JP2006112964A - Optical sensor device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an optical sensor device having a simple structure, capable of easily and accurately detecting the diameter of a round-bar object. <P>SOLUTION: Two sets of optics are prepared, each of which is constituted of a pair of a point source and a linear image sensor disposed opposite to the point source, and these two sets of optics are arranged, such that their optical axes are orthogonal to each other. Then, the positional coordinates of the round-bar object which is disposed in the optics, is obtained from the central position of a shadow form of the round-bar object detected by the respective linear image sensors, and a measurement scale factor for the optics is determined, in accordance with the positional coordinates, and the diameter of the round-bar object is computed accurately from the width 2a of the shadow form of the round-bar object detected, from the outputs of the liner image sensors, in accordance with the measurement scale factor. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an optical sensor device that can easily and accurately detect the diameter of a round bar, such as a drill blade, by directly using emitted light having a predetermined divergence angle emitted from a point light source. .

  Recently, the printed circuit board has been multi-layered with high density mounting, and through holes are formed in the printed circuit board to electrically connect a plurality of layers. Such a through hole is exclusively formed by using a fine drill blade having a diameter of about 50 to 100 μm, for example, and rotating the drill blade at a high speed to form a hole of a predetermined depth in the printed circuit board. At this time, of course, it is possible to select and use a drill blade of a predetermined diameter, attach the drill blade to the chuck of the drill without any runout, and further accurately grasp the tip position of the drill blade as desired. It is important to drill holes to the depth of However, it is generally very difficult to mechanically measure the diameter (drill diameter) of this type of small-diameter drill blade and to mechanically check the mounting state on the chuck. Optical measuring means are used (see, for example, Patent Documents 1, 2, and 3).

  However, the optical measurement method of a drill blade as shown in Patent Documents 1, 2, and 3 measures the light shielding width by a line sensor (linear image sensor) or the like using light shielding of the drill blade. Therefore, it was difficult to accurately measure the diameter of a fine drill blade having a diameter of 200 μm or less. That is, monochromatic parallel light such as laser light is exclusively used as the light source for this type of measurement. However, since diffraction of light occurs at the edge portion shielded by the drill blade, there is a problem that it is difficult to accurately measure the diameter of the drill blade due to the influence of this diffraction.

Therefore, the present inventor uses an approximate expression that approximates the diffraction pattern (intensity distribution) of the light that has been previously subjected to Fresnel diffraction using the hyperbolic second function sech (x), and easily and accurately obtains the edge position. A method was proposed (see, for example, Patent Document 4).
JP 2003-170335 A Japanese Patent Application Laid-Open No. 7-306020 JP 7-260425 A JP 2004-177335 A

  By the way, in the conventional optical measurement method described above, monochromatic parallel light such as laser light is exclusively used as the light source. Specifically, the laser light emitted from the laser diode (LD) is converted into a parallel light beam via an optical lens (collimator lens), and the parallel light beam (monochromatic parallel light) is directed to the light receiving surface of the line sensor. Projecting. The line sensor detects the shadow of the detection object (shielding object) positioned in the optical path and the Fresnel diffraction light pattern generated at the edge of the detection object.

  However, since the light source for projecting monochromatic parallel light requires an optical element such as the above-described optical lens (collimator lens), the configuration becomes large and the apparatus cost increases. Moreover, when trying to widen the beam bundle width of monochromatic parallel light, for example, a large-diameter projection lens is required, and generally, it is necessary to set a long optical distance between the LD and the projection lens. There are problems such as an increase in the size of the light source.

  The present invention has been made in view of such circumstances, and its purpose is to specifically emit radiation having a predetermined divergence angle emitted from a point light source without incurring a complicated configuration of the light source. An object of the present invention is to provide an optical sensor device that can easily and accurately obtain the diameter and position coordinates of a round bar, such as a drill blade, using light as it is.

  In order to achieve the above-described object, the present invention eliminates the need for an optical lens by using, for measurement, radiation light having a predetermined divergence angle emitted from a point light source such as a laser diode (LD). In addition to greatly simplifying the system, the error caused by the displacement of the detection target (round bar) caused by using the above-mentioned synchrotron radiation having a predetermined divergence angle is considered to be a geometrical characteristic of the optical system. By correcting based on the relationship, error distribution, etc., the diameter of a round bar such as a drill blade and its position coordinates are obtained with high accuracy.

That is, the optical sensor device according to the present invention is basically as described in claim 1.
(a) First and second linear elements each having a light receiving surface in which a plurality of light receiving cells are arranged in a straight line, and arranged so that the light receiving cells are arranged in different directions and the visual field areas of the respective light receiving surfaces intersect each other. An image sensor;
(b) a first point light source and a second point light source that respectively emit monochromatic light having a divergence angle that is disposed opposite to each linear image sensor and reaches the entire light receiving surface of each linear image sensor;
(c) The output of each linear image sensor indicates the center position of the shadow generated on the light receiving surface of each linear image sensor and the width of the shadow by the round bar positioned between the point light source and the linear image sensor. Shadow analysis means to analyze and obtain each;
(d) A distance between each linear image sensor and the round bar is obtained according to the center position of the shadow obtained by each linear image sensor, and the distance is obtained by each linear image sensor. A diameter calculating means for obtaining a diameter of the round bar from the shadow width is provided.

Specifically, two sets of an optical system composed of a pair of a point light source and a linear image sensor arranged to face the point light source are prepared, and the optical axes of these two sets of optical systems are mutually different. Place them so that they intersect at right angles. When the detection target (round bar) is positioned in the optical path of the optical system, the center of the shadow generated by the detection target (round bar) detected by each linear image sensor is within the optical path. The position coordinates of the detection target (round bar) are determined, and the distance WD between each linear image sensor and the detection target (round bar) is determined according to the position coordinates. The round bar according to the geometrical relationship between the opposing distance SD between the point light source and the linear image sensor, the separation distance WD, and the shadow width 2a of the round bar detected from the output of each linear image sensor. For example, 2r = 2a (SD−WD) / {(2a) ² + SD ² } ^1/2
It is characterized by obtaining with high accuracy.

Incidentally, the shadow analyzing means analyzes the output of each linear image sensor using an approximate expression of Fresnel diffraction as described in claim 2, and detects the edge position X _R , X _L of the light receiving pattern by each linear image sensor. And the center position (X _R + X _L ) / 2 of the shadow and its shadow width 2a (= X _R −X _L ) are respectively determined from these edge positions. Further, in the diameter calculating means, it is preferable that the shadow for correcting the shadow width detected by the linear image sensor in accordance with the center position of the shadow detected by each of the linear image sensors. It is desirable to provide width correction means.

An optical sensor device according to the present invention is as described in claim 4.
(e) The shadow of the round bar (reference pin) measured when a round bar (reference pin) with a known diameter is provided at a predetermined position between the point light source and the linear image sensor. Parameter storage means for determining a parameter for correcting the measurement span of the linear image sensor from the width of the linear image sensor and storing the parameter;
(f) Thereafter, when measuring the diameter and / or position of the round bar (measurement object), the measurement span for correcting the shadow width respectively obtained from the output of the linear image sensor using the parameters stored in the parameter storage means And a correction unit.

The predetermined position between the point light source and the linear image sensor is, for example, an opposing distance SD ₁ between the first point light source and the first linear image sensor, as described in claim 5. The parameter for correcting the measurement span of the linear image sensor is determined at that time by setting the opposing distance SD ₂ between the second point light source and the second linear image sensor as an intermediate point that bisects each. What is necessary is just to obtain | require as a coefficient which correct | amends the arrangement | positioning pitch w of the light receiving cell in the said linear image sensor so that the said shadow width 2a may be twice the diameter 2r of the said round bar (reference | standard pin).

In addition, when radiation light having a predetermined divergence angle emitted from a point light source is used, a measurement error depending on the arrangement pitch w of the light receiving cells in the linear image sensor cannot be denied. Since it depends on the distance WD between the linear image sensor and the round bar, for example, as described in claim 6
(g) A distance between the other linear image sensor and the round bar obtained based on the center position of the shadow obtained by one linear image sensor, and the distance obtained by the other linear image sensor. According to the shadow width of the round bar, an error pattern of the shadow edge position depending on the arrangement pitch of the plurality of light receiving cells in the other linear image sensor is obtained, and the output of the linear image sensor is analyzed according to the error pattern. It is desirable to provide edge position correcting means for correcting the edge position of the received light reception pattern to determine the edge position of the shadow.

Preferably, in the above-described optical sensor device as described in claim 7,
It is also useful to have verification means for comparing the diameters of the round bars obtained from the outputs of the first and second linear image sensors with each other and verifying the suitability of correction for the outputs of the linear image sensors. It is.
Furthermore, in the above-described optical sensor device according to claim 8, the shadow width required when the round bar positioned between the point light source and the linear image sensor is not rotating, and the round bar A center blur detecting means for detecting a center blur amount during rotation of the round bar according to a shadow width required when the body is rotating is provided, and from each output of the first and second linear image sensors. You may make it determine the appropriateness | suitability of a core blur detection by mutually comparing the amount of core blur calculated | required each.

  It is also useful to provide coordinate output means for outputting the center position of the round bar obtained by analyzing the outputs of the first and second linear image sensors, respectively.

  According to the optical sensor device of the present invention, it is only necessary to use monochromatic light having a predetermined divergence angle emitted from a point light source as it is, and it is not necessary to form a parallel light beam using an optical lens. Can be greatly simplified. In addition, the two optical units arranged with their optical axes substantially orthogonal to each other detect the positions on the linear image sensor of the shadows that are shielded from light by the round bar, so the optical positions are detected from these detection positions. The detection position (coordinates) of the round bar in the system can be specified geometrically. Further, the round bar is geometrically determined from the width of the shadow detected by the linear image sensor in accordance with the optical positional relationship between the detection position (coordinates) of the round bar and the point light source and the linear image sensor. Since the diameter of the body is obtained, it is possible to obtain the diameter with high accuracy without regard to the detection position (position shift) of the round bar in the optical system.

  That is, when measuring the diameter of a round bar using synchrotron radiation, a measurement error occurs if the detection position is shifted. However, if it is known that the detection position of the round bar has shifted, the measurement value can be corrected based on geometric calculations, error distribution, etc. If the position (coordinates) of the axis of the round bar can be measured accurately, good. Therefore, in the present invention, two pairs of lasers (point light sources) and linear image sensors are arranged at right angles, the axial center position of the round bar is monitored, and the positional deviation (coordinates) is measured two-dimensionally. Based on the information, the measured values are corrected according to geometric calculation and error distribution.

  By the way, the axis position of the round bar is not affected by synchrotron radiation because it is only necessary to find the position of the edge on both sides in the shadow and find the middle position, and it is relatively accurate regardless of where the round bar is located. It can be measured well. If the error due to the positional deviation is corrected accurately, the diameter of the round bar and the amount of runout measured by the two optical units described above should match each other. Thus, it can be determined whether or not the measurement (correction) is accurate. In addition, according to such measurement, not only the diameter and runout measurement of the round bar, but also the detection position coordinates can be accurately detected, so that the origin on the XY plane can be detected. Further, if the round bar is moved back and forth in the axial direction, and the tip position of the round bar is detected from the change in the output of the linear image sensor at that time, the three-dimensional coordinates of XYZ can be detected.

  Even if the two optical systems described above (the optical axis of the radiated light) are not arranged at right angles, if the span of the measurement system is adjusted during installation, the installation error of the optical system can be simplified. It can be corrected. Therefore, even if radiated light emitted from a point light source is used as it is, it is possible to perform highly accurate optical measurement, and its practical advantage is coupled with the fact that the configuration is simple and can be manufactured at low cost. It is a great deal.

Hereinafter, an optical sensor device according to an embodiment of the present invention will be described with reference to the drawings.
As shown in FIG. 1, the optical sensor device is arranged so as to be opposed to the first and second point light sources 1x and 1y with a predetermined distance SD from each of the point light sources 1x and 1y. First and second linear image sensors 2x and 2y are provided. The first and second point light sources 1x and 1y are formed of laser diodes (LD) that emit monochromatic light (laser light) having a predetermined divergence angle. In the first and second linear image sensors 2x and 2y, as shown in FIG. 2, a plurality of light receiving cells 3 are linearly arranged at a predetermined pitch W (for example, 85 μm) to form a light receiving surface having a predetermined width. It consists of a so-called line sensor such as a CCD.

  The first point light source 1x and the first linear image sensor 2x are positioned in the optical path of monochromatic light (laser light) emitted from the first point light source 1x toward the first linear image sensor 2x. A first optical system that detects the detection object (round bar) 4 is configured, and the second point light source 1y and the second linear image sensor 2y are connected to the second linear image from the second point light source 1y. The second optical system is configured to detect the round bar 4 positioned in the optical path of monochromatic light (laser light) irradiated toward the sensor 2y.

  In particular, each of the linear image sensors 2x and 2y in the first and second optical systems has the arrangement direction of the plurality of light receiving cells 3 constituting the light receiving surface orthogonal to each other and intersects the direction in which each light receiving surface is viewed. Is provided. And these 1st and 2nd optical systems detect the detection target object (round bar body) 4 located in the area | region where each said monochromatic light (laser beam) cross | intersects from a X-axis direction and a Y-axis direction, respectively. It has become a thing.

Incidentally, the detection object (round bar body) 4 is detected by partially blocking the monochromatic light emitted from the point light sources 1x and 1y by the round bar body 4 positioned in the optical path. For example, as shown in FIG. 2, a shadow A of the round bar 4 is generated on the light receiving surfaces of the sensors 2 x and 2 y, and this shadow A is detected. In particular, the detection of the shadow A is performed by analyzing the outputs of the linear image sensors 2x and 2y in the measuring units 5x and 5y as described later. Specifically, the edge positions E _L and E _R of the shadow A are detected from the outputs of the linear image sensors 2x and 2y (edge detection processing 6), and the width of the shadow A is determined from these edge positions E _L and E _R. Is calculated as [E _L −E _R ] (shadow width detection process 7), and the center position of the shadow A is determined as [(E _L + E _R ) / 2] from the edge positions E _L and E _R (center). The position detection process 8) is performed.

  In the calculation unit 9, the presence of the detection object (round bar) 4 in the optical path from the position of the shadow A on the linear image sensors 2x and 2y detected by the measurement units 5x and 5y, respectively. The position, that is, the position coordinate (X, Y) in the X-axis direction and the Y-axis direction is detected. Further, in the calculation unit 9, an optical positional relationship (geometric positional relationship) between the detection object (round bar) 4, the point light sources 1x and 1y, and the linear image sensors 2x and 2y obtained from the detection position coordinates. ) And the width of the shadow A detected by the linear image sensors 2x and 2y, as described later, the detection object (round bar) The diameter of 4 is obtained.

  In the above optical system, when the first and second point light sources 1x and 1y are driven to emit light simultaneously, the light reaching the first and second linear image sensors 2x and 2y by the reflected light from the round bar 4 May interfere. In order to prevent such interference, for example, the light emission timing of the point light sources 1x and 1y or the light reception timing of the linear image sensors 2x and 2y may be shifted. Specifically, when measurement is performed by driving each point light source 1x, 1y in a pulsed manner for 4 μs each, the timing of the light emitters of these point light sources 1x, 1y is shifted alternately by 4 μs. The light emission may be driven.

  Here, the detection process of the shadow A by the round bar 4 based on the outputs of the linear image sensors 2x and 2y described above will be described. In the above-described detection process of the edge position of the shadow A, a part of the light emitted from the light source is obtained. Paying attention to the fact that Fresnel diffraction occurs at the edge (edge) when blocked by the shield, the position of the edge (edge) of the shield according to the light intensity distribution on the light receiving surface of the linear image sensor Is detected with high accuracy.

That is, when the shielding object is a plate-like object that blocks the optical path of monochromatic parallel light from one end side in the arrangement direction of the light receiving cells in the linear image sensor, the light intensity distribution due to Fresnel diffraction at the edge of the shielding object is shown in FIG. As shown, it rises steeply in the vicinity of the edge position and converges while vibrating as it moves away from the edge position. Such light intensity distribution characteristics are as follows: the wavelength of monochromatic parallel light is λ [nm], the distance from the edge of the shield to the light receiving surface is z [mm], and the edge position x [μm] on the light receiving surface is [ 0], ∫ is an operation symbol indicating an integration from [x = 0] to [(2 / λz) ^1/2 · x]. Light intensity = (1/2) {[1/2 + S (x) ] ² + [1/2 + C (x)] ² }
S (x) = ∫sin (π / 2) · U ² dU
C (x) = ∫cos (π / 2) · U ² dU
Represented as: However, U is a temporary variable.

As for the functions S (x) and C (x) in the above equation, the Fresnel function is used exclusively as shown in the mathematical formulas, and S (x) ′ ≈ (1/2) − where x is large. (1 / πx) cos (πx 2/2)
C (x) '≒ (1/2 ) + (1 / πx) sin (πx 2/2)
Can be approximated respectively. Therefore, basically, by using the approximate expressions S (x) ′ and C (x) ′, the edge position described above can be calculated from the received light intensity of each light receiving cell of the linear image sensor.

  However, when actually calculated, as shown in FIG. 4, the functions S (x) and C (x) and the approximate expressions S (x) ′ and C (x) ′ are converged after the rise ( Although it approximates very well in the second and subsequent peaks, it cannot be denied that there is a large shift in the first rising portion (first mountain). In particular, the characteristic of the first rising portion plays an important role in edge detection, and the deviation of the characteristic causes a decrease in the detection accuracy of the edge position.

Therefore, the present inventor previously described in Patent Document 4 that the first rising portion of the light intensity distribution on the light receiving surface by Fresnel diffraction of monochromatic parallel light, particularly the distribution characteristics of the first mountain, a, b, c respectively. As coefficient y = a · sech (bx + c)
It was found that the hyperbolic second function sech (x) is very well approximated. Then, the output (light intensity) of the linear image sensor is analyzed using this hyperbolic second function sech (x), and the light intensity (relative value) in the light intensity distribution on the light receiving surface by Fresnel diffraction described above is [0. 25] is proposed as the edge position in the arrangement direction of the light receiving cells of the shield.

Specifically, when the above-described hyperbolic second function is applied to the above-described formula of the light intensity distribution by Fresnel diffraction and approximated to the first rising portion (first mountain) of the light intensity, the hyperbolic second function sech (x ) Is the light intensity = 1.37 sech {1.98 (2 / λz) ^1/2 x-2.39}
As shown. This approximate expression agrees with the theoretical expression of the light intensity distribution with an accuracy of about three digits. Where λ is the wavelength of light [nm], z is the distance from the edge to the light receiving surface [mm], and x is the coordinate [μm] where the edge position on the light receiving surface is [0]. The above coefficients are set under practical units.

An algorithm of edge position detection processing using such a hyperbolic second function sech (x) will be described. When calculating the inverse function of the approximated light intensity using the hyperbolic second function sech (x),
Y = (y / 1.37)
X = 1.98 (2 / λz) ^1/2 x
Anyway,
X = 2.39−ln {[1+ (1−Y ² ) ^1/2 ] / Y}
Can be expressed as

  Therefore, in the edge detection processing described above, basically, for example, according to the procedure shown in FIG. 5, first, the normalized received light intensity y1, y2,... Ym obtained from a plurality (m) of light receiving cells in the linear image sensor. The light receiving cell Cn that has received light intensity greater than the above-mentioned reference light intensity [0.25] and the light receiving cell Cn− that has received light intensity smaller than the above-mentioned reference light intensity [0.25]. 1 is obtained (step S1). That is, two adjacent light receiving cells Cn and Cn-1 having a light receiving intensity of [0.25] in each of the plurality of light receiving cells 1a (C1, C2, to Cm) are obtained. Then, the received light intensity yn, yn-1 of each of the light receiving cells Cn, Cn-1 is divided by the coefficient [1.37] described above to be converted into light intensity Yn, Yn-1 on the XY coordinates ( Step S2).

Thereafter, the positions Xn and Xn-1 on the light receiving surface of the light receiving cells Cn and Cn-1 from which the light receiving intensities Yn and Yn-1 of the light receiving cells Cn and Cn-1 are obtained are expressed by the above-described approximate expression. Xn = 2.39-ln accordance ^{{[1+ (1-Yn 2} ) 1/2] / Yn}
Xn-1 = 2.39-ln {[1+ (1-Yn-1 ² ) ^1/2 ] / Yn-1}
As a result, the relative position on the X-axis is calculated by performing inverse transformation (step S3). Then, as shown in FIG. 6, the difference Δx between the position of the light receiving cell Cn from these positions Xn and Xn−1 and the edge position where the light receiving intensity is [0.25] is expressed as Δx = W · [Xn / (Xn-Xn-1)]
It calculates by the interpolation calculation which becomes (step S4).

The difference Δx is the distance from the edge position xo where the received light intensity is [0.25] to the position of the light receiving cell Cn, and is thus measured from the first light receiving cell C1 on the entire light receiving surface of the linear image sensor. When the absolute position x of the edge is the cell number of the light receiving cell where the light quantity Y2 is obtained and the array pitch of the light receiving cells is W, x = n · W−Δx
Can be obtained as The relative positions Xn and Xn-1 obtained in the inverse transformation are
X = 1.98 (2 / λz) ^1/2 x
As shown below, the value is multiplied by [1.98 (2 / λz) ^1/2 ], but this term is substantially eliminated by obtaining the ratio by the above-described interpolation calculation.

  This interpolation calculation may be executed using the above-described approximate expression. However, if the change in light intensity between the two light receiving cells Cn and Cn-1 can be regarded as linear, it is simple. Simple linear interpolation may be used. Here, a position where the light intensity is [0.25] is found between adjacent light receiving cells, and the two light receiving cells Cn and Cn-1 having the position as the cell boundary are specified. Two or more light receiving cells may be specified. However, in this case, it is only necessary to prevent the calculation accuracy from being lowered by performing the interpolation calculation using the approximate expression described above. Further, the inverse transformation described above can be executed instantaneously, for example, by using a table in which the calculated values are stored in advance, thereby greatly reducing the calculation processing load.

  By the way, when the shield is a rod-shaped body having a very small diameter as described above, for example, a drill blade, Fresnel diffraction of monochromatic parallel light occurs at both sides of the drill blade (round bar body). Therefore, the Fresnel diffraction pattern on the light receiving surface of the linear image sensor 2x (2y) converges while vibrating on both sides of the center position of the drill blade (round bar) as shown in FIG. 7A, for example. The pattern has symmetry. Further, the received light intensity in each light receiving cell of the linear image sensors 2x and 2y is as shown in FIG. Moreover, when the diameter of the round bar (drilling blade) 4 is 200 μm or less, the received light intensity may not decrease to [0.25]. Therefore, in this case, there is a possibility that the position where the light quantity becomes [0.25] cannot be accurately obtained using the approximate expression of Fresnel diffraction as described above.

However, the diffraction pattern shown in FIG. 7 (a) is considered to be a composite of Fresnel diffraction generated on both sides of the shield (round bar) as shown in FIG. 7 (c). obtain. Therefore, for example, the intensity A (x) of the light passing through the periphery of the round bar (shield) 4 having the radius r is the light intensity A (x) L of the left diffraction pattern and the light intensity of the right diffraction pattern. A (x) R synthesized with A (x) R A (x) = A (x) L + A (x) R
= 1.37 sech {-1.98 (2 / λz) ^1/2 (x + r) -2.39}
+1.37 sech {1.98 (2 / λz) ^1/2 (x−r) −2.39}
Can be understood as However, as the diameter of the round bar (drilling blade) becomes smaller, the overlap of the left and right diffraction patterns near the [0.25] in the light intensities A (x) L and A (x) R greatly affects the linear pattern. Since the light intensity on the light receiving surface of the image sensor greatly exceeds [0.25], the position where the light amount is [0.25] cannot be directly detected as the edge position as described above. .

Therefore, in the diameter measurement process, the light intensity A (x) L, A (x) R of the left and right diffraction patterns described above is a part that is hardly affected by the other diffraction pattern at the first rising portion. Specifically, attention is paid to a part where the light intensity (light quantity) is [0.5 to 0.9], and the light intensity (light quantity) is, for example, [0.75] as shown in FIG. , Approximate edge positions X _R and X _L are obtained (steps S11 and S12). Then, a shadow width 2a in which the amount of light is [0.75] in the diffraction pattern A (x) is obtained from these rough left and right edge positions X _R and X _L (step S13). The diameter (diameter) 2r is obtained by calculating back the radius r of the round bar (step S14).

Specifically, the edge position X _R where the light intensity is [0.75] is obtained from the right diffraction pattern A (x) _R as follows. That is, the light intensity y
y = 1.37 sech {1.98 (2 / λz) ^1/2 X−1.21}
In
Y = y / 1.37
And put
sech ⁻¹ (Y) = ± ln [{1+ (1−Y ² ) ^1/2 } / Y]
= X'-1.21
However, 0 <y ≦ 1.37, 0 <Y ≦ 1.0, X ′ = 1.98 (2 / λz) ^1/2 X
It becomes.

Therefore, the measured values (normalized data y0, y1, y2,..., Y101) in each light receiving cell of the linear image sensor 2x (2y) composed of 102 cells are the [n−1] th cell and the nth cell. If there is a position where the light intensity is [0.75] between the cells and the light intensity in the [n-1] th and nth cells is yn-1, yn,
Yn-1 = yn-1 / 1.37, Yn = yn / 1.37
As in the case shown in FIG. 6 described above, the position where the light intensity is [0.75] is X′n−1 = 1.21−ln [{1+ (1−Yn−1 ² ) ^1/2. } / Yn-1]
X′n = 1.21−ln [{1+ (1−Yn ² ) ^1/2 } / Yn]
Can be mapped respectively. As a result, the edge position X _R is changed from these mapping positions to X _R [μm] = w {n−X′n / (X′n−X′n−1)}.
Can be obtained easily and accurately by interpolation processing. Where w is the width of the cell. As described above, X′n and X′n−1 are not original cell positions but are values multiplied by 1.98 (2 / λz) ^1/2. Since this term is substantially eliminated by obtaining, the interpolation process can be performed accurately even if the distance z is unknown.

Similarly, an edge position X _{L at} which the light amount is [0.75] is obtained from the left diffraction pattern A (x) L. Then, according to the edge positions X _R and X _L obtained from these diffraction patterns A (x) R and A (x) L, the light shielding width (shadow width) 2a at the position where the light quantity is [0.75] is set to 2a. = X _R -X _L
Asking.

Next, the above light quantity [0.75] and half the value a of the shadow width 2a are added to the expression representing the diffraction pattern obtained by combining the light intensities A (x) R and A (x) L of the right and left diffraction patterns. Substituting and calculating the radius r of the round bar 4 by back calculation. For the reverse calculation of r, for example, numerical calculation may be performed using the Newton method. Specifically, f (r) = 1.37 sech {-1.98 (2 / λz) ^1/2 (a + r) -2.39}
+1.37 sech {1.98 (2 / λz) ^1/2 (ar) -2.39}
-0.75
Then, the derivative is f ′ (r) = − 2.71 (2 / λz) ^1/2
Xsech {-1.98 (2 / λz) ^1/2 (a + r) -2.39}
Xtanh {-1.98 (2 / λz) ^1/2 (a + r) -2.39}
-2.71 (2 / λz) ^1/2
× sech {1.98 (2 / λz) ^1/2 (ar) -2.39}
X tanh {1.98 (2 / λz) ^1/2 (ar) -2.39}
It becomes.

Therefore, the initial value r0 of r is set to r0 = {2a−0.845 (λz) ^1/2 } / 2
age,
r _n = r _n−1 −f (r _n−1 ) / f ′ (r _n−1 )
n = 1,2,3, ...
If it is repeatedly calculated until the absolute value of [r _n −r _n−1 ] is, for example, 0.01 or less, the converged r can be obtained as the radius of the round bar 4. Therefore, the diameter of the round bar 4 can be obtained as twice the radius r, specifically as 2r.

If the diameter of the round bar is somewhat thick and the interference between the right and left diffraction patterns A (x) R and A (x) L is negligible, the light intensity of one diffraction pattern By simply using A (x) R and A (x) L, for example, 0.75 = 1.37 sech {1.98 (2 / λz) ^1/2 (ar) -2.39}
Just solve
2r = 2a-0.845 (λz) ^1/2
The radius r can be obtained as follows. That is, the diameter (drill diameter) 2r of the round bar is simply obtained by simply subtracting [0.845 (λz) ^1/2 ], which is the prescribed value of the optical system, from the shadow width 2a at the light amount [0.75]. be able to.

  By the way, as described above, the optical sensor device according to the present invention is configured to project the monochromatic light respectively emitted from the point light sources 1x and 1y directly toward the linear image sensors 2x and 2y. In this case, the light projected from the point light sources 1x and 1y toward the line sensors 2x and 2y has a predetermined divergence angle and is not the above-mentioned monochromatic parallel light. Fresnel diffraction does not occur at the edge of the body. However, the divergence angle of the laser light emitted from the LD as the point light sources 1x and 1y is generally relatively narrow at about 17 ° at most, so that this is regarded as parallel light, that is, Fresnel diffraction occurs. Therefore, even if edge detection is performed as described above, almost no measurement error occurs. Therefore, when the above-described laser diode (LD) that emits monochromatic light having an divergence angle of about 17 ° is used as the point light sources 1x and 1y, by using the above-described edge detection method as it is, It becomes possible to detect the width 2a of A with high accuracy.

  Further, in this optical sensor device, light having a predetermined divergence angle is emitted as it is from the point light sources 1x and 1y, so that the optical system is configured from a point light source 1x (1y) as schematically shown in FIG. A shadow of the emitted monochromatic light (laser light) is magnified, and the magnified optical system is projected onto the light receiving surface of the linear image sensor 2x (2y). The shadow A by the round bar 4 is a ratio between the distance SD between the point light source 1x (1y) and the linear image sensor 2x (2y) and the distance WD between the round bar 4 and the linear image sensor 2x (2y). Is projected onto the linear image sensor 2x (2y) at an enlargement ratio determined by

  When monochromatic parallel light is used as measurement light, the shadow A of the round bar 4 corresponds to the diameter of the round bar 4 and, for example, the linear image sensor 2x (2y) as it is as shown in FIG. In the case where monochromatic light having a predetermined divergence angle from the point light source 1x (1y) is used, the shadow A of the round bar 4 is enlarged and linear as shown in FIG. 10 (b). It is projected onto the image sensor 2x (2y). Accordingly, if the width 2a of the shadow A of the enlarged round bar is obtained as described above, the round bar is obtained from the measured value (shadow width) 2a on the condition that the enlargement ratio of the above-mentioned enlargement optical system is clear. 4 can be obtained.

That is, when the diameter of the round bar 4 is 2r, the shadow width 2a of the round bar 4 is 2a = 2r · SD / (SD−WD) according to the above-mentioned magnifying optical system.
It is detected by expanding. Therefore, if the shadow width 2a of the round bar 4 is measured as described above, the diameter 2r of the round bar 4 is calculated back from the enlargement factor 2r = 2a (SD-WD) / SD.
Can be obtained as At the same time, since the measurement error can be reduced by [(SD−WD) / SD], the measurement accuracy can be increased.

Strictly speaking, since the shield is a round bar, the edge position with respect to the monochromatic light is a position on a tangent line passing through the point light source 1x (1y) as shown in FIG. Therefore, the width 2a of the shadow A obtained on the linear image sensor 2x (2y) is slightly narrower than the diameter of the actual round bar 4. However, if the distance SD between the point light source 1x (1y) and the linear image sensor 2x (2y) and the distance WD between the round bar 4 and the linear image sensor 2x (2y) are respectively clear, the geometric relationship can be obtained. 2r = 2a (SD-WD) / {(2a) ² + SD ² } ^1/2
As a result, the true diameter 2r of the round bar 4 can be easily calculated.

  Incidentally, when the round bar 4 is a drill blade to be mounted (chucked) on a drill, the mounting position is normally defined in advance, so that, for example, the round bar (drill blade) 4 in the optical system described above, for example. Is defined as an intermediate point (1/2 point) of the opposing distance SD between the point light sources 1x, 1y and the linear image sensors 2x, 2y, the magnification rate is doubled from the above geometric relationship ( As a result, the diameter of the round bar (drill blade) 4 can be easily obtained. However, the round bar (drill blade) 4 is not necessarily positioned accurately at the intermediate point of the optical system. If the round bar (drill blade) 4 is deviated from a predetermined position (for example, an intermediate point), the distance WD between the round bar 4 and the linear image sensor 2x (2y) naturally changes. So the magnification will be different.

  Specifically, for example, as shown in FIG. 11, when the round bar 4 is displaced in the optical axis direction, the enlargement ratio changes according to the distance WD. Therefore, the linear image sensor 2x (2y ) Also changes the width 2a of the shadow A formed on the light receiving surface. That is, as the distance WD between the round bar 4 and the linear image sensor 2x (2y) becomes shorter, the enlargement ratio becomes smaller, and accordingly the width 2a of the shadow A becomes smaller.

As shown in FIG. 12, when the round bar 4 is displaced in the direction perpendicular to the optical axis, not only the position of the shadow A formed on the light receiving surface of the linear image sensor 2x (2y) but also its width. 2a also changes. For example, if the round bar 4 is shifted from the optical axis by an angle φ, the center position of the shadow A on the light receiving surface of the linear image sensor 2x (2y) is displaced by [SD · tan φ], and the size 2a of the shadow A 2a = SD · [tan (θ / 2 + φ) + tan (θ / 2−φ)]
It becomes. However, θ is an angle formed by the optical axis passing through the center of the round bar 4 and the tangent line of the round bar 4 passing through the point light source 1x (1y).

  Incidentally, if the round bar 4 is present on the optical axis (φ = 0), the shadow 2 has a minimum width 2a of [2 · SDtan (θ / 2)]. The error due to the deviation in the direction perpendicular to the optical axis is larger as the diameter of the round bar 4 is smaller, and is larger as the distance WD is larger. However, the error is theoretically smaller than the above-described deviation in the optical axis direction, and can be almost ignored in practice. Therefore, as described above, the linear image sensors 2x and 2y of the first and second optical systems, which are arranged to face each other with a predetermined distance SD from the point light sources 1x and 1y and arranged in directions orthogonal to each other, If the width 2a of the shadow A generated by the round bar 4 positioned in the optical path and the center position thereof are respectively obtained, the round position can be determined from the geometric positional relationship between the point light sources 1x and 1y and the center position of the shadow A. The two-dimensional position coordinates (x, y) of the rod 4 can be obtained.

  The distance WD between the round bar 4 and the linear image sensors 2x and 2y can be obtained from the position coordinates (x, y) of the round bar 4 respectively. In other words, the distance WD between the round bar 4 and the linear image sensor 2x is monitored from the center position of the shadow A obtained from the output of the linear image sensor 2y, and the enlargement ratio of the shadow A detected by the linear image sensor 2x. Can be requested. Conversely, the distance WD between the round bar 4 and the linear image sensor 2y is monitored from the center position of the shadow A obtained from the output of the linear image sensor 2x, and the enlargement ratio of the shadow A detected by the linear image sensor 2y is determined. Can be sought.

As a result, from the width 2a of the shadow A obtained from the output of the linear image sensor 2x (2y) and the separation distance WD between the linear image sensor 2x (2y) and the round bar 4, as described above. 2r = 2a (SD-WD) / {(2a) ² + SD ² } ^1/2
The diameter 2r of the round bar 4 can be calculated geometrically with high accuracy. Further, if the diameters 2r of the round bars 4 obtained from the outputs of the first and second linear image sensors 2x and 2y as described above are compared with each other, the rounds by the first and second optical systems are thereby compared. It is possible to easily verify whether or not the diameter measurement of the rod 4 has been performed correctly.

  By the way, each light receiving cell 3 in the linear image sensor 2x (2y) in which a plurality of light receiving cells 3 are arranged at a pitch of 85 μm outputs a signal obtained by integrating the amount of light received in the cell over the cell width. Therefore, in the light receiving cell 3 applied to the edge of the shadow A, as described above, the output differs depending on the position of the cell width where the edge of the shadow A exists. In addition, the light partially incident on the light receiving cell 3 at the edge portion of the shadow A does not cause Fresnel diffraction in a strict sense as described above, and thus becomes an error factor when obtaining the edge position.

  Further, as described above, the light quantity distribution of the left and right edges caused by diffraction affects the change in the light quantity at the other edge, so that the above-mentioned also depends on the diameter of the round bar 4 corresponding to the width of the left and right edge positions. The magnitude of the error has changed. Further, when the round bar 1 is a drill blade and the diameter thereof is 100 μm or less, the spiral pitch of the drill blade is smaller than the pitch (85 μm) in the arrangement direction of the light receiving cells 3 as schematically shown in FIG. Thus, a halfway shadow is made on the light receiving cell 3. For this reason, it cannot be denied that an error occurs in the above-described interpolation calculation.

  Accordingly, a plurality of round bars (drilling blades) 4 having actually known diameters are used, and these round bars 4 are arranged along the arrangement direction of the light receiving cells 3 in the linear image sensor 2x (2y), that is, the point light source 1x ( 1y) is moved at a constant speed in the direction perpendicular to the optical axis of the emitted light, and changes in the left and right edge positions (distances from the left and right ends of the linear image sensor) are recorded, and from the movement distance (straight line) I examined the deviation (error component). Then, when the component of the cell width period is extracted from this error distribution using a digital filter, it has an error distribution that changes with the cell width period as shown in FIGS. 14 and 15, for example. FIG. 14 shows an error component when the magnification ratio of the optical system is doubled, and shows a change like a sine wave. FIG. 15 shows an error component when the magnification ratio of the optical system is 1.2 times, and has a periodically distorted shape.

Further, such an error distribution (shape, amplitude, and phase of the left and right edges) of the cell width period not only changes depending on the diameter of the round bar 4, but also depends on the distance (enlargement magnification) from the linear image sensor 2x (2y). Change. Then, when various round rods 4 having different diameters were used and the error magnification of the right and left edge positions was similarly determined by changing the measurement magnification in the magnifying optical system in various ways, the following became clear.
(a) Since synchrotron radiation is used as the light source, the width of the shadow A changes when the measurement magnification changes depending on the position of the round bar 4, and accordingly, the amplitude and shape of the error distribution, and also the phase between edges. Changes.
(b) The magnitude of the error amplitude with respect to the shadow width is generally parabolic, and the amplitude of the error distribution increases when the shadow width is generally small and when the shadow width is large.
(c) Further, when the shadow magnification is made constant by changing the measurement magnification according to the diameter of the round bar 4, the phase between the left and right edges does not change, but the amplitude and shape of the error distribution change.

When the diameter of the round bar 4 is as small as about 0.1 mm, the error increases due to the influence of diffraction at both side edges, and when the diameter is as large as about 2 mm, the diffraction pattern of the radiated light is increased. It seems that the error increases.
From the above facts, it has been clarified that the error depending on the cell width period cannot be corrected accurately unless the separation distance WD between the round bar 4 and the linear image sensor 2x (2y) is known. In other words, if the light shielding width and the distance WD by the round bar 4 are clear, the pattern (shape) of the error distribution generated in the cell width period can be known. ) Is determined, for example, the shape of the error distribution is approximated by a sine function or the like, and the function for correcting the error depending on the cell width period is obtained by determining the amplitude of the error distribution according to the light shielding width. be able to. Further, regarding the above-described error due to the spiral pitch unique to the drill blade, if the light shielding width and the distance WD are clear, an approximate diameter can be obtained. Therefore, for example, if a special error correction function considering the helical pitch with respect to the drill diameter is created, the error due to the helical pitch can be corrected using this correction function.

Accordingly, when the left and right edge positions are obtained from the output of the linear image sensor 2x (2y), the edge position ΔX on the X axis obtained as described above according to the error distribution pattern (correction function f) is expressed as follows: ΔX ′ = ΔX−f (Δx, WD, 2a)
As a result, the detection accuracy can be increased.

  Incidentally, with respect to the correction function f, for example, in an optical system having an optical path length SD of 48 mm, when the distance WD between the linear image sensor 2x (2y) and the round bar 4 is 24 mm and the measurement magnification is doubled, According to the light shielding width, for example, a cell width of 85 μm as shown in Table 1 may be given as a sine function (sin function), and the distance WD in the optical system is 8 mm (measurement magnification is 1.). 2 times), a cell width of 85 μm as shown in Table 2 may be given as a cosine function (cos function) with one period.

  When there is a possibility that a deviation of about ± 1 mm occurs in the distance WD, a correction function f corresponding to the deviation amount is obtained in advance for each deviation pitch of 0.1 mm, for example, and a table is formed. The function f may be selectively used. Further, instead of the correction function f, the actual measurement values in the optical system may be tabulated in advance and the edge position may be corrected using the table. If the measurement error of the edge position depending on the cell width period of the linear image sensor 2x (2y) is corrected in this way, the measurement accuracy can be further improved effectively.

  By the way, when the round bar 4 described above is a drill blade, not only the diameter and the position coordinates of the round bar 4 are obtained as described above, but also whether or not the drill blade is rotated without centering. It is also important to confirm. However, since the drill blade is engraved in a spiral shape on its peripheral surface, the above-described optical image (image pattern) projected on the light receiving surface of the linear image sensor 2x (2y) is the drill blade (round bar). As the 4 rotates, the left and right are shaken.

Incidentally, when the drill blade 4 rotates at a low speed around its axis, the vibration itself is periodically generated according to the rotation angle. For example, the amount of received light required at the time of measuring the drill diameter is 0. shadow width 2a at a position where the _{.75] (= X R -X L} ) when calculating the, the center position c of simultaneous _{c = (X R + X L} ) / 2
If this is obtained, the runout of the drill blade 4 can be monitored by monitoring the change width of the center position c. If the change width of the center position c exceeds a predetermined threshold value, it can be determined that there is a center shift.

However, when the drill blade 4 rotates at a high speed, the concave and convex patterns formed by the spirally engraved blades overlap each other at a high speed, and are thus projected onto the light receiving surface of the linear image sensor 2x (2y). It cannot be denied that the shadow A is totally blurred (so-called defocused state). Therefore, it is difficult to obtain the shadow width 2a (= X _R −X _L ) at the position where the received light amount is [0.75].

  Therefore, in order to detect the run-out of the drill blade 4 rotated at high speed, for example, a round bar (dummy) 4 is attached to a spindle (drill chuck) to which the drill blade 4 is attached instead of the drill blade. What is necessary is just to detect the runout of the round bar 4. That is, when a round bar (dummy) having the same diameter as the drill blade is used, an optical image (image pattern) obtained by combining the left and right diffraction patterns diffracted on both sides as described above can be obtained. However, if the round bar body (shielding object) 4 rotates at a high speed while causing a runout, the position of the round bar body (shielding object) 4 is shaken in the arrangement direction of the light receiving cells of the linear image sensor 2x (2y). Schematically, the diffraction pattern when shifted to the left side and the diffraction pattern when shifted to the right side are combined to form a so-called defocused optical image (image pattern). In other words, when the runout occurs, the edge itself becomes blurred, and the light amount is originally blocked by the round bar 4 and the amount of light in the shadow portion increases, and the light amount in the peripheral portion decreases. Accordingly, when the diameter of the round bar 4 is measured in this state as described above, the diameter of the round bar 4 is required to be reduced by the amount of light at the original edge boundary.

  On the other hand, when the round bar body (drilling blade) 4 is rotating at high speed around its axis (in a state where there is no runout), or when the round bar body 4 is stopped from rotating, its optical image Since the influence of the above-mentioned center blur does not occur in (image pattern), the diameter of the round bar 4 can be accurately measured by the above-described method. Further, when the total amount of light received by the linear image sensor 2x (2y) when the round bar 4 is positioned on the optical path of the parallel light and when the round bar 4 is removed, the round bar 4 blocks the light. Since the amount of light received by the linear image sensor 2x (2y) is reduced by the amount obtained, the drill diameter can be determined relatively easily from the ratio of the amounts of received light even if high measurement accuracy cannot be expected. In addition, since the amount of light irradiated per unit time and unit area is constant, the ratio of the amount of received light is constant regardless of whether or not the round bar 4 is decentered. Accordingly, the diameter of the round bar 4 obtained in the initial state with no runout or the round bar obtained in the state where the round bar 7 is rotated at a high speed as described above from the ratio of received light quantity thus obtained. If the diameter of the body 4 is compared, it is possible to detect whether or not the round bar 4 has run out.

Therefore, in the present invention, for example, the presence or absence of runout of the round bar 4 rotated at high speed or low speed is determined according to the processing procedure shown in FIGS. That is, first, as an initial adjustment operation, the total light receiving pattern Ai is obtained by the linear image sensor 2x (2y) without the round bar 4 in the optical path of the monochromatic light 4 (step S21), and the amount of light received by each light receiving cell. The total value S ₀ (= ΣAi) is obtained (step S22). Then, the normalization parameter Ni (= 1 / Ai) is obtained so that the amount of light received by each light receiving cell becomes [1] (step S23).

Thereafter, the round bar 4 is positioned at a predetermined distance WD from the linear image sensor 2x (2y), for example, a position where the enlargement ratio is doubled (WD = SD / 2), and the light receiving pattern Yi at that time is obtained. (Step S24). And the light receiving patterns Yi using normalization parameters Ni described above and each normalized as [yi = Yi × Ni] (step S25), and the total values S ₁ of the light-receiving amount when inserting a round bar 4 (= Σyi ) Is obtained (step S26). Thereafter, the radius R _{OPT of the} round bar body 4 is set to R _OPT = 8670 (S from the ratio of the light quantity S ₁ when the round bar body 4 is put into the optical path and the light quantity S ₀ when the round bar body 4 is not put. _1/4 · S ₀₎
(Step S27). The coefficient [8670] indicates the overall length of the linear image sensor 2x (2y) in which 102 light receiving cells having a pitch width w of 85 μm are arranged.

Next, paying attention to the diffraction pattern at the above-mentioned right edge of the light receiving pattern in the linear image sensor 2x (2y), two points yn-1 and yn sandwiching the position where the light receiving amount is [0.75] are searched for ( Step S28) As described above, the position Xn-1, Xn is obtained by inversely mapping yn-1, yn to the X axis using the inverse Fresnel function (step S29). Then, the right edge position X _R is converted into X _R = 85 [n−Xn / (Xn−Xn−1)] by interpolation from the inversely mapped positions Xn−1 and Xn.
(Step S30).

Next, paying attention to the diffraction pattern at the left edge of the light receiving pattern, the left edge position X _L is obtained (step S31). And calculated as a light-shielding width at a position where the amount of receiving a difference 2a of the right edge position X _R and the left edge position X _L is [0.75] (step S32), further the right edge position X _R and the left edge From position X _L c = (X _R + X _L ) / 2
As a result, the position c of the axis is obtained (step S33).

Next, the diameter (radius) R _CCD of the round bar 4 is obtained from the shadow width 2a by the above-described reverse calculation (step S33). Then, as described above, the difference between the diameter R _{OPT of the} round bar 4 obtained by the light quantity method and the diameter (diameter) 2R _CCD of the round bar 4 obtained by analyzing the output of the linear image sensor 2x (2y) is obtained. Then, it is determined whether or not the difference is smaller than a predetermined threshold value R _MAX to determine whether or not there is a center shift (step S34). Further, it is determined whether or not the displacement of the axial center position c obtained as described above exceeds a specified value to determine the presence or absence of the center deviation (step S35). Specifically, when the difference (R _OPT −2R _CCD ) exceeds a predetermined threshold value R _MAX , it is determined that there is a runout. When the difference (R _OPT −2R _CCD ) is smaller than the predetermined threshold value R _MAX , it is further determined whether or not the displacement of the axial center position c exceeds a specified value, and the displacement c of the axial center position is determined. Is determined not to be a specified value, and it is determined that there is no runout. However, if the displacement of the axial center position c exceeds a specified value, it is determined that there is a runout. By using these two methods for determining runout, whether or not the round bar 4 is rotating at low speed or high speed can easily detect the presence or absence of runout from one of the determination results. It becomes possible.

  In this way, according to the method for determining the presence or absence of runout of the round bar 4, the presence or absence of runout is effective in a state where the round bar 4 is mounted on the chuck of the drill and rotated at high speed. In addition, it can be detected reliably. In addition, as seen in the conventional general method, it is possible to detect the presence or absence of the runout without monitoring the change in the axial center position of the round bar 4. In other words, since the round bar 4 rotates at high speed, it is possible to effectively detect the presence or absence of the runout even in a situation where the change of the axial center position cannot be detected. Its practical advantages are enormous. In particular, if the amounts of runout detected from the outputs of the first and second linear image sensors 2x (2y) are compared with each other, it is easy to evaluate whether the runout is correctly detected. Is possible.

  The present invention is not limited to the embodiment described above. For example, when measuring the diameter of the round bar 4, in this example, the light shielding width 2a at the position where the light quantity is [0.75] is obtained. The width 2a may be obtained. Practically, for example, it is sufficient to obtain the light shielding width 2a at an arbitrary position in a range where the light quantity is [0.5 to 0.9].

  In addition, when the optical sensor device having the above-described configuration is assembled, a round bar (reference pin) 4 having a known diameter is previously installed at, for example, an intermediate position of each optical system, and the width of the shadow A measured at that time The span parameter may be adjusted and stored so that becomes twice the round bar (reference pin) 4. If the span parameters are stored in this way, when the width of the shadow A is actually measured using the optical sensor device, correction can be made using the span parameters, so that the point light sources 1x, 1y and The shadow width can be accurately obtained without being confused with the positional relationship with the linear image sensors 2x and 2y. Therefore, the diameter and position of the round bar 4 can be detected with high accuracy without being affected by the deviation of the optical arrangement relationship between the point light sources 1x, 1y and the linear image sensors 2x, 2y.

  If the detection position coordinates of the round bar 4 obtained as described above are output, it is possible to control the position of the round bar 4 by using, for example, a drill blade as the round bar 4. It can be effectively used for drilling control. In addition, the present invention can be variously modified and implemented without departing from the scope of the invention.

The principal part schematic block diagram of the optical sensor apparatus which concerns on one Embodiment of this invention. The figure which shows the light-receiving surface of a linear image sensor. The figure which shows the intensity | strength pattern of the light which produced the Fresnel diffraction by the edge of the shield. The figure which contrasts the theoretical value of the light intensity distribution by a Fresnel diffraction, and the approximate characteristic using a function. The figure which shows an example of the procedure of the edge detection process from a Fresnel diffraction pattern. The figure which shows the processing concept of the edge detection shown in FIG. The figure for demonstrating the diffraction pattern produced with a micro diameter drill blade, and the measurement principle of this invention. The figure which shows the process sequence of the diameter measuring method which concerns on one Embodiment of this invention. The figure which shows the relationship between the diameter of the round bar in an expansion optical system, and the shadow width on a light-receiving surface. The figure which shows the change of the shadow width on a light-receiving surface. The figure which shows the change of the shadow on the light-receiving surface accompanying the shift | offset | difference to the optical axis direction of the round bar in an expansion optical system. The figure which shows the change of the shadow on the light-receiving surface accompanying the shift | offset | difference to the orthogonal direction with the optical axis direction of the round bar in an expansion optical system. The figure which shows typically the error factor resulting from the helical pitch of a drill blade. The figure which shows the error distribution of the edge position depending on a cell width period. The figure which shows the error distribution of the edge position depending on a cell width period. The figure which shows the example of the specific process sequence of core blur detection. The figure which shows the process sequence following the process sequence shown in FIG.

Explanation of symbols

1x, 1y Point light source 2x, 2y Linear image sensor 4 Round bar 5x, 5y Measurement unit 6 Edge detection process 7 Shadow width detection process 8 Position detection process 9 Calculation unit

Claims (9)

A first linear image sensor and a second linear image sensor each having a light receiving surface in which a plurality of light receiving cells are arranged in a straight line; ,
A first point light source and a second point light source, each of which emits monochromatic light having a divergence angle that is opposed to each linear image sensor and reaches the entire light receiving surface of each linear image sensor;
By analyzing the output of each linear image sensor, the center position of the shadow generated on the light receiving surface of each linear image sensor and its shadow width by a round bar positioned between the point light source and the linear image sensor are analyzed. Each of the shadow analysis means to find,
The distance between each linear image sensor and the round bar is determined according to the center position of the shadow determined by each linear image sensor, and the distance and the distance determined by each linear image sensor. An optical sensor device comprising: a diameter calculating means for obtaining a diameter of the round bar from the shadow width.   The shadow analysis means analyzes an output of each linear image sensor using an approximate expression of Fresnel diffraction to obtain an edge position of a light receiving pattern by each linear image sensor, and calculates the center position of the shadow from these edge positions. The optical sensor device according to claim 1, wherein each of the shadow widths is obtained.   The diameter calculating means includes shadow width correcting means for correcting a shadow width detected by the linear image sensor in accordance with a center position of a shadow detected by each of the linear image sensors. 2. The optical sensor device according to 1. The optical sensor device according to claim 1,
From the width of the shadow of the round bar measured when a round bar having a known diameter is provided at a predetermined position between the point light source and the linear image sensor, a measurement span of the linear image sensor is obtained. Parameter storage means for determining parameters for correcting and storing the parameters;
Thereafter, at the time of measuring the diameter and / or position of the round bar, it is provided with measurement span correction means for correcting the shadow width respectively obtained from the output of the linear image sensor using the parameters stored in the parameter storage means. An optical sensor device. The predetermined position between the point light source and the linear image sensor is the distance between the first point light source and the first linear image sensor, and the second point light source and the second linear sensor. It is an intermediate point that divides the facing distance with the image sensor into two parts,
The parameter for correcting the measurement span of the linear image sensor is obtained as a coefficient for correcting the arrangement pitch of the light receiving cells in the linear image sensor so that the width of the shadow is twice the diameter of the round bar. 5. The optical sensor device according to 4. The optical sensor device according to claim 1,
The distance between the other linear image sensor and the round bar obtained based on the center position of the shadow obtained by one linear image sensor, and the round bar obtained by the other linear image sensor An error pattern of the edge position of the shadow depending on the arrangement pitch of the plurality of light receiving cells in the other linear image sensor is obtained according to the shadow width of the other, and the light reception obtained by analyzing the output of the linear image sensor according to the error pattern An optical sensor device comprising edge position correcting means for correcting an edge position of a pattern to obtain an edge position of the shadow. In the optical sensor device according to any one of claims 1 to 6,
The apparatus further comprises verification means for comparing the diameters of the round bars obtained from the outputs of the first and second linear image sensors with each other and verifying the suitability of correction for the outputs of the linear image sensors. An optical sensor device. In the optical sensor device according to any one of claims 1 to 6,
The round bar according to the shadow width required when the round bar positioned between the point light source and the linear image sensor is not rotating and the shadow width required when the round bar is rotating A center blur detection means for detecting the amount of core blur at the time of rotation of the body,
An optical sensor device that determines whether or not the center shake detection is appropriate by comparing the amounts of center shake calculated from the outputs of the first and second linear image sensors.   The coordinate output means which outputs the center position of the said round bar obtained by analyzing each output of the said 1st and 2nd linear image sensor, respectively is provided. Optical sensor device.

Images