and in the case of the above-described perspective distortion detection method of Makiguchi, etc., the perspective distortion is detected by actually observing an object through a windshield. For this reason, when a perspective distortion exceeding the threshold distortion angle is found, it is necessary to redesign the shape of the windshield and recreate the heated bending die or to partially correct the heated bending die. Therefore, it takes a long time to manufacture the heating bending mold, and waste is likely to occur in the process.

[0007] In order to efficiently manufacture a heating bending die, attempts have been made to predict and calculate perspective distortion by an optical drawing method using a design drawing of a windshield.
However, in this method, since a great deal of time is required for optical drawing, it is difficult to predict and calculate the perspective distortion over the entire area of the windshield, and therefore, the method is not practical.

The present invention has been made in view of the above-described problems,
It is an object of the present invention to provide a method capable of detecting a perspective distortion of a windshield of an automobile in a short time with high accuracy by performing a simulation calculation without actually detecting a perspective distortion.

[0009]

Means for Solving the Problems The present invention is automatic in order to achieve the above object, obtained from the given geometry data
In the method for inspecting the windshield by dividing the shape model of the three-dimensional curved surface of the windshield of the car into a plurality of small areas, simulating the perspective distortion for each of the plurality of small areas, and inspecting the perspective distortion, Obtaining a shape model of the three-dimensional curved surface of the windshield from given shape data, and a virtual plane including a plurality of targets having at least three virtual points;
Ipoin as an observation reference point for the above shape model
A step of obtaining on the opposite side of the bets, at least 3 representing a direction in which said at least three virtual rays toward the direction of the eye point from each virtual point proceeds after being refracted by one side and the other side of the geometric model and the straight line of the present, the small for at least three and obtaining a point of intersection, the constitute the same target at least three temporary <br/> virtual point and the at least three virtual points between the virtual plane
Calculating a distortion angle for each of the small regions of the shape model corresponding to the target from at least three intersections .
It is related to the windshield inspection method.

[0010] In the present invention, the vehicle wins
It is preferable that the method further includes a step of displaying the distortion angle of each of the small regions so that the perspective distortion of the shield can be easily evaluated .

[0011]

According to the present invention, before manufacturing a mold such as a heating bending mold used for actually manufacturing a windshield of an automobile, the perspective distortion of the windshield as seen by a driver can be accurately determined in a short time. Can be eliminated, so that useless steps such as re-forming the mold can be omitted.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A method for inspecting an automobile windshield according to one embodiment of the present invention will be described below.

FIG. 1 shows a window of an automobile according to this embodiment.
FIG. 2 is a flow chart of a method of inspecting a shield , FIG. 2 schematically shows an inspection system for implementing the method of the present embodiment, and FIG. And the positional relationship between the virtual plane and the virtual plane.

As shown in FIGS. 1 and 2, in the method of inspecting a windshield of an automobile according to the present embodiment,
First, a shape model of the three-dimensional curved surface is generated from the shape data (step S1 or S2) of the windshield (step S3). As shown in FIGS. 1 and 2, when the windshield is designed by the CAD system, the shape data is transmitted to the computer 12 by the magnetic reproducing device 11 from a magnetic tape or the like on which the data (step S1) in the shape design stage is written. Input or input the design drawing 13 (step S2) to the digitizer 14
May be input to the computer 12 by plotting.

As shown in FIG. 3, the generated shape model 10 includes an inner surface SU1 and an outer surface SU2, and the shape model 10 of the inner surface SU1 and the outer surface SU2.
Are represented as an aggregate of patches (surface elements) 10, respectively. And each patch, for example, Coons (Co
ons) It can be mathematically represented by a curved surface, a Bezier surface, or a B-spline surface.

FIG. 4 shows a flow of a modeling process of the shape model 10 of the three-dimensional curved surface of the windshield 1. In FIG. 4, first, a number of points P1, P on a plan view and a side view of a windshield 1 called a Mylar diagram as shown in FIGS. 5A and 5B.
2. The three-dimensional coordinate data of P3 ‥‥ is designated by the digitizer 14 or the like and input to the computer 12 (step S
11). Next, a lattice-shaped three-dimensional spline curve passing through the point sequence is generated (step S12). Next, the surface is divided into quadrilateral patches with the spline curve as a boundary, and a bicubic parametric surface such as a Coons surface represented by parameters u and v along each side of the patch is generated (step S13). This curved surface is used as a shape model of the inner surface SU1 of the windshield 1, and then a shape model of the outer surface SU2 offset by the plate thickness is generated (step S14). The final shape model 10 is formed by combining the shape models of the inner surface SU1 and the outer surface SU2.

As shown in FIGS. 1 to 3, the computer 12 generates a shape model 10 of the windshield 1 based on the input shape data, and then places the observation model at a position corresponding to the position of the shape model. An eye point EP is determined (step S4). In this case, not only the data on the shape model (glass specifications such as plate thickness), but also the hip point (the position of the driver's hip) and the torso angle (the tilt angle of the seat) are considered as necessary.
Further, the computer 12 generates a virtual plane 21 on the opposite side of the eye point EP with respect to the shape model (that is, on the outer surface side of the shape model) (step S5). The virtual plane 21 is generated at a position appropriately distant from the shape model 10 so as to be substantially parallel to the shape model 10.
In addition, the virtual plane 21 is formed by three virtual points P substantially at right angles.
A number of targets 22 (corresponding to the targets in FIG. 14B) composed of O, POa, and POb are provided on the eye point EP side (the same number as the number of small regions described later).
Furthermore, when the virtual plane 21 is observed from the eye point EP through the shape model 10, the target 22 divides the surface of the shape model 10 more finely (for example, into 500 to 5,000 pieces) than the patch ( (Hereinafter referred to as “small region”) 23 are arranged on the virtual plane 21 such that one target 22 is observed.

Next, the computer 12 operates the shape model 1
Using 0, the eye point EP, and the virtual plane 21, a calculation for obtaining the perspective distortion angle is performed. As a calculation procedure, first, an eye point E is calculated from each of the three virtual points PO, POa, and POb that constitute a certain target 22.
The virtual ray heading in the direction of P is the outer surface S of the shape model 10.
The intersection between the virtual plane 21 and a straight line representing the direction of travel after refraction at U2 and the inner surface SU1 is determined (step S6). The position of this intersection is approximately the virtual point P
O, POa, and POb can be regarded as positions where they are actually observed. And three virtual points PO, POa, PO
From b and the three above intersections for these virtual points,
A distortion angle is calculated for the small region 23 corresponding to the target 22 forming these virtual points (step S7). By repeating such a calculation procedure, data on the distortion angle obtained for each of the many small areas 23 is displayed in the form of a distortion angle distribution diagram by the graphic display 15, the printer 16, and the hard copy device 17 (step). S8).

FIG. 6 shows the procedure for calculating the perspective distortion angle, and FIG. 7 is a plan view for explaining the procedure for finding the intersection in more detail. The calculation of the perspective distortion angle uses, in principle, the fact that the traveling direction of the light beam changes due to the refraction effect when the windshield 1 is provided or not. First, three substantially perpendicular virtual points P0, P0 constituting a certain target 22 on the virtual plane 21 are set.
One virtual point P0 of 0a and P0b is set as a calculation start point (step S21). Next, a vector VK0 in the traveling direction of the virtual ray 24 from the virtual point P0 toward the eye point EP is obtained (step S22). Next, an intersection P1 between the straight line passing through the vector VK0 and the outer surface SU2 of the shape model 10 is determined (step S23), and a normal vector VV1 of the outer surface SU2 of the shape model 10 at the intersection P1 is determined (step S24).

Further, the vector VK0 and the vector VK
From V1, a vector VK1 representing the path after the virtual ray is refracted on the outer surface SU2 of the shape model 10 according to the law of refraction is determined (step S25). That is, the angle of incidence of the virtual ray 24 from the virtual point P0 on the outer surface SU2 (with respect to the normal) is i, and the outer surface SU of the virtual ray 24 is
Assuming that the angle of refraction at 2 (with respect to the normal) is r and the refractive index of the windshield to air is n, sin i / si
Since n r = n holds, the vector VK0, the vector V
The refraction angle can be determined from V1 and the known refractive index n, from which the vector VK1 can be obtained.

Next, the intersection P2 between the straight line passing through the vector VK1 and the inner surface SU1 of the shape model is determined (step S2).
6), the inner surface S of the shape model 10 at the intersection P2
A normal vector VV2 of U1 is obtained (step S2
7). Further, a vector VK2 representing the path after the virtual ray 24 has been refracted on the inner surface SU1 of the shape model 10 is obtained from the vector VK1 and the vector VV2 according to the above-described law of refraction (step S28). Finally, an intersection P3 between the straight line passing through the vector VK2 and the virtual plane 21 is obtained (step S29). The position of the finally obtained intersection (virtual point) P3 is approximately the eye point E
When observed from P, the virtual point P0 can be regarded as a position where the virtual point P0 is actually observed.

Next, for the other virtual points P0a and P0b constituting the same target 22, step S
By executing 21 to S29, the corresponding intersection points (virtual points) P3a and P3b are obtained (step S30). From these data, the vertical and horizontal distortion angles of the small region 23 of the shape model 10 corresponding to the target 22 having the three virtual points PO, POa, and POb are calculated in accordance with the definition of the distortion angle shown in FIG. (Step S31). In this case, the elongation percentage in each of the vertical and horizontal directions can be calculated. In FIG. 9,
The lateral elongation is the ratio of (line A'B'-horizontal line AB) to the horizontal line AB, and the vertical elongation is the ratio of (line A'C'-vertical line AC) to the vertical line AC. Is defined respectively.

Further, by executing steps S21 to S31 for the other targets 22, the vertical and horizontal distortion angles of the small region 23 corresponding to those targets 22 are calculated (step S32). By repeating the above operation, it is possible to obtain the distortion angle for almost the entire area of the shape model 10.

The obtained data on the distortion angle is recorded in a file, and the distribution of the distortion angle of the entire windshield 1 can be grasped at a glance in FIG.
As shown in (1), the magnitude of the horizontal distortion angle or the vertical distortion angle is divided into appropriate ranges and displayed on a display and / or printed as a color-coded distortion angle distribution map (step S33). In FIG. 8, the distortion angle increases in the order of the color-coded areas C ₁ and C ₂ ‥‥ C ₁₁ .
Also, in this display, it is possible to indicate whether or not a perspective distortion that can be sensed by a human is generated in the windshield 1 by using a discriminant function as shown in FIG. further,
In this display, the observation zone G1 as shown in FIG.
If G4 is additionally displayed, it is possible to indicate for each observation zone of the windshield 1 whether or not there is a perspective distortion that can be sensed by a human.

As described above, in the present embodiment, the perspective distortion generated in the windshield of the automobile is detected by simulating using a computer, so that the perspective distortion angle of the windshield can be quickly and quickly determined. It is possible to easily detect .

[0026]

According to the present invention, as described above,
At least three virtuals that make up the same target
From the point of view, the shape model of the three-dimensional curved surface of the windshield of the car
Refracting through Dell and going towards the eye point
At least three straight lines representing the directions of the three virtual rays
And at least three intersections with the virtual plane
The at least three virtual points and at least these
At least the three intersections for the three virtual points
From the small area of the shape model corresponding to each target,
The distortion angle for each is calculated. this
For, found through relatively simple simulation translucency <br/> Miibitsu automobile windshield viewed from the driver, it is so arranged to inspect the windshield by examining the perspective distortion, driving Of the car seen from the driver
The above-mentioned window , which is capable of detecting the perspective distortion of the shield in a short time and with high accuracy, and inspecting the perspective distortion.
The inspection of the shield can be performed in a short time and with high accuracy. In addition, since the windshield of an automobile at the design stage (ie, before actually manufacturing a mold such as a heating bending mold) can be a simulation target, the perspective distortion seen from the driver who actually needs to correct the above is If it is found to occur in the windshield , the mold can be tailored accordingly, thus eliminating the need to remanufacture the windshield mold of the vehicle .
Therefore, since the final specification of the windshield of the vehicle that satisfies the standard with respect to the perspective distortion seen by the driver can be determined before the production of the mold, the degree of freedom of design is increased, and the design of the shape of the windshield of the vehicle is increased. From the production of the mold to the molding of the windshield of the automobile can be greatly shortened, and the production cost can be reduced.

[Brief description of the drawings]

FIG. 1 is a view of an automobile windsea according to one embodiment of the present invention.
5 is a flowchart of a field inspection method.

FIG. 2 is a schematic diagram of an example of an inspection system for performing the method shown in FIG. 1;

FIG. 3 is a diagram showing a positional relationship between a shape model of a windshield, an eye point, and a virtual plane in the method shown in FIG. 1;

FIG. 4 is a flowchart of a modeling process of a shape model used in the method shown in FIG. 1;

5A is a plan view of a windshield used in the modeling process shown in FIG. 4, and FIG. 5B is a side view of the same.

FIG. 6 is a flowchart showing a procedure for calculating a perspective distortion angle shown in FIG. 1;

7 is a shape model for explaining the procedure shown in FIG. 6,
FIG. 4 is a diagram illustrating a positional relationship between an eye point and a virtual plane.

FIG. 8 is a front view of a windshield showing a distribution of perspective distortion angles obtained by executing the procedure shown in FIG. 6;

FIG. 9 is a diagram for explaining a definition of a perspective distortion angle of a plate-like body.

FIG. 10 is a diagram showing distinction of observation zones of a windshield when observed from an eye point of a car.

11 is a graph showing a discriminant function for each observation zone shown in FIG.

FIG. 12 is a diagram illustrating an example of a conventional method for detecting perspective distortion of a windshield of an automobile.

FIG. 13 is a diagram showing a system for detecting perspective distortion of a windshield of an automobile in Japanese Patent Application No. 4-141001 invented jointly by the present inventors with another person.

14A is a front view of the LED plate shown in FIG. 13 with a middle portion cut away, and FIG. 14B is a partially enlarged view thereof.

[Explanation of symbols]

 Reference Signs List 1 windshield 10 shape model 12 computer 21 virtual plane 22 target 23 small area 24 virtual ray EP eye point SU1 inner surface SU2 outer surface

Claims (2)

(57) [Claims] 1. An automobile obtained from given shape data
Dividing the shape model of the three-dimensional curved surface of the windshield into a plurality of small areas, simulating the perspective distortion for each of the plurality of small areas, and inspecting the perspective distortion to inspect the windshield. and obtaining a geometric model of a three-dimensional curved surface of the windshield from the shape data that is, a plurality comprises virtual plane a target having at least three virtual point, the observation reference point for the shape model
A step of obtaining on the opposite side of the eye point of Te, the from each of said at least three virtual point Aipo
Represents the direction in which the virtual ray heading for the direction of the point travels after refraction on one and the other surface of the shape model
At least three straight lines and at least the virtual plane
Obtaining three intersections, and at least three of said at least three points making up the same target
Calculating, from a virtual point and the at least three intersections of the at least three virtual points, a distortion angle for each of the small regions of the shape model corresponding to the target. Feature
Inspection method of windshield of automobile . 2. The method of claim 1, further comprising the step of displaying the distortion angle for each of the sub-regions.