JPH095046A - Three-dimensional shape measuring apparatus - Google Patents

Abstract

PURPOSE: To provide an apparatus to measure a three-dimensional shape at a high speed and high precision by image processing in which an occluding edge problem does not occur, of which transverse resolution function for an original image is not deteriorated, by which processing is carried out easily, and which has a sufficient speed for in-line use and has constant image focusing magnification against the change of focus point. CONSTITUTION: The three-dimensional shape measuring apparatus is composed of a confocal scanning image-pickup system 3 consisting of a confocal scanning optical system 1 and a photoelectric sensor 2 to carry out photoelectric transduction of the image obtained by the system 1, a focus point changing means 4 for the confocal scanning image-pickup system 3, and an image processing apparatus 6 to calculate the three-dimensional shape of an object by calculating the convergence position of each point of images at precision exceeding the gap of focus points of images based on density information of a plurality of images with different focus points obtained by the confocal scanning image- pickup system 3 and the focus point changing means 4.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a three-dimensional shape measuring device for optically measuring the three-dimensional shape of an object.

[0002]

2. Description of the Related Art Conventionally, as an optical three-dimensional shape measuring method focusing on a focus position, Japanese Patent Application Laid-Open No. 3-63507 or paper [1] "Shape from focus system" has been used.
temfor rough surfaces ", Pr
ocedings of the Image Und
erstanding, 1992, S.M. K. Naye
The one described in r is known (hereinafter, this method will be referred to as S
(Hope From Focus method, abbreviated as SFF method). In the SFF method, a texture pattern is projected on the object surface, a plurality of images with different focal positions are taken by operating the z-axis stage, and for each local region of each obtained image, the contrast information in the local region is used. This is a method for measuring the three-dimensional shape of an object by performing a process of obtaining a focus measure and obtaining a focus position (maximum position of the focus measure) by interpolation calculation for each corresponding local area between images over the entire image. , Is known as a highly accurate three-dimensional shape measuring method for a minute object.

[0003]

However, this SF
The F method has some problems. (1) When a steep solid edge as shown in FIG. 7 exists, measurement cannot be performed in the area near the edge below the solid edge. In this region, even if the focus of the image pickup system is located at this position, the blurring of the upper edge is superposed, and therefore the change in contrast cannot be confirmed.
(Hereafter, this problem is called the occluding edge problem). (2) The lateral resolution is low (the lateral resolution is the resolution in the xy directions when the height direction is the z direction). In order to obtain the contrast information, which is a focus measure, a local area having a size enough to include texture is required, so that a smoothing effect is produced as compared with the original image, and the lateral resolution is reduced. (3) The processing is complicated. The calculation of obtaining the focusing measure from the contrast information in the local area requires the differential processing and the processing of the local sum or the processing of obtaining the variance value of the histogram. In software processing, this portion takes a very long time, so the processing circuit Need to be manufactured. Further, as the size for differentiating or performing the local sum, the optimum value thereof differs depending on the object, so it is necessary to select it for each object.

In addition, there are the following two problems in practical use.
(4) The conventional method of moving the focus position by operating the z-axis stage is cumbersome to control, and it is difficult to cope with the high-speed response required for in-line measurement. (5) Since the problem of the change of the imaging magnification when the focus position is changed is not taken into consideration, the measurement accuracy is lowered.

Therefore, according to the present invention, there is no problem of an occluding edge, the horizontal resolution is not deteriorated with respect to the original image, the processing is easy, the speed can withstand in-line use, and the focus position changes. On the other hand, it is an object of the present invention to provide a three-dimensional shape measuring device in which the imaging magnification does not change.

[0006]

In order to achieve the object, according to the present invention, a confocal scanning optical system includes a confocal scanning optical system and a photoelectric sensor for photoelectrically converting a two-dimensional optical image obtained by the confocal scanning optical system. An image pickup system, a focus position changing means for setting a focus position of the confocal scanning optical system, and a plurality of images having different focus positions obtained by the confocal scanning image pickup system and the focus position changing means From the density value of each image point that changes according to the change of the, the focus position that gives the maximum density value with accuracy that exceeds the focus position interval of the captured image The image processing apparatus is configured to perform processing for estimating and estimating the estimated focus position as the height of the point.

Further, the focal position changing means uses a plurality of parallel plate type transparent bodies having different thicknesses or a plurality of parallel plate type transparent bodies having different refractive indexes from each other and the confocal point of the confocal scanning optical system. The confocal scanning optical system is configured to be sequentially inserted into the optical path between the image planes, and the confocal scanning optical system is configured to be telecentric on the object side.

[0008]

By configuring the device in this way, the following actions are obtained. (1) Since the confocal scanning imaging system forms an image by scanning spot light, the density value at a certain point is not affected by the blurring at another point, and the problem of occluding edge does not occur. (2) The density value of each point of the image formed by the confocal scanning imaging system becomes higher as it approaches the in-focus position. Since this is the property itself of the focusing measure of the SFF method, the density value may be used as it is for the calculation of the focusing measure, and the calculation of the focusing measure is not necessary. Therefore, it is not necessary to set the local area, and the height of a certain point is determined from the density value of only that point of each image, so that the horizontal resolution does not decrease. (3) (2)
Similarly, since it is not necessary to obtain the focusing measure, it is not necessary to obtain the differentiation, the local sum, and the variance value of the histogram, and the processing circuit is simplified. Further, there is no need to be bothered to select the processing size. (4) Since the movement of the focal position is achieved by the change in the optical path length due to the insertion of the parallel plate type transparent body, it is not necessary to move the z-axis stage, and high-speed image input is possible. (5) Since the confocal scanning optical system is telecentric, the change in imaging magnification does not occur.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Before describing the embodiments of the present invention, a general description will be given of a confocal scanning optical system. First, the confocal optical system will be described. FIG. 2 shows the basic configuration of the confocal optical system. The illumination light emitted through the pinhole 201 is condensed by the objective lens 202 and converges on the focal plane 203. When the object surface 204 is located at this position, the reflected light of the object converges on the pinhole 201 in the completely opposite process to the illumination light, and most of the reflected light incident on the objective lens 202 passes through the pinhole 201. However, when the object surface 204 moves away from the focal plane 203, the convergent point of the reflected light also moves away from the pinhole 201, and the amount of light passing through the pinhole 201 decreases. At this time, if the light intensity | V (z) | ² passing through the pinhole 201 is k, the wave number of the illumination light is k, the NA of the objective lens 202 is sin θ, and the distance between the converging point of the reflected light and the pinhole 201 is z | V (z) | = | sin
kz (1-cosθ) | / | kz (1-cosθ) | is known (Paper [2] “De
pth response of confocal o
optical microscopes ", OPTIC
S LETTERS, Vol. 11, No. 12, 19
1986, T. R. (See Corle et al.), When the wavelength of the illumination light is 550 nm and the NA is 0.1, the light intensity changes as shown in FIG. Thus, in the confocal optical system, the pinhole 2
The position of the object surface 204 can be determined by the amount of reflected light passing through 01, and since the illumination light is applied to the object as a spot, there is no influence of light from other points on the object. That is, blurring at other points does not occur.

Next, the confocal scanning optical system will be described. The confocal scanning optical system is an optical system that obtains a two-dimensional optical image by scanning the confocal optical system two-dimensionally, and an example thereof is shown in FIG. FIG. 3 also shows a photoelectric sensor 2 that photoelectrically converts a two-dimensional optical image obtained by the confocal scanning optical system together with the confocal scanning optical system, and constitutes a confocal scanning imaging system as a whole. . The light emitted from the light source 101 is condensed by the condenser lens 102 and applied to the Nipkow disk 103. Ni
The pkow disk 103 is a disk in which a large number of pinholes are spirally provided, and the large number of pinholes on the disk each form a confocal optical system in combination with the objective lens 105. A sufficient distance is provided between the pinholes so that lights do not interfere with each other. When the Nipkow disk 103 is rotated at a high speed by the motor 104, the confocal optical system is two-dimensionally scanned. The reflected light from the object 5 that has passed through the pinhole on the Nipkow disk 103 is reflected by the half mirror 1.
It is reflected at 07 and guided to the area type photoelectric sensor 2 by the lens 108. This part is the photoelectric sensor 2 Nipko
The optical arrangement is such that a pinhole on the w disk 103 forms an image, and an image corresponding to the intensity of light passing through the pinhole is formed on the photoelectric sensor 2.

Next, an embodiment of the present invention will be described with reference to FIGS. 1 to 6. FIG. 1 shows the configuration of one embodiment of the present invention. The two-dimensional optical image obtained by the confocal scanning optical system 1 becomes an electric video signal by the photoelectric sensor 2 (this video signal generating unit composed of the confocal scanning optical system 1 and the photoelectric sensor 2 is confocal scanned. (Imaging system 3). The confocal scanning imaging system 3 has the configuration shown in FIG. 3 described in detail above. FIG.
The confocal scanning image pickup system 3 is a Nipkow disk10.
3 is rotated and scanned, but in the present invention, a more general structure in which a laser beam is scanned using a rotating mirror, a galvanometer mirror or an AO element may be used. In this case, the object 5 may be moved by the xy table. Any configuration may be used as long as an image is obtained by two-dimensionally scanning an optical system that is optically equivalent to the confocal optical system shown in FIG.

Ni sandwiching the objective lens 105 in FIG.
The optical path between the pkow disk 103 and the object 5 is provided with a focal position changing means 4 as shown in FIG. 1 so that the focal position can be changed. A plurality of images with different focal positions (for example, three images of the object 5 obtained at the focal positions of z1, z2, and z3) obtained by the focal position changing means 4 and the confocal scanning imaging system 3 are image processing devices. 6 and the three-dimensional shape of the object 5 is calculated by the image processing device 6 from these images.

Next, a method for obtaining the three-dimensional shape of the object 5 from a plurality of images having different focal positions will be described. The densities of the points representing the same position on the object 5 in the respective images are taken as the focus position coordinates (z
If they are arranged on the (coordinates), this will be a discretization of the continuous waveform shown in FIG. An example of discretization is shown by a dotted line in FIG. Since the relationship between the z coordinate and the light intensity (density in the image) can be accurately expressed by the model | V (z) | ² described above, the peak position of the center lobe (hereinafter referred to as the peak position), that is, the in-focus position is calculated from the discrete information. Can be accurately estimated. For example, if a Gaussian function, which is a function very similar to the shape of a center lobe, is used, the peak position can be obtained analytically, and the maximum value v1 of discrete values and one value v2 before or after the maximum value of two values in total. Therefore, the peak position p is expressed as follows: p = p1 + (1 + a2 (v2-v1)) / 2. Here, p1 is the z coordinate of v1, and a is a parameter indicating the spread of lobe, and the wavelength of illumination and the N.V. A. Is a constant determined by. The peak position calculation method is not limited to this, and three or more points may be used, or a function having another similar shape such as a quadratic function may be used. Of course, the model formula | V (z) | ² can also be used directly.
Besides, calculation using moment is possible. In order to speed up these arithmetic processes, the arithmetic result may be stored in the LUT in advance and the result may be referred to. Since the obtained in-focus position indicates the height of that point of the object, the three-dimensional shape of the object 5 can be obtained by executing the above calculation for all the points in the image.

This calculation process will be compared with the solid shape calculation process of the SFF method. (1) First, the focus measure calculation (contrast evaluation), which was indispensable in the SFF method, is not required at all. (2) In the interpolation calculation, the peak position because the shape parameter of the model formula changes depending on the surface state of the object in the SFF method. At least three points are required for the calculation, and since the relationship between the focus measure and the model formula is not clear, it is necessary to select the focus measure calculation method by experiment. According to the paper [2], the model formula of the optical system is accurately known, and the shape parameter does not change even if the surface state and the inclination of the object 5 are different. Therefore, in the present invention, the peak position is more accurate and more efficient. (At least 2
It is what is required.

Next, the structure of the focus position changing means 4 is shown in FIG. A plurality of parallel plate-shaped transparent bodies 40 having different thicknesses
2 are arranged on the circumference of the rotary plate 401 (transparent body 4
As the material of 02, optical glass, optical resin, liquid, liquid crystal, etc. can be used). The rotating plate 401 is arranged so that this circumference exactly intersects the optical axis of the confocal scanning optical system 1 perpendicularly (so that the transparent body 402 is inserted in the optical path of the confocal scanning optical system 1). When the rotating plate 401 is rotated at high speed by the motor 403, the thickness of the transparent body 402 inserted in the optical path of the confocal scanning optical system 1 changes at high speed. When the thickness of the transparent body 402 changes, the focus position changes with the change of the optical path length, so that the focus position changes at high speed. Rotating plate 40
By synchronizing the rotation speed of No. 1 and the imaging speed of the confocal scanning imaging system 3 by using the sensor 404, it is possible to obtain images with different focus positions at the imaging speed (real time) of the confocal scanning imaging system 3. Becomes Although a plurality of parallel plate type transparent bodies 402 having different thicknesses are used as the means for changing the optical path length, the same can be applied to a plurality of parallel plate type transparent bodies 402 having different refractive indexes. In addition to this embodiment, for example, the objective lens 1 of the confocal scanning optical system 1
If 05 has a simple structure with a special optical element such as an aspherical lens, it is possible to change the focus position at high speed by directly moving the objective lens 105 by vibrating means such as a tuning fork.

In FIG. 3, the objective lens 105 is shown as a general image forming lens. However, in the general image forming lens, the magnifications of images having different focal positions are different from each other, and
When using a general imaging lens, it is necessary to accurately perform the height measurement calculation in consideration of this change in magnification. Therefore, by making the objective lens 105 a telecentric lens on the object side (a structure in which an aperture 601 is opened at the rear focal point of the objective lens 602 as shown in FIG. 6), the chief ray becomes parallel to the optical axis, Even if the focal position is different, the magnification does not change in each image.

With the above structure, there is no problem of occluding edges, the horizontal resolution does not deteriorate with respect to the original image, the processing is easy, and the speed is high enough to be used inline. In addition, it is possible to measure the three-dimensional shape in which the imaging magnification does not change with respect to the change in the focus position. This device enables in-line inspection such as LSI mounting inspection, for example, inspection of TAB inner leads for peeling or forming abnormality, bonding wire loop height inspection, and bump shape inspection.

[Brief description of drawings]

FIG. 1 is a diagram showing an embodiment of the present invention.

FIG. 2 is a diagram showing a basic configuration of a confocal optical system.

FIG. 3 is a diagram showing an example of a confocal scanning imaging system of the present invention.

FIG. 4 is a diagram showing a relationship between a focal position on the image side and light intensity.

FIG. 5 is a diagram showing an example of a focus position changing means of the present invention.

FIG. 6 is a diagram showing a telecentric confocal optical system.

FIG. 7 is a diagram for explaining an occluding edge.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Confocal scanning optical system 2 Photoelectric sensor 3 Confocal scanning imaging system 4 Focus position changing means 5 Object 6 Image processing device

Claims (3)

[Claims] 1. A device for optically measuring a three-dimensional shape of an object, which comprises a confocal scanning optical system and a confocal scanning device which photoelectrically converts a two-dimensional optical image obtained by the confocal scanning optical system. An image pickup system, a focus position changing unit that changes the focus position of the confocal scanning optical system, and a plurality of images having different focus positions obtained by the confocal scanning image pickup system and the focus position changing unit are captured. The focus position that gives the maximum density value is calculated from the density value of each image point that changes according to the change of A three-dimensional shape measuring apparatus, comprising: an image processing apparatus that performs processing of estimating and estimating the estimated focus position as the height of the point. 2. The focus position changing means comprises a plurality of parallel plate-shaped transparent bodies having different thicknesses or a plurality of parallel plate-shaped transparent bodies having different refractive indexes, which are used for the object and the confocal scanning optical system. The three-dimensional shape measuring apparatus according to claim 1, wherein the three-dimensional shape measuring apparatus is one that is sequentially inserted in an optical path between focal image formation planes. 3. The three-dimensional shape measuring apparatus according to claim 1, wherein the confocal scanning optical system is telecentric on the object side.