JP3937024B2 - Detection of misalignment, pattern rotation, distortion, and misalignment using moiré fringes - Google Patents

Description

The present invention relates to an accurate pixel correspondence between a plurality of projection display devices and an imaging device, an optical device that requires accurate alignment, and an inspection relating to flatness and perpendicularity of a planar object.

There is a need for high-speed and high-precision non-contact shape measurement techniques in various fields such as industrial fields. If shape data can be obtained in real time, it is possible to inspect a processed part efficiently. As the technique, shape deformation measurement by a phase shift method for shifting the phase of the projection grating is often used. However, since it is necessary to take a plurality of images in the phase shift method, there is a problem that an error increases for an object whose shape changes with time. There are proposals that can project phase analysis from RGB information by projecting a color grid, but there are some methods that can measure well depending on the color of the object and others that are not so general.
The present inventors have so far developed a DMD reflection type CCD camera (DMD camera) (Digital Micro-mirror Device camera). If a DMD camera is used, all necessary images can be captured in one frame of the CCD camera, and errors due to movement of moving objects can be greatly reduced.

In a DMD camera, it is important to accurately perform a 1: 1 pixel correspondence between DMD and CCD. However, since both the optical elements of DMD and CCD are very small, it is difficult to image a conventional calibration pattern and adjust it visually, and it is also difficult to adjust quantitatively.
Here, an adjustment method using a moire pattern caused by a difference in spatial frequency when DMD and CCD correspond to pixels is proposed. Further, by shifting the phase of the moire pattern, a phase distribution can be obtained, adjustment on the order of pixels or less is possible, and automatic adjustment of the optical system is also possible. As a result, each mirror of the DMD and each pixel of the CCD can be associated more accurately.

  The present inventors have so far researched and developed a DMD camera by combining a DMD and a CCD camera, and have proposed application to optical measurements such as phase analysis and shape measurement using the DMD camera.

For example, as disclosed in Patent Document 1, a projection image alignment device for a projection display device has been proposed, and the light output positions of two projection display devices are adjusted so that moire is not displayed. By inputting an adjustment signal that causes moiré in each of the plurality of upper parts, it is possible to know whether or not there is a shift for each part and easily adjust a complicated shift. However, since the phase of the moire fringe pattern is not obtained and the accuracy that can be adjusted is pixel order, and the adjustment signal is a sine wave, it is difficult to apply to other than a projection display device (see Patent Document 1) ).
See Japanese Patent Application Laid-Open No. 2004-69769.

Further, as disclosed in Patent Document 2, there is a projection display system and a projection position adjustment method, in which a position detection unit such as an image sensor or a position detection element is used to display the position of an inspection pattern projected together with a video signal on a screen. The amount of displacement is calculated by image processing or the like, and the projection position of the optical system is automatically adjusted so that the amount of displacement is eliminated. However, since the amount of displacement is detected by image processing, in principle, the accuracy higher than the pixel order cannot be obtained (see Patent Document 2).
JP-A-8-168039

Conventionally, when aligning a projected image, a pattern such as a cross hatch is displayed as a display image, and the position of the projected image is adjusted so that there is no visual shift. Accordingly, it has been impossible to accurately perform pixel correspondence of a very small size element.
The features of the present invention are as follows.
(1) Adjustment of pixel alignment can be easily performed without requiring a special device.
(2) A phase distribution can be obtained from a single image.
(3) By performing the phase analysis, adjustment can be made with an accuracy of pixel order or less.

The present invention is characterized in that the pixels of the two optical elements are made to correspond to 1: 1 precisely and moiré fringes generated by the difference in the spatial frequency of the pixel interval between the two optical elements are used.
According to the present invention, a black and white two-pixel pitch grid pattern is displayed in the vertical and horizontal directions in the visual adjustment method. When thinning out every two pixels, a moire pattern caused by a spatial shift can be seen. It is characterized in that the deviation of the optical system can be known from the moire pattern and the pixel can be easily handled.
According to the present invention, a black and white four-pixel pitch grid pattern is displayed in the vertical and horizontal directions in the computer adjustment method. A moire pattern can be seen from an image thinned out every four pixels from the first. Similarly, a total of four images obtained by thinning out every second pixel from the second, third, and fourth pixels are obtained. It is possible to obtain four images whose phases are phase shifted by 1/4 period. By using these images, it is possible to analyze the phase of the moire pattern indicating the shift amount. It is characterized in that pixel alignment is performed from a single image with an accuracy less than the cell order of the optical element.
The present invention is characterized in that the flatness and perpendicularity of a planar object can be known by pasting a calibration pattern in advance on the surface of the object having a planar shape to be observed.
The present invention is characterized in that if a lattice pattern is drawn on a part of a part that has been adjusted in advance, it is possible to perform accurate alignment so that the same position is always obtained by adjusting so as not to generate a moire pattern.

The present invention relates to a method for detecting an optical system displacement, pattern rotation, and position displacement, and a pattern in which a test pattern having a two-pixel pitch is displayed or pasted is photographed by an image sensor in which pixels are arranged two-dimensionally. The image processing for calculating the sum of the absolute values of the difference between the four pixels adjacent to the adjacent upper, lower, left and right pixels is performed on the generated image, and a moire pattern having a predetermined number of moire fringes is obtained.
The present invention is an optical system misalignment, pattern rotation, and position misalignment detection method, in which a pattern on which a test pattern having a pitch of four pixels is displayed or pasted is photographed by an image sensor in which pixels are arranged two-dimensionally. The image processing for thinning out every four pixels while changing the start point for thinning out the calculated image, calculating the phase distribution of the moire pattern from the obtained four-step phase-shifted image, and the phase distribution of the moire pattern It is detected from.
In the optical system deviation, pattern rotation, and position deviation detection method, the present invention further takes a pre-deformation image as the first state, obtains the phase distribution of the moire pattern generated at that time, and obtains the second state. And detecting a phase difference between the first state and the second state by obtaining a phase distribution of the moire pattern generated at that time, and obtaining a phase difference between the first state and the second state. And
According to the present invention, in an optical system misalignment, pattern rotation, and position misalignment detection method, a test pattern having one cycle in the horizontal direction or the vertical direction may be two pixels or four for an image sensor in which pixels are arranged two-dimensionally. It is characterized by being a test chart before and after pixels.
According to the present invention, in an optical system misalignment, pattern rotation, and position misalignment detection method, a test pattern having one cycle in the horizontal direction or the vertical direction may be two pixels or four for an image sensor in which pixels are arranged two-dimensionally. A test chart having pixel front and rear is used, and a pattern in which the horizontal and vertical directions are combined is used as the test chart.
The present invention relates to a detection method of optical system deviation, pattern rotation, and position deviation, which is divided into four regions on a diagonal line, a vertical test pattern in the left and right regions, and a horizontal test pattern in the upper and lower regions. As a test chart combining the above, a moiré pattern in the horizontal direction and the vertical direction is obtained at the same time.
The present invention relates to a detection method of optical system deviation, pattern rotation, and position deviation, which is divided into four regions on a diagonal line, a vertical test pattern in the left and right regions, and a horizontal test pattern in the upper and lower regions. As a test chart combining the above, a phase distribution of moire patterns in the horizontal direction and the vertical direction is obtained at the same time.
According to the present invention, a pattern in which a pixel-related adjustment method pitch is displayed or pasted with a test pattern including 2 pixels and 4 pixels at the same time is photographed by an image sensor in which pixels are two-dimensionally arranged. Calculate the sum of the absolute values of the differences between the four pixels on the top, bottom, left and right adjacent to each other and perform image processing to thin out every four pixels while changing the start point of the thinned image. Calculate the phase distribution of the moire pattern from the phase-shifted image, perform rough alignment using the fine pitch part from the phase distribution of the moire pattern, and calculate the phase using the coarse pitch part It is characterized by precise adjustment.

According to the present invention, the following effects can be obtained.
(1) Pixel alignment and alignment can be easily performed.
(2) Adjustment can be made with an accuracy of pixel order or less.
(3) Since only one image needs to be taken, moiré fringes due to displacement can be observed in real time.
(4) The phase distribution of moire fringes can be obtained from one image. Does not require special equipment.
(5) Flatness and perpendicularity can be easily determined using a moire pattern.
(6) By performing automatic adjustment so that moire fringes are not always generated, precise alignment can be performed.

First, pixel correspondence of a DMD reflection type CCD camera (hereinafter referred to as “DMD camera”) will be described.
FIG. 1 is an explanatory diagram of the basic operation of the DMD camera. An image of the object to be imaged is formed on the surface of the DMD element by the lens 1. At this time, the DMD and the lens surface of the lens 1 are arranged parallel to each other. At this time, the lens 1 has a tilt angle of 24 degrees. In this way, the light emitted from one point on the surface of the target object, which is the positional relationship of image formation, reaches each DMD mirror on the surface of the DMD. Next, the lens 2 is arranged so that the DMD surface and the CCD surface are in a positional relationship of image formation. With this arrangement, the light that reaches the DMD pixel receiving the on signal is reflected in the CCD direction, and is imaged by the lens 2 on the corresponding pixel of the CCD. Since the switching time of the direction of the DMD pixel is about 1/1000 compared with the CCD shooting time, each CCD pixel has a shutter function that can be controlled independently at high speed.

  An overall view of the DMD camera device used by the present inventors is shown in FIG. FIG. 3 illustrates the relationship between the on / off pattern of the DMD and the photographed image when the subject to be photographed is the pattern as shown in FIG. First, when all the mirrors of the DMD are on, the image to be imaged is directly captured as a digital image by the CCD as shown in FIG. 3 (b). Next, when some DMD pixels are off as shown in FIG. 3 (c), only the on pixels are photographed as shown in FIG. 3 (d). Also, as shown in Fig. 3 (e), by sending a signal of a pattern in which the exposure time of pixels in a part of the DMD is adjusted to the DMD, only the part of the tree has uniform brightness as shown in Fig. 3 (f). You can also shoot to become.

By using the DMD camera in this manner, a high-speed shutter that operates independently at a sub-millisecond order speed is attached to each CCD pixel, and can be used for various purposes. However, to achieve this, the 1: 1 correspondence of DMD and CCD pixels must be adjusted as accurately as possible. Since both the DMD and CCD optical elements are very small, visual adjustment is difficult, and there is a problem that quantitative adjustment is difficult.
In the conventional apparatus, the positions of the four corners and the center are visually adjusted using a characteristic pattern such as a cross hatch. However, in the case of an extremely small element, the accuracy of visual adjustment is limited, and it has not been possible to easily adjust with an accuracy of a pixel order or less.

Next, a visual adjustment method using a moire pattern will be described.
In the DMD camera, in the pixel alignment apparatus as shown in FIG. 4, it is important that each mirror of the DMD and each pixel of the CCD correspond exactly 1: 1. Here, a method that can be accurately adjusted by using a moire fringe generated by a difference in spatial frequency between DMD and CCD pixels for adjusting the optical system of the DMD camera will be further described. With this method, the DMD and CCD pixels can be 1: 1.

  FIG. 5 shows an example of a calibration pattern. First, the DMD displays a rectangular wave lattice pattern with two pixels of one pitch as shown in FIG. Install the CCD and lens 2 at a position that satisfies the relationship of image formation. At this time, if the CCD surface is not parallel to the DMD surface or if the size of the CCD and the size of the DMD image formed on the CCD do not match, that is, each mirror of the DMD and each pixel of the CCD can support 1: 1. In a state in which it is not, moire fringes as shown in FIG. 6 appear. Since the moire fringes change regularly depending on the position and tilt of the CCD camera, the optical system shift corresponding to the pixels can be known from the moire fringe pattern. As the deviation increases, the number of fringes increases. Each pixel of mirror and CCD corresponds to 1: 1. In order to make it easier to observe the moiré pattern, applying an edge detection filter to the moiré pattern photographed with a CCD camera makes it possible to see a clearer moiré pattern and doubles the number of moiré fringes. . In FIG. 7, when the image forming magnification is changed by shifting the position in the z direction when the pattern of FIG. 5 (a) is displayed, the angles α, β, and θ are inclined about the x, y, and z axes, respectively. The simulation result after performing the image processing of the edge detection filter for the moire pattern that appears in the case is shown.

In FIG. 8, when the image forming magnification is changed by shifting the position in the z direction when the pattern of FIG. 5 (b) is displayed, the angles α, β, and θ are inclined about the x, y, and z axes, respectively. The simulation result after performing the image processing of the edge detection filter for the moire pattern that appears in the case is shown.
For this reason, when the CCD image is not tilted but the magnification of the image formed on the CCD is different, linear moire fringes appear in the horizontal direction as shown in FIG. 7 (a), but the CCD camera tilts. In such a case, the moire fringes are curved or inclined as shown in FIGS. 7 (b), (c), and (d).
Therefore, first, the adjustment is performed by rotating the CCD camera so that the moire fringes are linear. Next, the front and rear positions of the lens and the CCD are adjusted with respect to the DMD so as not to generate moire fringes. Finally, by matching the corners of the DMD pattern display range and the CCD shooting range, each mirror of the DMD and each pixel of the CCD can be made to correspond.

Next, a computer adjustment method using a moire pattern will be described.
Although the visual adjustment method using the moire pattern is a simple visual adjustment method, it is difficult to discriminate moiré fringes when the displacement is about one or two pixels. In addition, it is difficult to distinguish from the brightness distribution caused by the influence of light and darkness due to ambient light. However, evaluation with one pixel or less is not possible. On the other hand, an adjustment method using a computer capable of quantitatively evaluating the deviation of the optical system below the pixel order will be described below.
In the adjustment method by the computer, first, as shown in FIG. 5A, a black and white four-pixel pitch lattice pattern is displayed in the vertical and horizontal directions. As shown in FIG. 9, a moire pattern can be seen from an image thinned out every four pixels from the first. Similarly, a total of four images are obtained by thinning out every second pixel from the second, third, and fourth. It is possible to obtain four images whose phases are shifted by 1/4 period. The phase of the moire pattern indicating the amount of deviation can be analyzed using a phase shift method for obtaining the phase from the four most commonly used luminance values. In this adjustment method, a shift of 4 pixels corresponds to a phase value of 2π. Therefore, if there is a phase difference of π / 2 at both ends of the analysis range, a shift of 1 pixel occurs. Pixel alignment can be performed from one image with an accuracy less than the cell order of the optical element.
According to the present invention, it is possible to obtain a two-dimensional phase distribution of a moiré pattern caused by a shift in the spatial position of the CCD and DMD, and more accurately perform CCD and DMD pixel correspondence. In addition, the optical system can be automatically adjusted by obtaining the shift amount such as the magnification and the tilt angle using the two-dimensional phase distribution data.

Next, the operation of the pixel alignment method according to the present invention will be described in detail.
First, the correspondence between DMD and CCD pixels will be described. An adjustment device in the case of corresponding to a DMD camera pixel is shown below. In the visual adjustment method using the moire pattern, as shown in FIG. 5 (c), a rectangular wave lattice having 2 pixels each in the horizontal direction in the upper and lower areas and in the vertical direction in the left and right areas is displayed on the DMD. An example of a moire pattern actually obtained is shown in FIG. By using this calibration pattern, the moire patterns in the x and y directions can be observed simultaneously. Fig. 10 (a) shows the moiré pattern when the size is shifted by about 12 pixels due to deviation from the correct imaging position, and Fig. 10 (b) shows the moiré pattern when shifted by about 10 pixels from the correct imaging position. It is a pattern. It can be seen that the smaller the deviation due to the imaging magnification, the smaller the number of moire fringes. FIGS. 10C to 10H are moire patterns generated by the rotation angle β around the z-axis as shown in FIG. Moire fringes that are greatly rotated can be observed due to a slight shift in the rotation angle θ. FIGS. 10 (i) and 10 (j) are moire patterns generated by the rotation angle β around the y-axis. When the rotation angle β is small, the square moiré fringes are not formed, but the shape is shifted diagonally. FIG. 10 (k) shows a moire pattern when β = 0, which is a square moire pattern. FIG. 10 (l) shows a moire pattern generated by the rotation angle α around the x axis. It can be seen that when the rotation angle α is slightly, the pitches of the moire fringes in the x and y directions are greatly different. FIG. 10 (m) shows a moire pattern when α = 0, and moire fringes in the x and y directions have the same interval. FIG. 10 (n) shows a moire pattern in which the pixel correspondence is slightly shifted. The moire pattern when it finally matches is shown in FIG. 10 (o).

Next, a computer adjustment method using a moire pattern will be described.
FIGS. 11 to 13 show the obtained results of the phase distribution of the moire pattern generated when corresponding to the pixels of the DMD camera. FIGS. 11A to 11D show four shifted images obtained by thinning out every four pixels by the method described in the phase shift moire method. The result of the phase analysis is shown in FIG.
FIG. 12 shows the phase distribution obtained from the result of the phase analysis, in the case of 14 pixel deviation, where (a) shows the x direction and (b) shows the y direction.
At this time, there is a shift of about 14 pixels. In this case, it can be sufficiently confirmed that the shift is visually. FIG. 13 shows a phase distribution when there is a shift of about 3 pixels, for example, 3.26 pixels, between DMD and CCD. (A) shows the x direction, and (b) shows the y direction.
It is somewhat difficult to confirm the shift visually. FIG. 14 shows the phase distribution when the pixel correspondence of DMD and CCD match. (A) shows the x direction, and (b) shows the y direction.
With respect to the analysis range of 640 pixels, the phase difference between both ends is 0.08 (corresponding to a 0.05 pixel shift), which can be said to be almost completely coincident. FIG. 15 shows a result of pixel correspondence of the DMD camera, and shows a pattern obtained in a state where the pixel correspondence is not achieved and in a state where it is possible.
(A) shows an ideal pattern, (b) shows a case where the pixel correspondence is shifted, and (c) shows a case where the pixel correspondence matches. According to the present invention, the state corresponding to the pixel is known.

Although the embodiments of the present invention have been described above, some specific embodiments will be described in detail.
Image processing that calculates the sum of the absolute value of the difference between the four pixels on the top, bottom, left, and right adjacent to the captured image by photographing a pattern with a test pattern with a pitch of 2 pixels displayed or pasted with a CCD camera, Compared with the conventional moire pattern obtained by thinning out every two pixels, the number of moire fringes is doubled, and a moire pattern with high sensitivity and good contrast is obtained.
That is, as represented by the following expression (1) in FIG. 16, when attention is paid to a certain pixel of the photographed image, this is a process of calculating the sum of the absolute values of the differences between the upper, lower, left and right four pixels.

By doing so, as shown in FIG. 17, the result of comparing the moire pattern obtained by the conventional thinning-out process for every two pixels and the moire pattern obtained by the method of the present configuration described above can be understood. It can be seen that the number of moire stripes is doubled. Also good contrast. That is, FIG. 17 shows a moiré pattern obtained by image processing, (a) an image taken by a CCD camera, (b) a moiré pattern obtained by a conventional thinning process for every two pixels, and (c) obtained by this configuration. The resulting moire pattern is shown.

Next, in a further embodiment, a pattern in which a test pattern with a pitch of 4 pixels is displayed or pasted is photographed by a CCD camera, and image processing is performed to thin out every 4 pixels while changing the start point for thinning out the photographed image. Then, the phase distribution of the moire pattern is calculated from the obtained 4-step phase-shifted image. From the phase distribution of the moire pattern, optical system deviation, pattern rotation and distortion, and positional deviation are detected.
FIG. 18 shows the phase distribution of the moire pattern obtained by this method. In this method, four phase-shifted moire pattern images are obtained from the photographed image by thinning-out processing, and the phase distribution is obtained from these four images. Since the phase information is analyzed, it is possible to detect optical system deviation, pattern rotation and distortion, and positional deviation with higher accuracy than the above method.
That is, FIG. 18 shows the phase distribution of the moire pattern obtained by this method. (A) is an image taken with a CCD camera, (b) from the first pixel, (c) from the second pixel, (d) From the third pixel, (e) From the fourth picture, the moire pattern obtained by thinning out every four pixels, (f) is the moire pattern obtained from the four phase-shifted images of (b) to (e). The phase distribution of a pattern is shown.

Next, in another embodiment, a pattern in which a test pattern with a 4-pixel pitch is displayed or pasted is photographed with a CCD camera, and image processing is performed to thin out every four pixels while changing the start point for thinning out the photographed image. The phase distribution of the moire pattern is calculated from the obtained 4-step phase-shifted image. In the method of detecting the optical system deviation, pattern rotation and distortion, and positional deviation from the phase distribution of the moire pattern, an image before deformation is taken as the first state, and the phase distribution of the moire pattern generated at that time is obtained. Taking the deformed image as the second state, obtaining the phase distribution of the moire pattern generated at that time, and obtaining the phase difference between the phase distribution of the first state and the second state, It detects optical system displacement, pattern rotation and distortion, and positional displacement.
FIG. 19 shows a phase difference distribution image. Using this method, phase distribution images of the first state and the second state are obtained, respectively, and the phase obtained by subtracting the phase distribution of the first state (phase difference) from the phase distribution of the second state From the distribution image, optical system deviation, pattern rotation and distortion, and positional deviation are detected.
That is, FIG. 19 shows a phase difference distribution image according to this method, (a) is a phase distribution image in the first state, (b) is a phase distribution image in the second state, and (c) is the first state distribution image. Phase distribution image obtained by subtracting the phase distribution of the first state from the phase distribution of the two states (phase difference) (this phase difference distribution image shows how much deviation is present in the two states) ).

  In another embodiment, in the above-described method, as shown in FIGS. 20A and 20B, the test pattern of one cycle in the horizontal direction or the vertical direction is about 2 pixels or about 4 pixels for the CCD camera. A test chart having such a test chart or a combination of the horizontal direction and the vertical direction can be used as the test chart.

  In still another embodiment, in the above-described method, as shown in FIG. 21, the test area is divided into four areas on the diagonal line, and a vertical test pattern is formed in the left and right areas, and a horizontal test pattern is formed in the upper and lower areas. As a combined test chart, it is possible to obtain a test chart that is easier to visually adjust than the above method by simultaneously obtaining a moire pattern or its phase distribution in the horizontal and vertical directions.

Further, the test chart is configured as described in FIGS. 22 (a) and 22 (b), that is, a pattern on which a test pattern including one pixel having two and four pixels is displayed or pasted is photographed with a CCD camera. Calculates the sum of the absolute values of the difference between the upper, lower, left, and right four pixels adjacent to the photographed image, and performs image processing that thins out every four pixels while changing the start point of the photographed image. Then, the phase distribution of the moire pattern is calculated from the obtained four-step phase-shifted image, and the coarse pitch portion is roughly aligned using the fine pitch portion from the phase distribution of the moire pattern. It is possible to use an adjustment method corresponding to a pixel so that the phase is calculated by using it and a precise adjustment is performed.
In this way, rough positioning can be performed using the fine pitch portion, and the phase can be calculated using the coarse pitch portion to facilitate more precise adjustment.

  The detection and adjustment methods described above can be applied when performing alignment and pixel correspondence of a plurality of devices in a projection device apparatus such as a DMD or a liquid crystal projector.

An example in which the present invention is applied to a flatness and perpendicularity inspection apparatus will be described.
FIG. 23 shows a schematic diagram of a flatness inspection apparatus for a flat object. A calibration pattern for observing a moire pattern is attached in advance to the four corners (and / or the center) of a planar object to be inspected. After adjusting the calibration pattern pitch interval and the CCD camera pixel interval to be the same, shoot with the CCD camera. By the above-described method, moire patterns at four corners can be obtained. At this time, if the moire pattern is similar, it can be determined that there is flatness and perpendicularity. If the moire patterns are different, it can be seen that the planar object is distorted or tilted.
In this way, the verticality can be inspected by applying the calibration pattern.

Next, an example in which the method according to the present invention is applied to an alignment apparatus in a semiconductor exposure apparatus (stepper) or the like will be described.
FIG. 24 shows an alignment apparatus for a semiconductor exposure apparatus (stepper). In a semiconductor exposure apparatus (stepper), it is necessary to precisely match a new pattern with the position of a pattern already drawn on the wafer. The present invention can be applied to alignment when repeatedly performing exposure on the same wafer. As shown in FIG. 24, a grid pattern of 0.1 to 1 mm square is prepared in advance as an alignment mark on the end portion of the wafer to be exposed on the rotary stage, so that it is always fixed at the same position. It can be seen from the state of the moire pattern photographed from the camera, and the position can be precisely aligned by automatically adjusting so that moire fringes do not always occur. It can also be solved by changing the pitch of the lattice pattern according to the accuracy to be adjusted.

  In the above embodiment, an example using a CCD camera has been described, but it is a matter of course that a CMOS camera can be used as an image pickup device in which pixels are two-dimensionally arranged.

  As an optical device that requires accurate pixel correspondence between a plurality of projection display devices and photographing devices, an inspection device for flatness and perpendicularity of a planar object, and a positioning device for a semiconductor exposure device (stepper) Applicable.

Configuration of Embodiment of DMD Reflective CCD Camera (DMD Camera) Device DMD reflection type CCD camera (DMD camera) device Images taken with a DMD camera System explanatory diagram of pixel alignment device Calibration pattern Moire pattern taken with a CCD camera Various moiré patterns in the x direction (a) When the magnification does not match, (b) When tilted to the axis, (c) When tilted to the y-axis, (d) When tilted to the z-axis Various moiré patterns in the y direction (a) When the magnifications do not match, (b) When tilted to the x axis, (c) When tilted to the y axis, (d) When tilted to the z axis Computerized adjustment method using moire fringes Moire pattern obtained in the adjustment process experiment Four phase-shifted images obtained by the phase shift moire method Phase distribution obtained from the result of phase analysis (in the case of 14 pixel deviation) (a) x direction, (b) y direction Phase distribution obtained from the result of phase analysis (in the case of 3 pixel deviation) (a) x direction, (b) y direction Phase distribution obtained from the result of phase analysis (when matched) (a) x direction, (b) y direction DMD camera pixel support results Relationship between a pixel P and surrounding pixels Moire pattern obtained by image processing (a) Image taken with a CCD camera, (b) Moire pattern obtained by conventional thinning process for every two pixels, (c) Moire pattern obtained by this configuration Phase distribution of moire pattern obtained by the method of this configuration (a) Image taken with a CCD camera, (b) From the first pixel, (c) From the second pixel, (d) From the third pixel, (e) 4 Moire pattern obtained by thinning out every four pixels from the pixel, and phase distribution of moire pattern obtained from four phase-shifted images (f) (b) to (e) Phase difference distribution image (a) Phase distribution image of the first state, (b) Phase distribution image of the second state, (c) Subtracting the phase distribution of the first state from the phase distribution of the second state ( Phase distribution image obtained by (phase difference) (this phase difference distribution image shows how much the difference is in the two states) (A), (b) Test chart Test chart (A), (b) Test chart Flatness inspection of planar objects Positioning device in a semiconductor exposure apparatus (stepper)

Claims (8)

  A pattern in which a test pattern having a pitch of 2 pixels is displayed or pasted is photographed by an image sensor in which pixels are arranged two-dimensionally, and the absolute value of the difference between the four pixels adjacent to the captured image is measured An optical system misalignment, pattern rotation, and position misalignment detection method, characterized by performing image processing for calculating a sum and obtaining a moire pattern having a predetermined number of moire fringes.   Image processing in which a test pattern with a pitch of 4 pixels is displayed or pasted with an image sensor in which pixels are two-dimensionally arranged, and the start point for thinning out the captured image is changed and thinned out every 4 pixels And calculating the phase distribution of the moire pattern from the obtained four-step phase-shifted image and detecting from the phase distribution of the moire pattern, pattern rotation, distortion, and position Deviation detection method.   In the detection method according to claim 2, as a first state, a pre-deformation image is taken, a phase distribution of a moire pattern generated at that time is obtained, and as a second state, a post-deformation image is taken, Determining the phase distribution of the moire pattern generated at that time, and detecting the phase difference between the phase distribution of the first state and the second state, and detecting the optical system shift and pattern rotation Distortion and misalignment detection method.   4. The detection method according to claim 1, wherein a test pattern of one cycle in the horizontal direction or the vertical direction is 2 pixels or around 4 pixels for an image pickup device in which pixels are arranged two-dimensionally. An optical system misalignment, pattern rotation and distortion, and misregistration detection method, characterized by using a test chart as described above.   4. The detection method according to claim 1, wherein a test pattern of one cycle in the horizontal direction or the vertical direction is 2 pixels or around 4 pixels for an image pickup device in which pixels are two-dimensionally arranged. 5. A method for detecting optical system misalignment, pattern rotation and distortion, and positional misalignment, characterized in that the test chart is a pattern having a test chart and a combination of the horizontal and vertical directions.   4. The detection method according to claim 1, wherein the region is divided into four regions on a diagonal line, and a vertical test pattern is combined in the left and right regions, and a horizontal test pattern is combined in the upper and lower regions. A method for detecting misalignment of an optical system, pattern rotation or distortion, and misalignment, characterized by obtaining a moire pattern in the horizontal direction and the vertical direction simultaneously as a test chart.   4. The detection method according to claim 1, wherein the region is divided into four regions on a diagonal line, and a vertical test pattern is combined in the left and right regions, and a horizontal test pattern is combined in the upper and lower regions. A method for detecting optical system misalignment, pattern rotation and distortion, and positional misalignment, characterized by simultaneously obtaining a phase distribution of moire patterns in the horizontal and vertical directions as a test chart.   A pattern in which a test pattern including 2 pixels and 4 pixels at the same time is displayed or pasted is photographed by an image sensor in which pixels are two-dimensionally arranged. The sum of the absolute values of the differences is calculated, and image processing is performed for every four pixels while changing the starting point for thinning the captured image, and the moire pattern of the four-step phase-shifted image obtained is obtained. The phase distribution is calculated, and from the phase distribution of the moire pattern, rough positioning is performed using the fine pitch portion, and the phase is calculated using the coarse pitch portion, and precise adjustment is performed. An adjustment method corresponding to a characteristic pixel.