JPH11237216A - Evaluation method for optical distortion and evaluation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To quantify optical distortion regardless of the kind of the distortion, by calculating correlation value between a perspective image or a reflection image and an ideal image, and evaluating the optical distortion of a measured object based on the calculated correlation value. SOLUTION: A reference pattern 2 from a light source transmits a measuring object 1 such as a plate glass for measuring optical distortion and is received by an imaging device 4. The observed perspective image 3 by the imaging device 4 is inputted in an operator device 5 such as a computer. The operator device 5 evaluates optical distortion of the measuring object 1 from the observed perspective image 3 obtained when the reference pattern 2 passes the measuring object 1. Here, the operator device 5 realizes a correlation value calculation means and evaluation means. By this, the optical distortion can be quantified regardless of the kind of the distortion and the quantification of the optical distortion matching to human sense is possible. Since the optical distortion is quantified with correlation operation, a strong distortion which was difficult to quantify with a method measuring the distortion of rectangular elements constituting a grid can also be quantified.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for measuring optical distortion such as perspective distortion and reflection distortion of a transparent plate, and more particularly to perspective distortion of automobile glass and reflection distortion of architectural glass. The present invention relates to a method and an apparatus for evaluating optical distortion suitable for measuring the optical distortion.

[0002]

2. Description of the Related Art A glass sheet for an automobile and a glass sheet for an architectural structure include a flat glass plate and a bent glass plate. In particular, a bent glass plate with a large curvature or a glass plate with poor smoothness may appear distorted when viewing an object through the glass plate or when viewing an object reflected on the glass plate . If the distortion generated when observing the object through the glass plate is too large, the appearance is impaired and the visibility is impaired. Therefore, it is necessary to evaluate whether the distortion caused by the glass plate is within an allowable range. Several methods have been proposed as methods for evaluation.

[0003] One of such methods is to observe a perspective image or a reflection image of an object to be measured in an orthogonal lattice pattern, and to determine the perspective distortion of the perspective image or the reflection image based on the inclination angle of a line segment constituting the lattice. There is a way to quantify the degree. In addition, there is a method of measuring a perspective distortion or a reflected image based on a variation in the shape of a square element constituting a grid. Japanese Patent Application Laid-Open No. 7-20059 describes a method for quantifying the degree of perspective distortion or reflection distortion based on the amount of decrease in the contrast of an observation perspective image or observation reflection image.

[0004]

However, the above-described method has the following problems. In the method of obtaining the inclination angle of the line segment forming the grid, distortion that does not deform the grid in an oblique direction cannot be quantified correctly. Further, in the method of measuring the variation in the shape of the square elements forming the grid, it is necessary to obtain the coordinate values of the grid in the perspective image or the reflection image, and the measurement load is large. In particular, when distortion occurs such that the lattice is greatly deformed, it is difficult to obtain the coordinate value itself, and it is difficult to quantify the distortion.

In the method of quantifying optical distortion based on the amount of decrease in the contrast of a perspective image or a reflected image, the degree of optical distortion can be quantified by simple image processing. For, the degree cannot be quantified correctly.

An object of the present invention is to address such a problem, and an object of the present invention is to provide an optical distortion evaluation method and an evaluation apparatus capable of quantifying optical distortion regardless of the type of distortion.

[0007]

According to the present invention, there is provided a method for evaluating optical distortion, comprising the steps of: irradiating an object with a reference pattern;
A step of observing a perspective image or a reflection image of the reference pattern from the object to be measured, a step of calculating a correlation value between the perspective image or the reflection image and the ideal image, and an optical system of the object to be measured based on the calculated correlation value. Evaluating the distortion.

When calculating a correlation value between a perspective image or a reflection image and an ideal image, a correlation between a local image that is a part of the perspective image or the reflection image and an ideal pattern of the same size as each local image is calculated. A value may be calculated, and a strain existing area in the measured object may be specified from each calculated correlation value.

Then, a perspective image or a reflection image of the reference pattern from the object to be measured is observed by an image pickup means having a plurality of pixels, and the gradation value of the observed perspective image or the reflection image pixel and the ideal image pixel are compared. The correlation value may be calculated using the gradation value.

An apparatus for evaluating optical distortion according to the present invention includes an image pickup means for picking up a perspective image or a reflection image of a reference pattern by an object to be measured, and a perspective image or a reflection image obtained by inputting a perspective image or a reflection image from the image pickup means. The apparatus comprises a correlation value calculating means for calculating a correlation value with an image, and an evaluation means for evaluating optical distortion of the device under test based on the correlation value by the correlation value calculating means.

[0011]

Embodiments of the present invention will be described below. FIG. 1 is a conceptual diagram showing an embodiment of an optical distortion evaluation apparatus according to the present invention. As shown in the figure, a reference pattern 2 from a light source is received by an imaging device 4 after transmitting through a DUT 1 such as a plate glass for which optical distortion is measured.
The observation perspective image 3 from the imaging device 4 is input to an arithmetic device 5 such as a computer. The arithmetic unit 5 evaluates the optical distortion of the DUT 1 from the observation perspective image 3 obtained by passing the reference pattern 2 through the DUT 1. Here, the arithmetic unit 5 implements a correlation value calculation unit and an evaluation unit.

As the reference pattern 2, a regular pattern such as a checker pattern, an orthogonal lattice pattern or a dot pattern is used. As the imaging device 4, any device that can input the observation perspective image 3 as image data, such as an area camera, a line camera, a video camera, a still camera, and an array of photo sensors, may be used.

Next, the operation will be described with reference to the flowchart of FIG. 2 and the explanatory diagram of FIG. Arithmetic unit 5
Inputs the observation perspective image 3 from the imaging device 4 (step S
1) A correlation operation is performed between the input observation perspective image 3 and the ideal perspective image (step S2).

FIG. 3 is an explanatory diagram for explaining a specific example of the correlation operation. In FIG. 3, the observation perspective image 3 is
It is represented as an image of (m × n) pixels. This image is expressed as f (xi, yi). Here, f represents the gradation value of the pixel, and xi and yi represent the pixel position in the image. The arithmetic unit 5 extracts a local image 6 of (u × v) pixels from the observation perspective image 3. And the extracted local image 6
And the correlation value C between the local image 6 and the ideal perspective image 7 of the same size are calculated based on the equation (1).

[0015]

(Equation 1)

In the equation (1), g (xi, yi) represents the local image 6 extracted from the observation perspective image 3, and
(Xi, yi) represents an ideal perspective image 7 of the same size as g (xi, yi). Equation (1) is obtained by multiplying the gradation value between the local image 6 and the ideal perspective image 7 for each corresponding pixel, and calculating the sum of the multiplication results. It is normalized by dividing by the sum of the square values. In addition,
When each pixel is binarized, the denominator of equation (1) is (u ×
v). Accordingly, the correlation value C increases as the distortion of the local image 6 decreases, that is, as the local image 6 approaches the ideal perspective image 7.

The arithmetic unit 5 performs the above-described correlation operation by:
This is executed for the entire observation perspective image 3 of (m × n) pixels. Then, by evaluating the optical distortion of the entire observation fluoroscopic image 3 based on the obtained average value or maximum value of the correlation values C, the optical distortion is evaluated over the entire DUT 1 (step S3).

For example, when the average value is smaller than a predetermined reference value, it can be evaluated that distortion exists in the DUT 1. Alternatively, when the minimum value is less than a predetermined reference value, it can be evaluated that distortion exists in the DUT 1. If there is a correlation value C that is significantly lower than the average value, it can be evaluated that there is a large distortion in a portion corresponding to the local image 6 indicating the correlation value C. Alternatively, it can be evaluated that there is a large distortion in a portion corresponding to the local image 6 indicating the correlation value C indicating the minimum value lower than the reference value.

The evaluation of the optical distortion in step S3 may be performed by the evaluator based on the result of the correlation calculation, or the arithmetic unit 5 may calculate the average or maximum value of each correlation value C and the reference value. May be compared and a comparison result may be output.

As described above, in this embodiment, the optical distortion of the DUT 1 is evaluated based on the correlation value between the pattern transmitted through the DUT 1 and the ideal pattern. Regardless of the direction of distortion, the distortion can be detected.

As the ideal perspective image 7, a perspective image of the DUT 1, which has been confirmed to have an ideal shape without distortion, and the reference pattern 2 were observed directly without passing through the DUT 1. The observation image in that case may be used. When the reference pattern 2 is a periodic pattern, a local image adjacent to the local image 6 to be calculated may be used as the ideal perspective image 7.

Further, in this embodiment, a case has been described in which an observation perspective image 3 based on a pattern transmitted through the DUT 1 is handled. The method and the device according to the invention can likewise be applied.

[0023]

Embodiments of the evaluation method according to the present invention will be described below. [Example 1] Assume that an observation perspective image 3 as shown in FIG. 4 is obtained. As shown in FIG. 4, this perspective image has distortion in which the checker pattern expands and contracts only in the left-right direction. The observation perspective image 3 was an image of (512 × 512) pixels, and the local image 6 was an image of (8 × 8) pixels.
In calculating the correlation value C, a signal value of -1 was given to a black area and a signal value of +1 was given to a white area in the checker pattern. Then, the higher the correlation between the two images, the more the value of the correlation value C becomes 1.
, And the value of the correlation value C approaches 0 as the correlation between the two images is lower.

FIG. 5 is an explanatory diagram showing a calculation result of the correlation value C. In the figure, a white portion indicates a portion where the correlation value is large in the calculation result, and a black portion indicates a portion where the correlation value is small. As is clear from FIG. 5, the correlation value obtained by the method according to the present invention is small in the portion where the distortion is large in the observation perspective image 3. in this way,
In the method according to the present invention, the region where the distortion exists in the observation perspective image 3 can be specified, and the degree of the distortion can be quantified by the magnitude of the correlation value. It is to be noted that the conventional method for obtaining the inclination angle of the line segments constituting the grid cannot correctly quantify the distortion as shown in FIG.

[Embodiment 2] In Embodiment 1, each pixel of the observation perspective image 3 and the ideal perspective image 7 is binarized (black: -1, white: +
After 1), the correlation value C of both images is calculated. However, if each pixel of the observation perspective image 3 and the ideal perspective image 7 is multi-valued, the correlation value can be calculated with higher accuracy. As a result, the region where the strain exists in the DUT 1 can be specified with higher accuracy.
The degree of distortion can be quantified more accurately.

For example, when quaternary processing is performed, +0.75 for a white area, -0.75 for a black area, +0.25 for a black adjacent area in a white area, and -0 for a white adjacent area in a black area. .25. That is, as for the ideal perspective image 7, as shown in a part of the checker pattern illustrated in FIG. 6, each of the white region 8, the black region 10, the black adjacent region 9 in the white region, and the white adjacent region 11 in the black region. , +0.75, -0.75, +0.25, -0.
A signal value of 25 is given.

The observation perspective image 3 also has +0.75 and −0.75 respectively for the white area, black area, black adjacent area in the white area, and white adjacent area in the black area detected from the observed image.
Signal values of 0.75, +0.25, and -0.25 are given. Then, the correlation between the observation perspective image 3 and the ideal perspective image 7 is calculated using the equation (1).

When the quaternary processing is performed, the signal level is multi-valued, so that the degree of distortion can be quantified more accurately. In this embodiment, a signal level intermediate between the white area and the black area is given to the black and white boundary area.
It is possible to perform quantification by removing the influence of unnecessary moire distortion generated in the binarization processing.

[Embodiment 3] Next, the correspondence between the result of quantifying optical distortion by the method according to the present invention and the human sense of distortion will be described. Here, the relationship between the result of the threshold experiment by the subject and the optical distortion quantitative value by the present method was examined.

In the threshold experiment, a sinusoidal distortion in which the intersection point in the orthogonal grid pattern was moved by a sin function as shown in FIG. 7 was applied to the orthogonal grid pattern as shown in FIG. The one with distortion was used. Then, the strain pattern was presented to the subject by changing the strength of the strain, and a question was asked as to whether or not the subject looked distorted. In general, a pattern with a higher distortion has a higher probability of responding that the pattern is distorted, and a pattern with a lower distortion has a lower probability of responding that the pattern is distorted.
Here, assuming that the distribution of thresholds at which distortion starts to be sensed is a normal distribution, the probability (perceived probability) of responding that the object is distorted is converted into a Z score and used as the amount of distortion sensation.

On the other hand, the distortion pattern used in the above threshold experiment is expressed as an image of (512 × 512) pixels, and the local image 6 is an image of (8 × 8) pixels. The correlation with the target pattern was obtained. Then, the average value of the correlation values calculated for each local image 6 was used as the optical distortion quantitative value of the distortion pattern. Here, a signal value of -1 was given to a black area and a signal value of +1 was given to a white area in the pattern.

FIG. 9 shows the distortion perception amount (Z) obtained in the threshold experiment.
FIG. 4 is an explanatory diagram showing a relationship between a score) and an optical distortion quantitative value obtained by a method according to the present invention. In FIG. 9, the horizontal axis represents (1
-The average value of each correlation value) in the image. The larger the value, the smaller the correlation value, that is, the lower the correlation between the ideal image and the local image 6. As is clear from FIG. 9, the smaller the correlation value of the distortion pattern having the sine wave distortion and the distortion pattern having the linear distortion is (the more the rightward on the horizontal axis in FIG. 9),
The value of the Z score increases. That is, the probability that the subject answers that the subject is distorted is increased.

Therefore, for both the distortion pattern having a sine wave distortion and the distortion pattern having a linear distortion, a good correspondence between the optical distortion quantitative value obtained by the method according to the present invention and the distortion sensation amount is found. Was confirmed. This indicates that the method according to the present invention is a method capable of not only quantifying optical distortion irrespective of the type of distortion but also performing distortion measurement suited to human senses.

[0034]

As described above, according to the present invention, a method and an apparatus for evaluating optical distortion are provided for observing a see-through image or a reflection image of an object to be measured of a reference pattern, and observing the observed see-through image or reflection image. And calculated the correlation value between the ideal image, and configured to evaluate the optical distortion of the measured object based on the calculated correlation value, so that the optical distortion can be quantitatively evaluated regardless of the type of distortion, There is an effect that quantification of optical distortion suitable for human sense can be realized. Further, since the optical distortion is quantified by the correlation operation, the strong distortion, which was difficult to quantify by the method of measuring the deformation of the square element forming the grid, can be quantified.

The perspective image or the reflection image of the reference pattern by the object to be measured is observed by the imaging means having a plurality of pixels, and the gradation value of the observed perspective image or the reflection image pixel and the gradation value of the ideal image pixel are observed. In the case where the correlation value is calculated by using and, the correlation value is calculated by using the gradation value, so that the difference between the perspective image or the reflection image and the ideal image is more accurately converted into the correlation value. It is preferable because it can be reflected.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram showing an embodiment of an optical distortion evaluation device according to the present invention.

FIG. 2 is a flowchart showing a method for evaluating optical distortion according to the present invention.

FIG. 3 is an explanatory diagram for describing a specific example of a correlation operation.

FIG. 4 is an explanatory diagram showing an example of an observation perspective image.

FIG. 5 is an explanatory diagram showing a correlation value calculation result using the observation perspective image shown in FIG. 4;

FIG. 6 is an explanatory diagram showing an example of signal level assignment in the case of quaternary processing;

FIG. 7 is an explanatory diagram showing a pattern obtained by giving a sinusoidal distortion to an orthogonal lattice pattern.

FIG. 8 is an explanatory diagram showing a pattern obtained by applying a linear distortion to an orthogonal lattice pattern.

FIG. 9 is an explanatory diagram showing a relationship between a Z score obtained by a threshold experiment and a quantitative optical distortion value obtained by a method according to the present invention.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 DUT 2 Reference pattern 3 Observed perspective image 4 Imaging device 5 Arithmetic unit 6 Local image 7 Ideal perspective image 8 White region 9 Black adjacent region in white region 10 Black region 11 White adjacent region in black region

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Hiroaki Shimozono 1150 Hazawa-cho, Kanagawa-ku, Yokohama-shi, Kanagawa-ken

Claims (4)

[Claims] 1. A method for evaluating optical distortion of an object to be measured, comprising: irradiating a reference pattern to the object to be measured; observing a see-through image or a reflection image of the reference pattern from the object to be measured; An optical distortion evaluation method for calculating a correlation value between the obtained perspective image or reflection image and an ideal image, and evaluating optical distortion of the device under test based on the calculated correlation value. 2. When calculating a correlation value between a perspective image or a reflection image and an ideal image, a local image that is a part of the perspective image or the reflection image;
2. The method for evaluating optical distortion according to claim 1, wherein a correlation value between each local image and an ideal pattern of the same size is calculated, and a distortion existing area in the measured object is specified from each calculated correlation value. 3. A perspective image or a reflected image of a reference pattern from an object to be measured is observed by an image pickup means having a plurality of pixels, and a gradation value of the observed perspective image or a reflected image pixel and a pixel of an ideal image are observed. The method for evaluating optical distortion according to claim 1, wherein a correlation value is calculated using the gradation value. 4. An image pickup means for picking up a perspective image or a reflection image of a reference pattern by an object to be measured, and inputting the perspective image or the reflection image from the image pickup means and correlating the perspective image or the reflection image with an ideal image. An optical distortion evaluation device, comprising: a correlation value calculation unit that calculates a value; and an evaluation unit that evaluates optical distortion of the device under test based on the correlation value obtained by the correlation value calculation unit.