JPS62174606A - Dimension measuring instrument - Google Patents

Abstract

PURPOSE:To improve accuracy of dimensional measurement without being influenced by noise by approximating functionally analytically a video signal including a noise component in the neighborhood of a slice level. CONSTITUTION:A fall part and a rise part of the video signal of a round bar with a diameter (d) are made a gentle waveform 20. A bit (b) number between a fall and a rise is measured and the coefficients alpha and beta are calculated using a relational expression d=alphab+beta in a slice level value set optionally using two round bars whose dimensions are known in advance. Further, the video signal 20 is standardized using a maximum value and a minimum value of the signal 20 in order to standaridized the slice level. In other words, the signal 20 is processed on an area A between a preset point 22 which is made a voltage level higher than the slice level and a point 23 which is made the voltage level lower than the slice level. Especially, the signal 20 is processed so that an error is absorbed on the average in the neighborhood of the slice level. Then, the signal 20 can be approximated to a straight line. Then, the actual dimensions are calculated by substituting the bit number (b) of a difference between the calculated respective address values of the fall part and the rise part for the relational expression.

Description

[Detailed description of the invention] [Industrial application field] The present invention relates to a non-contact optical dimension measurement method.

[Conventional technology]

Conventionally, the most commonly used non-contact optical dimension measurement method is to irradiate the object to be measured with light and use a lens system to magnify the image produced by the object to be measured using a CCD image sensor.
A video signal is generated by receiving light at
CD image sensor li! ii corresponds to the cumulative number)
was detected and the size was measured from the number of bits of the difference between two address numbers.

[Problem that the invention seeks to solve]

When the CCD image sensor receives an image, the CCDC
C Image sensor non-uniformity of each pixel, such as pixel size and iIl! If there are variations in the Ji element pitch, variations in detection sensitivity for the incident light intensity of each pixel, etc., the video signal's 'E pressure' will change depending on the degree of the above-mentioned variation. There are fluctuations in the video signal due to quantization errors during D conversion, and image distortion due to optical system aberrations, interference, countermeasures, and the like.

When measuring dimensions from a video signal modulated by such various factors, if the video signal is modulated near the slice level voltage, which is the standard for binarization, it will be difficult to measure the falling edge when binarized. This has the disadvantage that the address value at the rising edge also changes, making it impossible to measure accurate dimensions.

SUMMARY OF THE INVENTION An object of the present invention is to provide a method for solving the above-mentioned conventional problems and performing highly accurate dimension measurement from a video signal modulated by disturbance factors.

[Means for Solving the Problems] In order to achieve the above object, the present invention comprises the following method.

The object to be measured whose dimensions are to be measured is irradiated with light, the image generated by the object to be measured is detected by a CCD image sensor to generate a video signal, and each of the falling and rising portions of the video signal due to the image is detected. , CCDC in a preset area across the slice level for binarization.
The address value of the C image sensor pixel is the independent variable X, and the output voltage of the video signal is the dependent variable y.The function is expanded into a function having a predetermined order, and the dependent variable value yo of the function corresponding to the slice level is Independent variable value Xo
is calculated, and the size is measured from the number of bits of the difference between the two calculated values of Xo.

[Effect]

When measuring dimensions using the above method, the video signal containing various noise components must be set at a certain level from a high voltage region to a low voltage region, especially across the voltage at the slice level where binarization is performed. Regarding the area where
The function form of a preset order such as a linear function or a quadratic function is expanded to y2f (xi) using a method such as the method of least squares. Generally, the noise component is not very large and the signal component is compared to the noise component In the case of smoothness, it is possible to minimize errors due to variations in each measurement value by performing function approximation.Furthermore, when determining the pixel address of the CCD image sensor that is equal to the voltage value of the slice level, Function y = f (x = determined by xi
g (just calculate yl, CCD image sensor -
Since addresses can be determined at a pitch finer than the pixel pitch of , the pitch between addresses between the falling and rising edges of the video signal can be determined with high precision.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a diagram showing the method of measuring dimensions of the present invention.

10 is the object to be measured whose dimensions are to be measured, 11 is the object to be measured 10
12 is a light receiving section including a light receiving optical system for generating an image generated by the object to be measured 10; 13 is a CCD image sensor 1;
14 is an analog video signal output from the CCD image sensor. 15 is an A/D conversion unit that digitizes the analog video signal 14; 16 is a slice level for performing binarization processing on each of the falling and rising parts of the digital video signal 151 generated by the object to be measured 1o; The video signal 151 is calculated in a preset area from the side where the voltage value is higher than the slice level of the video signal 151 to the side where the voltage value is lower than the slice level, sandwiching the values of the video signal 151. This is a functional calculation unit that expands into a functional form having an order.

This function calculation unit 16 uses a function y of a certain set order, with the address value of the pixel of the CCD image sensor 16 as an independent variable X and the output voltage of the video signal 151 as a dependent variable y.
2. Expand to (x).

When the degree is 1, the function is a linear equation y=ax'',b, and when the degree is 2, the function is a quadratic equation y2ax2+bx-1-c.

However, a, b, and c are constant coefficients of the function y = r (xl). 17 is the aforementioned function y = f that is equal to the voltage value yo of the slice level for binarization in the address calculation section.
(Independent variable value X corresponding to dependent variable value yo of xl.

is calculated for each of the falling and rising parts of the video signal 151. The independent variable value Xo calculated here
is the address value of the pixel serving as the slice level. 18
is a dimension conversion unit that calculates the at1/s value (
The size is calculated from the number of bits), and the size is determined using a size conversion coefficient K that converts the number of bits to the actual size, for example, μm.

FIG. 2 shows an example of a waveform diagram of a video signal, and the details of the operation will be explained.

Reference numeral 20 indicates a falling portion and a rising portion of a video signal generated by a round rod having a diameter do when the object to be measured 10 in FIG. 1 is a round rod.

In general, when the optical system described below is used, the rising and falling portions are caused by light interference, diffraction effects, and leakage of photocarriers between pixels of a CCD sensor, so that they do not have a clean waveform like the pulse waveform of an electrical signal. The result is a waveform that changes smoothly.

Reference numeral 21 denotes a slice level serving as a reference voltage level when performing binarization. The criteria for setting the slice level 21 may be arbitrary. Using two reference samples (diameters d+, d2) whose dimensions are known in advance, the number of bits (bl, b2) between the falling edge and the rising edge is measured by the method of the present invention at a certain arbitrarily set slice level value. Then, the relational expression d-α·b+β between the size and the number of bits is calculated. Using the coefficients α and β of this relational expression, the number of bits, which is a measured value, is converted into a dimension. Furthermore, in order to standardize the slice level value, the video signal may be normalized using the maximum value and minimum value of the video signal. The explanation of FIG. 2 will be made regarding the falling portion of the waveform. 22 is a preset point where the voltage level is higher than the slice level, 23 is a preset point where the voltage level is lower than the slice level, and the video signal is processed for the area A between the points 22 and 26. . As mentioned above, video signals contain various noise components, and if they are binarized using their original form, errors will occur.Especially near the slice level 210, the errors will be absorbed on average and the signal will become monotonous. The waveform is processed to become a decreasing or increasing waveform. For example, when slice level 21 is a point with an intensity of 15% of the standardized video signal, point 22 is 20%, point 26
When the voltage value is 10%, the video signal 20 can be sufficiently approximated to a straight line in a region where the width of the video signal intensity is 10%.

When approximating the straight line y=a x shown in 24 to
The error of x+b can be minimized.

If the video signal has large noise in region A, numerical smoothing processing such as a moving average method may be performed in advance, and then linearization function expansion may be performed.

When the slice level value is 'io for the function y = ax -) b mentioned above, x o = (y o
-b)/a The address value Xo when giving the slice level yO can be found.

Generally, the pixels of a CCD image sensor are not continuously distributed but are distributed discretely, so the voltage level of the obtained video signal also shows discrete changes, so between pixel i and pixel i+1 It was necessary to obtain the voltage value by interpolating yl and yij-1, but the method according to the present invention equivalently interpolates the voltage between pixels to an infinitesimal value.

The actual size is calculated by substituting the number of bits of the difference between the address values Xo+ and Xo2 of each of the falling and rising parts obtained as above into the above-mentioned relational expression d-αb+β.

Here, a major factor affecting measurement accuracy is the magnification of the optical system. That is, the measurement accuracy improves as the magnification increases.

FIG. 3 shows an embodiment of the optical system of the light transmitting section 11 and the light receiving section 12.

In the light transmitting section 11, 60 is an I-1c-N e laser, 61 is a cylindrical lens with a focal length of "i," and 32
is a plano-convex lens with a focal length of f2. By the combination of this cylindrical lens 61 and plano-convex lens 62, I
-1e-N The circular beam shape emitted from the e laser is converted into an elliptical beam that is approximately one-dimensionally expanded. Note that the reason why a laser beam is used as a light source is that the amount of irradiated light, that is, the intensity, is stable over a long period of time, and the S/N ratio of brightness and darkness of the video signal is good. The object to be measured 10 is placed at a position at the rear focal length of the plano-convex lens 62. In the light receiving unit 12, 63 is a plano-convex lens with a focal length r3;
is a plano-convex lens with a focal length r4, and 35 is a diameter P. A spatial filter having a filter is installed at the confocal position of the plano-convex lenses 66 and 64. 36 is a cylindrical lens with a focal length of f, and 67 is a plano-convex lens with a focal length of 6, which are placed at confocal positions with each other.

A CCD image sensor 16 is installed at the rear focal position of the plano-convex lens 67.

In the light receiving section 12 described above, the spatial filter 65 operates as a so-called low-pass filter, and is effective in improving the S/N ratio of the video signal because it cant the influence of high-order diffracted waves due to the diffraction effect.

Also, the magnification determined by the plano-convex lenses 66 and 64 '<
/'3, cylindrical lens 66 and plano-convex lens 67
By adding the combination of f4/r
It can be greatly enlarged to a magnification of 3.times.re/I's, and if a short focal length is used for the cylindrical lens 66, a high magnification optical system with a short optical path length can be easily realized. By the method described above, the magnification of the optical system is 20
When a 4096-pixel CCD image sensor was used, results were obtained in which dimensions could be measured with an accuracy of 0.5 μm or less when measuring diameters in the range of 50 μm to 500 μm.

〔Effect of the invention〕

As is clear from the above explanation, by analytically approximating a video signal containing noise components to a function near the slice level, pitches below the inherent pitch of pixels of a CCD image sensor can be measured without being affected by noise. This makes it possible to easily achieve highly accurate dimensional measurements.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing the dimension measurement method of the present invention, Fig. 2 is a waveform diagram showing an example of a video signal pattern, and Fig. 3 is an optical path showing an example of the configuration of an optical system for measuring dimensions. It is a diagram. 11... Light transmitting section, 12... Light receiving section, 1
3...CCD image sensor-116...
...Function operation section, 17...Address operation section, 18...Dimension operation section, 30...=...1-1 e-Nev-the, 6
5... Spatial filter. Figure 1 Figure 20

Claims (1)

[Claims] The object to be measured whose dimensions are to be measured is irradiated with light, the image generated by the object is received by a CCD image sensor to generate a video signal, and the video signal is binarized using a set slice level and converted into two. In the dimension measuring method, the dimension is measured by creating a digitized signal, detecting address numbers of pixels at falling and rising edges of the binary signal, and counting the number of bits of the difference between the address numbers. For each of the falling and rising parts of the video signal caused by the measurement object, in a preset area across the slice level, the address value of the pixel of the CCD image sensor is set as an independent variable, and the output voltage of the video signal is set as an independent variable. is expanded into a functional form having a predetermined order with as a dependent variable, and an independent variable value corresponding to a dependent variable value of the function corresponding to the voltage value of the slice level is calculated, and the calculated independent variable A dimension measurement method characterized by measuring dimensions by counting the number of bits of a difference between address values.