JP3062664B2 - Optical pattern recognition device having coordinate conversion function - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the field of optical information processing and optical measurement, in which two-dimensional images obtained from an imaging device such as a CCD camera are subjected to optical coordinate conversion and optical correlation processing using coherent light. To automatically perform pattern recognition and measurement by performing the following.

[0002]

2. Description of the Related Art As a correlator for optically performing pattern recognition, two systems of a VanderLugt type and a joint conversion type are generally well known. In either method,
Since it is based on the optical Fourier transform using a lens, the translation of the correlated object to be recognized can be recognized without any problem (shift invariance). On the other hand, it is known that the recognition ability is not invariable with respect to the rotation and the size of the correlated object, and the degree of the degree of the cognition decreases.

[0003] Conventionally, for example, like characters and parts,
When optically measuring and recognizing objects having different shapes, sizes, orientations, and positions, first, the correlated object is converted into a two-dimensional image (correlated image), and second, the correlated image is converted. The general procedure is to convert to a desired coordinate system that is invariant to changes required for recognition such as rotation and size change, and to perform measurement and recognition on the coordinate transformed image thirdly. is there.

[0004] The types of coordinate conversion include polar coordinate conversion when recognizing and measuring the rotation angle of an object having a different orientation, and recognition and measurement of the rotation angle and magnification for an object having a different orientation and size. For example, when performing (x, y) coordinates to (lnr, θ) coordinates,
Various types of coordinate transformations are used depending on the purpose. Here, r and θ are the length and argument of the moving radius in polar coordinates.

First, an outline of an optical coordinate conversion method is shown in FIG. In this method, a liquid crystal television 303 displaying an image to be converted and a coordinate conversion optical filter 304 are superposed and arranged on the front focal plane of the coordinate conversion lens 307, and parallel coherent light is emitted from behind the liquid crystal television 303. Thus, a desired coordinate conversion image can be obtained on the liquid crystal light valve 308 disposed on the rear focal plane of the coordinate conversion lens 307. Here, the coordinate conversion optical filter 304
Are produced by computer-generated holograms. In addition, as long as the liquid crystal television 303 and the coordinate conversion optical filter 304 are arranged close to each other, it goes without saying that either one may be in front.

As the light modulation material of the liquid crystal light valve 308, a material using TN liquid crystal is often used. Also, instead of the liquid crystal light valve 308, B
Sometimes used SO crystal (Bi ₁₂ SiO ₂₀₎ optical write type spatial light modulator using a crystal such as. These spatial light modulators are irradiated with a coordinate conversion image and displayed and stored, and then irradiated with coherent light to read out the coordinate conversion image and use it for processing such as pattern recognition. In addition, instead of an optically-written spatial light modulator, an image-acquisition device such as a CCD camera is used to receive a coordinate-converted image, and the obtained image signal is converted into an electrically-written spatial-light modulator such as a liquid crystal television. There is also a way to input.

Next, as a method of pattern recognition using coordinate transformation, a method of applying an ordinary VanderLugt type correlator to an image subjected to coordinate transformation as preprocessing is already known. FIG. 3 shows a VanderLugt having a conventional coordinate conversion function.
1 shows a configuration diagram of a type of correlator. FIG. This method will be described with reference to the drawings. First, as a first step, a matched filter of a reference image is prepared. In this step, the reference image is displayed on the liquid crystal television 303. The coherent light emitted from the laser 301 is expanded into a parallel light having a predetermined beam diameter by a beam expander 302, and then split into two by a beam splitter 305. One of the luminous fluxes passes through the liquid crystal television 303 displaying the reference image and the coordinate conversion optical filter 304, and passes through the coordinate conversion lens 3.
At 07, the writing surface of the liquid crystal light valve 308 is irradiated with the Fourier transform. In this way, the coordinate transformation intensity distribution image of the reference image is displayed on the liquid crystal light valve 308. The other light beam split by the beam splitter 305 is reflected by the mirror 306, the beam splitter 309, and the polarization beam splitter 310, and then irradiates the readout surface of the liquid crystal light valve 308 to convert the coordinate conversion intensity distribution image into a coherent image. .

[0008] The coherent image passes through the polarizing beam splitter 310 and passes through the Fourier transform lens 31.
Then, the photographic dry plate 312 is subjected to Fourier transform at 1, and is irradiated as signal light at the time of producing a hologram. At this time, the luminous flux transmitted through the beam splitter 309 at the same time is reflected by the mirror 314 via the shutter 313 in the open state, and irradiates the photographic dry plate 312 as reference light when producing a hologram. At this time, the signal light and the reference light are set at a predetermined angle to the photographic dry plate 31.
2 to form a hologram on the photographic plate 312. Of course, the liquid crystal light valve 308 is disposed on the front focal plane of the Fourier transform lens 311, and the photographic plate 312 is disposed on the rear focal plane of the Fourier transform lens 311. The photographic dry plate 312 on which the hologram is formed in this manner is removed and developed, and after development, it is placed again at the original position. The photographic dry plate 312 on which the hologram is recorded and developed is called a matched filter.

Next, in a second step, a correlation process is actually performed. The description of the same steps as those in the first step is omitted or simplified. In this step, the correlated image is displayed on the liquid crystal television 303. In the same manner as in the first step, the coordinate-transformed intensity distribution image of the correlated image is
08, the image is read out, and Fourier-transformed by a Fourier-transform lens 311 to irradiate a photographic dry plate 312 as a matched filter. At this time, unlike the first step, the shutter 313 is in the closed state, and the reference light does not irradiate the photographic dry plate 312. The light transmitted through the photographic dry plate 312, which is a matched filter, is converted into a Fourier transform lens 315.
Is subjected to Fourier transform again, whereby a correlation peak representing the correlation coefficient between the reference image and the correlated image can be obtained on the light receiving element 316 disposed on the conversion plane. Here, the photographic plate 312 is disposed on the front focal plane of the Fourier transform lens 315, and the light receiving element 316 is disposed on the rear focal plane.

[0010]

However, according to the above method, there are three problems caused by using a photographic dry plate as a medium for recording a hologram as a matched filter. First, after forming a hologram on a photographic plate, it is necessary to remove and develop it.
It took a lot of trouble and time. Second, when the photographic plate is developed and then set back to the original position, it is very troublesome and difficult to adjust the photographic plate completely to the original position. In particular, it is necessary not only to align the optical axis with the center of the hologram, but also to set the hologram so that there is no rotation or tilt. Third, if the reference image is changed, a new matched filter must be recreated or replaced with a previously created filter, and it is not possible to change the reference image in real time or at high speed. could not. For the above reasons, it has not been possible to configure a pattern recognition device having a coordinate conversion function capable of operating in real time at a practical level.

Another problem caused by the use of a VanderLugt type correlator is that a hologram formed on a photographic plate usually has a large angle of several tens of degrees between signal light and reference light, and therefore has several hundreds of holograms. The interval between interference fringes is very narrow, about [lp / mm]. Therefore, in the stage of manufacturing the matched filter, it is easily affected by vibration, wind, and the like, and if the interference fringes are swayed by them, an accurate hologram cannot be recorded, and the pattern recognition ability is reduced. .

Further, another problem caused by performing the optical coordinate transformation is that since the shift invariance is eliminated, accurate pattern recognition cannot be performed when the correlated image is translated. For example, consider the case where the direction and size of the correlated image are exactly the same as the reference image. In a normal VanderLugt-type pattern recognition device that does not perform optical coordinate conversion, as described above, there is shift invariance due to the Fourier transform feature, so the position of the correlated image on the input surface is Can be accurately recognized even if the position is different (parallel movement) from the position of the reference image. However, in the pattern recognition device that performs the optical coordinate transformation as shown in FIG. 3, the position of the correlated image displayed on the liquid crystal television 303 is different from the position of the reference image at the time of producing the matched filter and is translated. Then, since the relative positional relationship between the reference image or the correlated image and the coordinate conversion optical filter 304 changes, the shape of the image of the coordinate conversion image of the reference image differs from that of the correlated image. Therefore, if the correlation processing is performed using such a coordinate conversion image, there is a problem that accurate recognition cannot be performed naturally.

[0013]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention provides a two-dimensional image obtained from a CCD camera or the like by performing an optical correlation process using coherent light to obtain a required pattern. In an optical pattern recognition apparatus for automatically recognizing and measuring a coordinate conversion image obtained by converting at least one reference image including a required target and at least one newly input correlated image into a desired coordinate system, An optical coordinate conversion unit to be obtained, and a joint conversion correlator unit that obtains a correlation coefficient between the coordinate conversion image of the reference image and the coordinate conversion image of the correlated image obtained in the optical coordinate conversion unit, The optical coordinate conversion unit includes: at least one coherent light source; a unit configured to obtain the two-dimensional reference image and the correlated image; The reference image and the correlated image, which are converted so that the reference image and the correlated image are shift invariant, are converted into the shift invariant, and the at least one shift invariant image spatial light modulator is converted into the shift invariant image. Means for displaying;
The at least one coordinate conversion optical filter and the at least one lens are arranged so as to be superimposed on the shift invariant image spatial light modulator for converting into a desired coordinate system. The coordinate transformation images of the reference image and the correlated image obtained in the optical coordinate transformation unit are transformed into coordinate transformation intensity distribution images, and the coordinate transformation intensity distribution images are displayed on a coordinate transformation image spatial light modulator. Means, means for converting each coordinate transformed intensity distribution image displayed on the coordinate transformed image spatial light modulator into a coherent image, and Fourier transforming the coherent image using a lens, and the coordinate transformed intensity distribution image Means for obtaining a joint Fourier transform image, and transforming the joint Fourier transform image into a Fourier transform intensity distribution image, wherein the Fourier transform intensity distribution Means for displaying an image on a Fourier transform image spatial light modulator, means for reading the Fourier transform intensity distribution image displayed on the Fourier transform image spatial light modulator using coherent light, and means for reading the Fourier transform image. Correlation output image obtained by Fourier transforming the converted intensity distribution image again using a lens, means for converting to a correlation signal using an imaging device or a light receiving element, and signal processing the correlation signal and the reference image 2 of the correlated image
And a means for calculating a dimensional correlation coefficient.

[0014]

In the optical pattern recognition apparatus having the coordinate conversion function configured as described above, the correlated image and the reference image are converted by the optical coordinate conversion section with respect to a desired change such as rotation or change in size. Since the image is optically converted to the invariable coordinate system using the coordinate conversion optical filter, the pattern can be invariably recognized with respect to such a change in the correlated image.

Also, the present invention is based on a joint transform type correlator unlike the VanderLugt type correlator.
To change the coordinate conversion optical filter or change the reference image to change the type of coordinate conversion, simply change the reference image or the coordinate conversion optical filter, and record, develop and replace the matched filter again to adjust the optical axis. Since there is no need for troublesome operations, it is very easy and fast. Further, the optical axis adjustment and the like in the optical system of the entire pattern recognition apparatus are very easy as compared with the VanderLugt type correlator. Therefore, the influence of the vibration of the apparatus, the fluctuation of the air, and the like is reduced.

Further, the optical coordinate transformation unit performs pre-processing for transforming the reference image or the correlated image so as to be shift invariant electronically or optically, and then applies the shift invariant intensity distribution image to the shift invariant intensity distribution image. As a result, desired optical coordinate transformation that is invariant to rotation and size change is performed, so that pattern recognition can be performed not only for rotation and size change but also for shift. As a method of optically converting so as to be shift invariant, Fourier transform is performed on a correlated image or a reference image using coherent light. As a result, even if the displayed correlated image is translated on the spatial light modulator, the phase distribution of the Fourier transformed image only changes due to the Fourier transform feature, and the intensity distribution does not change. Therefore, by performing a desired optical coordinate transformation on the intensity distribution image (shift-invariant intensity distribution image) of the Fourier transform image and then performing a correlation process, the result of the correlation process is shift-invariant.

On the other hand, as a method of electronically making a shift invariant, a computer calculates a center of gravity of an image of a reference image or a correlated image, and the center of gravity of those images is always calculated on a shift invariant image spatial light modulator. Image processing by a computer is performed so as to be at a fixed position. Thus, the relative positional relationship between the reference image and the corresponding coordinate transforming optical filter is always the same as the relative positional relationship between the correlated image and the corresponding coordinate transforming optical filter. And
By performing the desired optical coordinate transformation on such an image-processed image and then performing the correlation processing, the result of the correlation processing becomes shift-invariant in the same manner as described above.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a configuration diagram of an embodiment of an optical pattern recognition device having a coordinate conversion function according to the present invention. As a configuration of the optical coordinate conversion unit, at least one coherent light source includes a shift invariant laser 61, a shift invariant beam expander 62, a coordinate conversion laser 5, a coordinate conversion beam expander 6, and a coordinate conversion polarization beam. The means for obtaining the two-dimensional reference image and the correlated image, which includes the splitter 12, includes the input image capturing device 2, the image processing device 42, and the image memory device 43, and optically or electronically stores the reference image. The correlated image is converted so as to be shift invariant, and the reference image and the correlated image converted so as to be shift invariant are at least one.
The means for displaying the shift invariant image spatial light modulators are the input image electric writing spatial light modulator 3, the shift invariant lens array 64, and the shift invariant optical writing spatial light modulator 65. The at least one coordinate conversion optical filter, which is disposed on the shift invariant image spatial light modulator so as to convert it into a desired coordinate system, is a coordinate conversion optical filter array 44.
And at least one lens is a coordinate conversion lens array 47.

The configuration of the joint transform correlator section is as follows.
The coordinate transformation images of the reference image and the correlated image obtained in the optical coordinate transformation unit are transformed into coordinate transformation intensity distribution images, and the coordinate transformation intensity distribution images are displayed on a coordinate transformation image spatial light modulator. The means for performing comprises a mask 48 and an optical writing type spatial light modulator 13 for coordinate conversion, and means for converting each coordinate conversion intensity distribution image displayed on the coordinate conversion image spatial light modulator into a coherent image. A Fourier transform laser 101, a Fourier transform beam expander 102, and a Fourier transform polarizing beam splitter 103. The coherent image is Fourier transformed using a lens to obtain a joint Fourier transform image of the coordinate transform intensity distribution image. The means comprises a first Fourier transform lens 104, and converts the joint Fourier transform image to a Fourier transform intensity. The means for converting to an image and displaying the Fourier transform intensity distribution image on the Fourier transform image spatial light modulator comprises a Fourier transform optical writing type spatial light modulator 105, and the Fourier transform image spatial light modulator. The means for reading out the Fourier transform intensity distribution image displayed in FIG. 3 using coherent light includes a correlation laser 201, a correlation beam expander 202, and a correlation polarization beam splitter 2.
Means for converting a correlation output image obtained by subjecting the read-out Fourier transform intensity distribution image to Fourier transform again using a lens into a correlation signal using an imaging device or a light receiving element. A unit that includes a conversion lens 204 and a light receiving element 205 and that performs signal processing on the correlation signal to obtain a two-dimensional correlation coefficient between the reference image and the correlated image, respectively, includes an identification circuit 206.

First, a reference object, which is a database serving as a reference for recognition, is arranged at the position of the correlated object 1, and this is photographed by the input image pickup device 2 to form a two-dimensional reference image. The image is stored in the image memory device 43 via the device 42. Next, similarly, the correlated object 1 to be recognized is photographed by the input image capturing device 2 to form a two-dimensional correlated image.
The reference image stored in the memory 3 is synthesized by the image processing device 42 and displayed on the electric writing spatial light modulator 3 for input image. Here, for example, when pattern recognition is performed on the correlated object 1 whose position changes with time, the correlated object 1
The position of the correlated image displayed on the input image electrical writing type spatial light modulator 3 also changes with the movement of. Even in such a case, to prevent the reference image and the correlated image from overlapping,
As shown in FIG. 4, the reference image and the correlated image are synthesized separately. Needless to say, if the correlated object 1 moves, the positional relationship such as the distance between the reference image and the correlated image changes.

The coherent light emitted from the shift invariant laser 61 is transmitted to the shift invariant beam expander 6.
2, the beam is expanded into a parallel beam having a predetermined beam diameter,
By irradiating the input image electric writing spatial light modulator 3 on which the correlated image and the reference image are displayed, these images are converted into coherent images. These coherent images are Fourier-transformed by each shift-invariant lens of the shift-invariant lens array 64, so that a Fourier transform image as a shift-invariant image is obtained on the conversion surface. Here, each shift invariant lens on the shift invariant lens array 64 corresponds to a correlated image or a reference image displayed on the input image electric writing spatial light modulator 3, respectively. They are separated by a predetermined distance L.

On the conversion surface from which a Fourier-transformed image is obtained, a shift-invariant optical writing type spatial light modulator 65 and a coordinate conversion optical filter array 44 are arranged immediately before it. With this arrangement, an image obtained by superimposing a coordinate conversion optical filter on each Fourier transform image of the reference image or the correlated image irradiates the writing surface of the shift-invariant optical writing spatial light modulator 65 as writing light. I do. By the above, the image obtained by superimposing the coordinate transformation optical filter on the Fourier transform image of each of the correlated image and the reference image is converted into the intensity distribution, and the shift invariant optical writing spatial light modulator is used as the shift invariant intensity distribution image. It is displayed on 65 at a distance L. There are many types of shift-invariant optical writing type spatial light modulators 65. In this embodiment, a configuration using a reflection type liquid crystal light valve whose light modulating material is a ferroelectric liquid crystal will be described. Further, the input image electric writing spatial light modulator 3 is arranged on the front focal plane of the shift invariant lens array 64, and the shift invariant optical writing spatial light modulator 65 and the coordinate conversion optics are arranged on the rear focal plane. The filter arrays 44 are arranged in an overlapping manner.

The coordinate conversion optical filter array 44 has a computer generated hologram (C).
GH), two coordinate transforming optical filters manufactured by GH) are separated by a predetermined distance L, and are arranged corresponding to the positions of the shift invariant lenses. In coordinate conversion, if the relative positional relationship between the coordinate conversion optical filter and the image to be converted is different, different coordinate conversion images will be obtained even if the same image is subjected to coordinate conversion. Therefore, the distance between the two coordinate transformation optical filters is defined as the distance L between each Fourier transform image.
(Equal to the distance between the shift invariant lenses), and each coordinate transform optical filter is set so that the relative positional relationship between each Fourier transform image of the correlated image and the reference image and each coordinate transform optical filter becomes the same. Deploy. Once arranged in this way, even if the position of the correlated image is translated on the input image electric writing spatial light modulator 3 as shown in FIG.
The phase of the Fourier transform image changes, but the intensity distribution does not change at all. Accordingly, the relative positional relationship between the Fourier transform image of the correlated image and the corresponding coordinate transform optical filter does not change, and the coordinate transform image remains unchanged.

The coherent light emitted from the coordinate conversion laser 5 is expanded by a coordinate conversion beam expander 6 into a parallel light beam having a predetermined beam diameter, and then reflected and read by a coordinate conversion polarization beam splitter 12. The reading surface of the shift invariant optical writing type spatial light modulator 65 is irradiated as light. Here, the polarization direction of the readout light and the direction of alignment of liquid crystal molecules (or a direction perpendicular thereto) aligned by initialization of the shift invariant optical writing type spatial light modulator 65 are previously adjusted, and the shift is performed. Invariant optical writing spatial light modulator 6
The light is transmitted through an analyzer arranged such that the polarization axis is perpendicular (or parallel) to the polarization direction of the readout light reflected by 5, and is displayed on the shift invariant optical writing type spatial light modulator 65. Can be read out as a positive or negative image. In this embodiment, a coordinate conversion polarizing beam splitter 12 is used as an analyzer.

In this manner, each shift-invariant intensity distribution image of the correlated image and the reference image is converted into a coherent image. Then, these coherent images are Fourier-transformed by the respective coordinate conversion lenses of the coordinate conversion lens array 47, so that coordinate conversion images converted to desired coordinate systems are obtained on the conversion surfaces. Then, a hole is formed in the mask 48 disposed immediately before the coordinate conversion optical writing spatial light modulator 13 so that, for example, only each + 1st-order coordinate conversion image is transmitted. As a result, unnecessary DC bias components and higher-order coordinate conversion images are cut off, and only necessary coordinate conversion images of the shift-invariant intensity distribution images of the reference image and the correlated image are transmitted. The writing surface of the writable spatial light modulator 13 is irradiated as writing light. As described above, each coordinate conversion image of each shift-invariant intensity distribution image of the correlated image and the reference image is converted into an intensity distribution, and the coordinate conversion intensity distribution image is displayed on the coordinate-writing optical writing spatial light modulator 13. Are displayed with a distance L apart. Although many types of optical writing spatial light modulators 13 for coordinate conversion are conceivable, similar to the shift invariant optical writing spatial light modulator 65, in this embodiment, the light modulating material is a ferroelectric liquid crystal. A configuration using a reflective liquid crystal light valve will be described.

Here, a shift invariant optical writing type spatial light modulator 65 is disposed on the front focal plane of the coordinate conversion lens array 47, and the coordinate converting optical writing spatial light modulator 13 is disposed on the rear focal plane. I do. Further, each coordinate conversion lens of the coordinate conversion lens array 47 has a 2
The positions are separated by a predetermined distance L corresponding to the positions of the two shifted invariant intensity distribution images to be converted.

The coherent light emitted from the Fourier transform laser 101 is expanded into a parallel beam having a predetermined beam diameter by a Fourier transform beam expander 102, and then reflected and read by a Fourier transform polarizing beam splitter 103. Optical writing type spatial light modulator 1 for coordinate conversion as light
The reading surface of No. 3 is irradiated. Here, similarly to the case of the shift invariant optical writing type spatial light modulator 65, each coordinate conversion intensity distribution image displayed on the coordinate conversion optical writing type spatial light modulator 13 is read. Can be. However, the polarization beam splitter 103 for coordinate conversion is used as an analyzer.

As described above, each coordinate transformation intensity distribution image displayed on the coordinate transformation optical writing spatial light modulator 13 is converted into a coherent image and Fourier transformed by the first Fourier transformation lens 104. Thereby, a joint Fourier transform image of each coordinate transformation intensity distribution image of the reference image and the correlated image is formed on the conversion plane. Therefore, by arranging the writing surface of the optical writing type spatial light modulator for Fourier transform 105 on the conversion surface, the intensity distribution of the joint Fourier transform image of each coordinate conversion intensity distribution image of the reference image and the correlated image is obtained. , Are displayed on the optical writing type spatial light modulator for Fourier transform 105 as Fourier transform intensity distribution images. Here, as in the case of the shift-invariant optical writing spatial light modulator 65 and the coordinate conversion optical writing spatial light modulator 13, the Fourier transform optical writing spatial light modulator 105 is a ferroelectric material. A reflective liquid crystal light valve using liquid crystal is used. Also,
An optical writing spatial light modulator for coordinate conversion 13 is arranged on the front focal plane of the first Fourier transform lens 104, and an optical writing spatial light modulator 105 for Fourier transformation is arranged on the rear focal plane.

The coherent light emitted from the correlation laser 201 is expanded into a parallel light beam having a predetermined beam diameter by a correlation beam expander 202, and then reflected by a correlation polarizing beam splitter 203 to be Fourier-read as read light. The reading surface of the conversion optical writing type spatial light modulator 105 is irradiated. Here, too, the shift invariant optical writing type spatial light modulator 65 is used.
Similarly to the case of the coordinate-writing optical writing spatial light modulator 13, the Fourier transform intensity distribution image displayed on the Fourier transformation optical writing spatial light modulator 105 is converted into a coherent image. However, a correlation polarizing beam splitter 203 is used as an analyzer.

The read coherent image is Fourier-transformed by the second Fourier transform lens 204 to form a correlation output image including a correlation peak on the light receiving element 205 arranged on the conversion plane. Light receiving element 2
Numeral 05 receives only the correlation peak in the correlation output image and converts it into a correlation signal. The correlation signal is input to the identification circuit 206, the correlation peak intensity is measured, and the correlation coefficient between the correlated image and the reference image is output. Here, the optical writing type spatial light modulator 105 for Fourier transform is arranged on the front focal plane of the second Fourier transform lens 204, and the light receiving element 205 is arranged on the rear focal plane.

In order to perform a strict Fourier transform, it is preferable to arrange an image to be Fourier-transformed between the front focal plane of each lens performing the Fourier transform or the lens and the rear focal plane. Then, a Fourier transform image is formed on the rear focal plane of the lens. Therefore, in this embodiment, the shift invariant lens array 64, the coordinate conversion lens array 47, the first Fourier transform lens 104, and the second Fourier transform lens 2
An image to be Fourier-transformed is arranged on the front focal plane 04, and the Fourier-transformed image is received on the rear focal plane.

In the above embodiment, the coordinate conversion optical filter array 44 is arranged immediately before the shift invariant optical writing type spatial light modulator 65, and the coordinate conversion optical filter is superimposed on the Fourier transform image of the reference image or the correlated image. The converted image is converted into an intensity distribution and displayed on the shift-invariant optical writing spatial light modulator 65 as a shift-invariant intensity distribution image. However, it is needless to say that the coordinate conversion optical filter array 44 may not be disposed immediately before (on the writing surface side) but immediately after (on the reading surface side) of the shift invariant optical writing type spatial light modulator 65. No. In this case, each Fourier transform image on which the coordinate transformation optical filter is not superimposed is converted into an intensity distribution,
The shift-invariant intensity distribution image is displayed on the shift-invariant optical writing type spatial light modulator 65, and the readout light passes through the coordinate conversion optical filter array 44 twice.

When the Fourier transform image on which the coordinate transforming optical filter is superposed is converted into an intensity distribution as in the above embodiment, the coordinate transforming optical filter must be an amplitude modulation type filter. However, when the coordinate conversion optical filter array 44 is disposed immediately after the shift invariant optical writing type spatial light modulator 65, the coordinate conversion optical filter may be an amplitude modulation type or a phase modulation type.

In this embodiment, the shift-invariant optical writing spatial light modulator 65 and the coordinate conversion optical writing spatial light modulator 1 are used.
3, a reflective liquid crystal light valve using ferroelectric liquid crystal is used as the optical writing spatial light modulator 105 for Fourier transform. When a TN liquid crystal, which is a well-known light modulation material, is used, gradation expression is possible, but the resolution is as low as about 30 [lp / mm], and the operation speed is slow at a video rate (30 Hz). There is a problem that. In order to solve this problem, in this embodiment, a ferroelectric liquid crystal was used as a light modulation material instead of a TN liquid crystal. As a result, the liquid crystal light valve using the ferroelectric liquid crystal has very excellent characteristics of a resolution of about 100 [lp / mm] and an operation speed of about several kHz. However, it should be noted here that a liquid crystal light valve using a ferroelectric liquid crystal normally binarizes and stores a written image because the liquid crystal itself has bistability. Therefore, in this embodiment, since the shift invariant intensity distribution image, the coordinate transformation intensity distribution image, and the Fourier transform intensity distribution image are binarized and stored, the binarization shift invariant intensity distribution image and the binarization coordinate transformation intensity distribution image And a binarized Fourier transform intensity distribution image. Of course, other light modulating materials besides liquid crystal include BSO
A device using an electro-optic crystal such as a crystal is also conceivable. Further, it goes without saying that the principle is the same in a transmissive optical writing type spatial light modulator instead of a reflective type. In addition, it goes without saying that, instead of using an optical writing type spatial light modulator, an imaging device and an electric writing type spatial light modulator can be used in combination.

Next, the structure and operation of a reflective liquid crystal light valve in which a light modulating material is a ferroelectric liquid crystal, which is used as an optical writing type spatial light modulator in this embodiment, will be described. The difference from the conventional liquid crystal light valve is that a ferroelectric liquid crystal having a clear bistability between light transmittance or light reflectance and applied voltage is used as a liquid crystal layer. FIG.
FIG. 3 is a cross-sectional view illustrating a structure of a liquid crystal light valve using a ferroelectric liquid crystal. Transparent substrates 131a and 131b, such as glass or plastic, for sandwiching liquid crystal molecules are provided with transparent electrode layers 132a and 132b, and silicon monoxide obliquely inclined at an angle in the range of 75 to 85 degrees from the normal direction of the transparent substrate. Orientation film layers 133a and 133b are provided.
The transparent substrates 131a and 131b have their alignment film layers 133a,
The ferroelectric liquid crystal layer 134 is sandwiched between the 133b sides by controlling the gap via a spacer 139. Further, the photoconductive layer 135, the light shielding layer 136, and the dielectric mirror 13 are formed on the transparent electrode layer 132a on the light writing side.
7 is laminated between the alignment film layer 133a and non-reflective coating layers 138a and 138b are formed on the outer surfaces of the cells of the transparent substrate 131a on the writing side and the transparent substrate 131b on the reading side.

In the above configuration, when the visible light reflectance of the dielectric mirror 137 is sufficiently large and the influence of the reading light on the photoconductive layer 135 is extremely small, the light shielding layer 136 can be omitted. Further, when the reflectance of the photoconductive layer 135 with respect to the readout light is sufficiently large, and the readout light is sufficiently small and the influence of the readout light on the photoconductive layer 135 is extremely small, the dielectric mirror 137 can be omitted. . However, in this embodiment, since the optical writing type spatial light modulator must be connected in series and writing must be performed to the next-stage optical writing type spatial light modulator using the reading light, the intensity of the reading light should be as small as possible. Stronger is better. Therefore, the shift invariant optical writing type spatial light modulator 65 and the coordinate conversion optical writing type spatial light modulator 13 are used.
Preferably has a light-shielding layer 136 and a dielectric mirror 137. On the other hand, the optical writing type spatial light modulator 1 for Fourier transform
In the case of 05, the sensitivity of the light receiving element 205 in the next stage is high, so that the intensity of the reading light may be low in many cases. Therefore, the optical writing type spatial light modulator 105 for Fourier transform is provided with the light shielding layer 136.
It is often sufficient to have no dielectric mirror 137.

Next, a method for initializing the liquid crystal light valve having the above structure will be described. The first method involves once irradiating the entire writing surface of the liquid crystal light valve with light, and superimposing a pulse voltage, a DC bias voltage, or an AC voltage of 100 Hz to 50 kHz sufficiently higher than the maximum value of the threshold voltage in the bright state. The applied DC bias voltage is applied to the transparent electrode layers 132a and 132
b, the ferroelectric liquid crystal molecules are aligned in a stable state in one direction, and the state is stored. The second method is
Without light irradiation, a pulse voltage or DC bias voltage sufficiently higher than the maximum value of the threshold voltage in darkness or 1
A DC bias voltage on which an AC voltage of 00 Hz to 50 kHz is superimposed is applied between the transparent electrode layers 132a and 132b to align the ferroelectric liquid crystal molecules in a stable state in one direction and store the state. Normally, the maximum value of the threshold voltage at the time of darkness is larger than that at the time of light irradiation.

The operation after the liquid crystal light valve is initialized as described above will be described. Without light irradiation, the threshold voltage is equal to or less than the maximum value of the threshold voltage in darkness, and is equal to or more than the maximum value of the threshold voltage in light irradiation.
Optical writing of an image is performed by a laser beam or the like while applying a DC bias voltage on which an AC voltage of 50 kHz is superimposed between the transparent electrode layers 132a and 132b. Carriers are generated in the photoconductive layer 135 in the region irradiated with the laser, and the generated carriers drift in the direction of the electric field due to the applied voltage. An applied voltage of the opposite polarity to that at the time of initialization above the threshold voltage is applied, and the molecules of the ferroelectric liquid crystal are inverted due to the inversion of spontaneous polarization, and the other stable state is transferred, so the image is binary. And stored.

The binarized and stored image is irradiated with readout light of linearly polarized light whose polarization axis is aligned with the direction of alignment of liquid crystal molecules (or a direction perpendicular thereto) aligned by initialization, and a dielectric material. By passing the light through an analyzer arranged so that the polarization axis is perpendicular (or parallel) to the polarization direction of the light reflected by the mirror 137, reading can be performed in a positive state or a negative state. In the embodiment of FIG. 1, a polarizing beam splitter is used instead of the analyzer.

The threshold value for binarizing an image is adjusted by adjusting the pulse voltage value or pulse width applied between the transparent electrode layers 132a and 132b, the frequency of the AC voltage, or the value of the DC bias voltage. Can be changed. Further, by adjusting the power of the laser to change the light intensity applied to the writing surface, substantially the same effect as when the threshold value is changed can be obtained. By appropriately adjusting the threshold value in this way, it is possible to obtain an accurate image with a small noise component even when the noise component is large and is normally buried in noise.

As shown in FIG. 1, when a plurality of optical writing type spatial light modulators are connected in series to record and read out images one after another, the image to each optical writing type spatial light modulator is It is needless to say that the writing, erasing and reading of the data must be performed in synchronization. In the case of the liquid crystal light valve using the ferroelectric liquid crystal in the present embodiment, the writing pulse and the erasing pulse applied to each liquid crystal light valve and the irradiation time of the reading light and the writing light are synchronized, and each liquid crystal light valve is synchronized. It is necessary that the writing light irradiation time during which the writing light is irradiated to the writing surface of the light valve and the writing voltage application time to the liquid crystal light valve coincide with at least a predetermined time. In such a driving method, the shift invariant optical writing spatial light modulator 65, the coordinate conversion optical writing spatial light modulator 13, and the Fourier transform optical writing spatial light modulator 105 have a frequency of 30 Hz to 2 kHz. It was able to drive at high speed.

Next, the input image electric writing type spatial light modulator 3 used in the above embodiment will be described. In the operation of the input image electric writing spatial light modulator 3, the shift invariant optical writing spatial light modulator 65, the coordinate conversion optical writing spatial light modulator 13, and the Fourier transform light are also used. As in the case of the write-type spatial light modulator 105, it is needless to say that it is necessary to operate in synchronization with another spatial light modulator or laser.

However, as described above, the shift invariant optical writing spatial light modulator 65, the coordinate conversion optical writing spatial light modulator 13, the Fourier transform optical writing spatial light modulator 105, and the like.
When a liquid crystal light valve using a ferroelectric liquid crystal is used, they can operate at a very high speed of about 30 Hz to 2 kHz. However, if a liquid crystal television or the like whose light modulation material is a TN liquid crystal is used as the input image electric writing type spatial light modulator 3, it can be operated only at a low speed of 30 Hz. The speed is limited by the speed of the input image electrical writing spatial light modulator 3. Therefore, as the input image electric writing type spatial light modulator 3, the light modulating material may be an electro-optic ceramic such as PLZT, a magneto-optic material such as yttrium iron garnet, a ferroelectric liquid crystal, or the like so as to enable high-speed operation. It is preferable to use Therefore, in this embodiment, an electric writing type spatial light modulator using a ferroelectric liquid crystal as a light modulating material is used as the input image electric writing type spatial light modulator 3. As a result, the input image electric writing spatial light modulator 3 can operate at a high speed in synchronization with other spatial light modulators.

In the ordinary correlation processing without using coordinate transformation, when the correlated object is rotated by about 10 degrees with respect to the reference object or differs in size by about 20 to 30%, the intensity of the correlation peak is greatly reduced. And couldn't recognize it correctly. However, in the present invention described above, by using a filter for converting to (lnr, θ) coordinates as a coordinate conversion optical filter, the rotation angle can be changed for all 360 degrees of rotation, and the change in size can be reduced to 50 degrees. Even if it changes by about%, it can be recognized without any problem, and pattern recognition that is invariant to changes in rotation and size has become possible. It was also possible to measure the rotation angle and the amount of change in the magnitude from the position where the correlation peak appears. Furthermore, even when a coordinate conversion optical filter for converting to another coordinate system is used, a pattern that is invariant in size, rotation, etc. can be recognized in accordance with the used coordinate conversion.

Further, it is possible to recognize a pattern invariant not only to a change in the rotation or size of the correlated image but also to a shift. For example, in the present embodiment, considering a reference image and a correlated image as shown in FIG. 4, the reference image is stored in the image memory device 43, so that the reference image on the electrical writing type spatial light modulator 3 for the input image is used. The position is unchanged.
On the other hand, as shown in FIG. 8, the position of the correlated image is changed according to the movement of the correlated object 1 by the electric writing type spatial light modulator 3 for the input image.
Move up. However, as described above, by performing Fourier transform to make the correlated image shift invariant as a pre-process of coordinate transformation, even if the position of the correlated image changes, the intensity distribution (shift invariant intensity) of the Fourier transformed image changes. Distribution image) does not change. As a result, since the relative positional relationship between each shift-invariant intensity distribution image and the coordinate conversion optical filter does not change, the coordinate conversion image does not change, and the result of the correlation processing is invariant to the shift of the correlated image. It becomes. Of course, conversely, even when the reference image is changed and the position is shifted, similarly, it is needless to say that the correlation process invariant to the shift of the reference image is possible.

As described above, when the correlated object 1 moves in the three-dimensional space, the size, the direction, the position, and the like of the correlated image on the input image electric writing spatial light modulator 3 change. By using an appropriate coordinate transformation optical filter for these changes, accurate pattern recognition of the correlated image has become possible. Regarding the recognition speed, when a spatial light modulator using a ferroelectric liquid crystal as a light modulating material is used, the spatial light modulator can operate at about 30 Hz to 2 kHz.
High-speed operation of about 0 Hz to 2 kHz was possible.

Further, in the present invention, since a joint conversion type correlator is used instead of a VanderLugt type correlator, the trouble of developing the photographic dry plate and setting the photographic dry plate after the development is unnecessary, and the pattern can be easily formed. Recognition is possible.
In addition, if the reference object is stored in the image memory device 43 in advance, it is possible to easily and quickly change the reference image. Further, since a spatial light modulator having a higher resolution than a photographic dry plate is not required, the influence of vibration and air fluctuation is small, and the optical axis of the optical system can be easily adjusted.

FIG. 6 is a block diagram of another embodiment of the optical pattern recognition device having a coordinate conversion function according to the present invention. This shows a method of using an electric writing type spatial light modulator instead of the shift invariant optical writing type spatial light modulator 65. Since the joint transform correlator is the same as that of the embodiment of FIG. 1, the illustration is omitted. Parts having the same functions as those of the embodiment in FIG. 1 are denoted by the same reference numerals, and the description is omitted or simplified.

The part of the optical coordinate converter different from that of the embodiment shown in FIG. 1 is optically or electronically transformed so that the reference image and the correlated image are shift invariant, and the shift invariant is obtained. Means for displaying the converted reference image and the correlated image on at least one shift-invariant image spatial light modulator, comprising: an input image electric writing spatial light modulator 3 and a shift-invariant lens. An array 64, a shift-invariant imaging device 66, a binarization processing circuit 67, and a shift-invariant electric writing spatial light modulator 68.

The reference image and the correlated image displayed on the input image electric writing spatial light modulator 3 are subjected to Fourier transform similarly to the embodiment shown in FIG. 1, and each Fourier transform image is directly shifted invariant. And is converted into an image signal. After the image signal is binarized by a binarization processing circuit 67, a shift invariant electric writing type spatial light modulator 68 is used.
, Each Fourier transform image of the reference image and the correlated image is converted into a binarized intensity distribution image, and the binarized shift invariant intensity distribution image is transmitted to the shift invariant electric writing spatial light modulator 68. Is displayed. As the shift-invariant imaging device 66 and the shift-invariant electric writing type spatial light modulator 68 in this embodiment, a transmissive spatial light modulator such as a CCD camera or a liquid crystal television is used.

If it is not necessary to binarize the shift-invariant intensity distribution image, it goes without saying that the binarization processing circuit 67 is unnecessary. However, by setting an appropriate threshold value for binarization, a clear binarized shift-invariant intensity distribution image with less noise can be obtained. Therefore, in this embodiment, binarization is performed. The coordinate conversion optical filter array 44 is disposed immediately before or immediately after the shift invariant electric writing spatial light modulator 68, and is a shift invariant intensity distribution image on which the coordinate conversion optical filter is superimposed by coherent light from the coordinate conversion laser 5. Is read. Subsequent processing is the same as in the embodiment shown in FIG.

FIG. 7 is a block diagram of another embodiment of the optical pattern recognition device having a coordinate conversion function according to the present invention. This shows a case where a reference image or a correlated image is not optically converted into a shift invariant, but is electronically performed using a computer. Since the joint transform correlator is the same as that of the embodiment of FIG. 1, the illustration is omitted. Parts having the same functions as those of the embodiment in FIG. 1 are denoted by the same reference numerals, and the description is omitted or simplified.

The optical coordinate converter differs from the embodiment shown in FIG. 1 in that at least one coherent light source is composed of a coordinate conversion laser 5 and a coordinate conversion beam expander 6, which is a two-dimensional reference. The means for obtaining the image and the correlated image includes the input image imaging device 2, the image processing device 42, and the image memory device 43, and optically or electronically shifts the reference image and the correlated image invariably. Means for displaying the reference image and the correlated image converted so as to be shift invariant on at least one spatial light modulator for shift invariant image, comprising a shift invariant computer 69 and a shift invariant computer. Electrical writing type spatial light modulator 68.

First, as in the embodiment shown in FIG. 1, a two-dimensional reference image is obtained from the input image pickup device 2. The reference image is subjected to image processing by the shift invariant computer 69 to calculate the position G of the center of gravity, and then transmitted to the image processing device 42 as a shift invariant intensity distribution image of the reference image.
Store in 3 Next, similarly, the correlated image is subjected to image processing to calculate the position G ′ of the center of gravity, obtain a shift-invariant intensity distribution image based on the position of G ′, and send the image to the image memory device 43 first. The shift invariant intensity distribution image of the reference image stored in the memory is synthesized by the image processing device 42,
It is displayed on the shift invariant electric writing type spatial light modulator 68.
In this case, as shown in FIG. 9, the distance GG ′ between the centers of gravity on the shift-invariant electric writing type spatial light modulator 68 is equal to the distance L between the coordinate transforming optical filters, as shown in FIG. And the coordinate conversion optical filters are combined so that the relative positional relationship between them is the same. The subsequent processing is shown in FIG.
6 is the same as the embodiment shown in FIG. It is needless to say that the coordinate conversion optical filter array 44 may be arranged immediately after the shift invariable electric writing type spatial light modulator 68.

Here, the contents of the image processing performed by the shift invariant computer 69 will be described. First, the reference image and the correlated image input from the input image capturing device are binarized. The reason for this is that an image taken from a normal imaging device has a gradation, and when such an image is processed by a computer, the amount of data increases and the time required for processing increases. Next, the position of the center of gravity G or G 'of the reference image or the correlated image is calculated. Finally, if the position of the center of gravity of the reference image matches the center of gravity of the correlated image, no further processing is performed, and if the center of gravity differs, the position of the correlated image is translated so that the center of gravity is the same. To be.

In the correlated image that has been subjected to the above image processing, the position of the center of gravity is invariant regardless of the movement of the correlated object 1, so that the image becomes a shift-invariant intensity distribution image. Then, the shift invariant intensity distribution images of the reference image and the correlated image are combined so that the distance between the centers of gravity becomes L according to the position of the coordinate transformation optical filter, and the shift invariant electric writing spatial light modulation is performed. Since the relative positional relationship between each shift-invariant intensity distribution image and the coordinate conversion optical filter is invariable when displayed on the unit 68, the correlation process invariant to the shift of the correlated image is performed as in the other embodiments. be able to.

In the above embodiment, the part using the optical writing type spatial light modulator is a combination of an imaging device such as a CCD camera and an electric writing type spatial light modulator such as a liquid crystal television. Needless to say, the principle is the same even if it is replaced with. In the above embodiment, the reference image and the correlated image are simultaneously displayed on the electrical writing type spatial light modulator for input image 3, and the process of making the shift invariant and the optical coordinate conversion are performed in parallel. By the way, after the reference image alone is converted in a shift-invariant manner or coordinate-converted in advance, the image is stored in the image memory device 43. And
The correlated images to be processed one after another are converted into shift-invariant or coordinate-converted only, and then the shift-invariant or coordinate-converted reference image and the correlated image are electrically synthesized by the image processing device 42. Display in parallel to the shift invariant or coordinate conversion image electric writing type spatial light modulator,
It goes without saying that the principle of the correlation processing method is the same.

In the above embodiment, the laser 5 having good coherence such as a gas laser or a semiconductor laser may be used as the coordinate conversion laser 5, the shift invariant laser 61, the Fourier transform laser 101, and the correlation laser 201. Needless to say. Although many lasers are used, it goes without saying that the output from one laser may be separated into many light beams by using a beam splitter.

In the above embodiment, an electric writing type and transmission type electric writing type spatial light modulator 3 for an input image is used as an example in order to display a reference image and a correlated image in parallel. It goes without saying that the principle is the same even if an optical writing type or reflection type spatial light modulator is used. In the above embodiment, the reference object is photographed by the input image pickup device 2,
Although used as a reference image in the image memory device 43, the reference object is always photographed together with the correlated object 1 by the input image imaging device 2 and displayed on the input image electric writing spatial light modulator 3. Needless to say, this may be done. In this case, since there is no need to combine the reference image and the correlated image, the image processing device 42 and the image memory device 43 become unnecessary.

In the above embodiment, the reference image is stored in the image memory device 43. However, the reference object and the correlated object 1 are photographed by different input image pickup devices, and the obtained reference image and correlated image are obtained. Are synthesized by the image processing device 42,
It goes without saying that the input image may be displayed on the electric writing type spatial light modulator 3. In this case, the image memory device 43 becomes unnecessary.

In the above embodiment, the reference image and the correlated image are combined by the image processing device 42 and displayed on the input image electric writing spatial light modulator 3. However, the reference image and the correlated image are combined. Needless to say, the information may be displayed on two input image electrical writing spatial light modulators arranged in parallel without performing the operation. In the above embodiment, one reference image and one
The correlation processing with the number of correlated images is performed, but if many correlated images are to be performed at once by using multiple correlated images or multiple reference images, the shift invariant By increasing the number of shift invariant lenses of the lens array 64, the number of coordinate conversion optical filters of the coordinate conversion optical filter array 44, and the number of coordinate conversion lenses of the coordinate conversion lens array 47, the number of reference images and correlated images obtained Needless to say, it is sufficient to arrange each of them in accordance with the position.

In the above embodiment, the shift invariant lens array 6 is used for the conversion to the optical shift invariant image and the coordinate conversion.
4 and the coordinate conversion lens array 47 are used, but are converted into a shift invariant image or a coordinate conversion image in a separate optical system without using a lens array, and each shift invariant image or coordinate conversion image is synthesized by a beam splitter. It goes without saying that the writing surface of the shift-invariant image optical writing spatial light modulator 65 or the coordinate conversion optical writing spatial light modulator 13 may be irradiated.

[0063]

According to the present invention, as described above, an optical coordinate conversion unit for performing coordinate conversion of a reference image or a correlated image is used as preprocessing of a joint conversion correlator unit.
Accurate recognition is possible even when the size or direction of the correlated object changes. In particular, since the optical coordinate conversion unit performs optical coordinate conversion after converting the reference image and the correlated image so as to be shift invariant, even when the position of the correlated image changes, invariable pattern recognition can be performed. Now you can. Furthermore, when an optical writing type or electric writing type spatial light modulator using a ferroelectric liquid crystal is used, a high-speed operation of about 30 Hz to 2 kHz is possible. In addition, it is possible to easily and quickly change the type of coordinate conversion and the reference image. From the above, very high-speed and accurate pattern recognition becomes possible, and effects such as speeding up and cost reduction can be expected in the field of recognition devices and inspection devices.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an embodiment of an optical pattern recognition device having a coordinate conversion function according to the present invention.

FIG. 2 is a schematic diagram showing an outline of an optical coordinate conversion method.

FIG. 3 is a configuration diagram of a conventional VanderLugt type correlator having a coordinate conversion function.

FIG. 4 is a diagram illustrating an example in which a reference image and a correlated image are separated and displayed on the input image electric writing spatial light modulator 3;

FIG. 5 is a sectional view showing a structure of a liquid crystal light valve using a ferroelectric liquid crystal used in the present invention.

FIG. 6 is a configuration diagram of another embodiment of an optical pattern recognition device having a coordinate conversion function according to the present invention.

FIG. 7 is a configuration diagram of another embodiment of an optical pattern recognition device having a coordinate conversion function according to the present invention.

FIG. 8 is a diagram showing an example in which a correlated image shifted from a reference image is displayed on the input image electric writing spatial light modulator 3;

9 is a diagram illustrating an example in which a shift invariant intensity distribution image of a reference image and a correlated image is displayed on a shift invariable electric writing spatial light modulator 68 at a distance L. FIG.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 correlated object 2 input image imaging device 3 input image electrical writing spatial light modulator 5 coordinate conversion laser 6 coordinate conversion beam expander 12 coordinate conversion polarization beam splitter 13 coordinate conversion optical writing space Light modulator 42 Image processing device 43 Image memory device 44 Coordinate conversion optical filter array 47 Coordinate conversion lens array 48 Mask 61 Shift invariant laser 62 Shift invariant beam expander 64 Shift invariant lens array 65 Shift invariant optical writing type Spatial light modulator 66 shift invariant imaging device 67 binarization processing circuit 68 shift invariant electric writing type spatial light modulator 69 shift invariant computer 101 laser for Fourier transform 102 beam expander for Fourier transform 103 polarization for Fourier transform Beamsplitter 104 First Fourier Interchangeable lens 105 Fourier transform optical writing spatial light modulator 131a, 131b Transparent substrate 132a, 132b Transparent electrode layer 133a, 133b Alignment film layer 134 Ferroelectric liquid crystal layer 135 Photoconductive layer 136 Light shielding layer 137 Dielectric mirror 138a 138b Anti-reflection coating layer 139 Spacer 201 Correlation laser 202 Correlation beam expander 203 Correlation polarization beam splitter 204 Second Fourier transform lens 205 Light receiving element 206 Identification circuit 301 Laser 302 Beam expander 303 Liquid crystal television 304 Coordinate conversion optical filter 305 Beam splitter 306 Mirror 307 Coordinate conversion lens 308 Liquid crystal light valve 309 Beam splitter 310 Polarization beam splitter 311 Fourier transform lens 312 Photographic dry plate 3 13 shutter 314 mirror 315 Fourier transform lens 316 light receiving element

──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Toshiharu Takei 1-24-1-403, Kurosunadai, Chiba-shi, Chiba (72) Inventor Yasuhiro Takemura 3-1-1, Shirai-machi, Shiraicho, Inba-gun, Chiba (56) References JP-A-1-300382 (JP, A) JP-A-2-198414 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G01B 11/00-11/30

Claims (5)

(57) [Claims] 1. An optical pattern recognition apparatus for automatically recognizing and measuring a required pattern by performing an optical correlation process using coherent light on a two-dimensional image obtained from a CCD camera or the like. An optical coordinate conversion unit that obtains a coordinate conversion image obtained by converting at least one reference image including a required target and at least one newly input correlated image into a desired coordinate system; and A joint conversion correlator unit for obtaining a correlation coefficient between the obtained coordinate conversion image of the reference image and the coordinate conversion image of the correlated image, wherein the optical coordinate conversion unit includes at least one coherent light source Means for obtaining the two-dimensional reference image and the correlated image, and Fourier transform using coherent light for the reference image and the correlated image Means for converting each to a shift invariant image by applying, a means for converting each shift invariant image to a shift invariant intensity distribution image and displaying it on a shift invariant image spatial light modulator, and converting to a desired coordinate system At least one coordinate transforming optical filter and at least one lens disposed so as to overlap with the shift invariant image spatial light modulator, and wherein the joint transform correlator unit is configured to perform the optical coordinate transforming. Means for converting each coordinate transformed image of the reference image and the correlated image obtained in the section into a coordinate transformed intensity distribution image, and displaying each coordinate transformed intensity distribution image on a coordinate transformed image spatial light modulator; Means for converting the respective coordinate transformed intensity distribution images displayed on the coordinate transformed image spatial light modulator into a coherent image, and performing a Fourier transform on the coherent image using a lens, Means for obtaining a joint Fourier transform image of the coordinate transform intensity distribution image; means for converting the joint Fourier transform image to a Fourier transform intensity distribution image; and displaying the Fourier transform intensity distribution image on a Fourier transform image spatial light modulator. Means for reading out the Fourier transform intensity distribution image displayed on the Fourier transform image spatial light modulator using coherent light, and performing a Fourier transform on the read out Fourier transform intensity distribution image again using a lens. Means for converting the obtained correlation output image into a correlation signal using an imaging device or a light receiving element, and means for processing the correlation signal to obtain a two-dimensional correlation coefficient between the reference image and the correlated image, respectively An optical pattern recognition device having a coordinate conversion function, comprising: 2. An optical pattern recognition device for automatically recognizing and measuring a required pattern by performing an optical correlation process using coherent light on a two-dimensional image obtained from a CCD camera or the like. An optical coordinate conversion unit that obtains a coordinate conversion image obtained by converting at least one reference image including a required target and at least one newly input correlated image into a desired coordinate system; and A joint conversion correlator unit for obtaining a correlation coefficient between the obtained coordinate conversion image of the reference image and the coordinate conversion image of the correlated image, wherein the optical coordinate conversion unit includes at least one coherent light source Means for obtaining the two-dimensional reference image and the correlated image, and shifting the reference image and the correlated image by image processing using a computer. Means for converting to a variable intensity distribution image and displaying the shift invariant intensity distribution image on at least one spatial light modulator for shift invariant image; and spatial light modulation for shift invariant image for conversion to a desired coordinate system. At least one coordinate conversion optical filter and at least one lens, the joint conversion correlator unit being configured to control the reference image or the target image obtained by the optical coordinate conversion unit. Means for converting each coordinate transformed image of the correlation image into a coordinate transformed intensity distribution image, and displaying each coordinate transformed intensity distribution image on a coordinate transformed image spatial light modulator; and displaying on the coordinate transformed image spatial light modulator. Means for converting each of the obtained coordinate transformed intensity distribution images into a coherent image, and a Fourier transform of the coherent image using a lens, Means for obtaining a Fourier transform image, means for converting the joint Fourier transform image to a Fourier transform intensity distribution image, and displaying the Fourier transform intensity distribution image on a Fourier transform image spatial light modulator; and Means for reading out the Fourier transform intensity distribution image displayed on the optical modulator using coherent light, and a correlation output image obtained by performing a Fourier transform on the read out Fourier transform intensity distribution image again using a lens; Means for converting into a correlation signal using an imaging device or a light receiving element, and means for processing the correlation signal to obtain a two-dimensional correlation coefficient between the reference image and the correlated image, respectively. An optical pattern recognition device having a characteristic coordinate conversion function. 3. The means for converting each shift invariant image into a shift invariant intensity distribution image and displaying the shift invariant image on a spatial light modulator for shift invariant image according to claim 1, An optical pattern recognition device having a coordinate conversion function, which is means for converting into an invariant intensity distribution image and displaying the image on the shift invariant image spatial light modulator. 4. The means for converting each shift invariant image into a binarized shift invariant intensity distribution image and displaying the image on the shift invariant image spatial light modulator according to claim 3, A coordinate conversion function for irradiating a shift invariant image to an optical writing type spatial light modulator for a shift invariant image using a ferroelectric liquid crystal having a bistable memory property and storing the shift invariant image An optical pattern recognition device having: 5. The means for converting each shift invariant image into a binarized shift invariant intensity distribution image and displaying the shift invariant image on the shift invariant image spatial light modulator according to claim 3. Coordinates are converted into an image signal by receiving light using a device, and after binarizing the image signal, input to an electric writing type spatial light modulator for shift invariant image and displayed. An optical pattern recognition device having a conversion function.