JPH0331973A - Optical pattern recognizing device - Google Patents

Abstract

PURPOSE:To execute an optical pattern recognition at a high speed by preparing sum information or difference information in superposing the phase conjugate information of optical information, further superposing the Fourier conversions, recording them to a spatial light modulator, thereafter, reading them again and executing the Fourier conversions. CONSTITUTION:Information recorded to an optical information input means 2 is inputted to a phase conjugate information preparing system 3, it is converted to the phase conjugate information, arbitrary two pieces of optical informations are superposed by a first wave synthesizing system 4, and the sum information is inputted to the one side of an optical switch 5, and the difference information is inputted to the other side. Only the sum information is Fourier-converted by a first Fourier converting system 6 and recorded through a second wave synthesizing system 7 to a spatial light modulator 8. Next, the Fourier-converted difference information is superposed so as to make the sum information and an optical axis to coincide, inputted to the spatial light modulator 8, read by the light from a reading optical system 9, thereafter, Fourier-converted by a second Fourier converting system 10 and detected by a photodetecting system 11. Thus, the optical pattern recognition can be executed at a high speed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an optical pattern recognition device that identifies two-dimensional image information by outputting the difference between input optical information based on the intensity of a correlation peak.

[Summary of the invention]

The present invention creates sum information or difference information of each optical information by superimposing the phase conjugate information of at least two pieces of optical information, and further superimposes the Fourier transform of these sum information or difference information to enable optical input. After recording on a spatial light modulator, it is read out again and subjected to Fourier transformation, allowing the cross-correlation or autocorrelation peak of each input information to be output, making it possible to perform optical pattern recognition at high speed without using an electrical system. The present invention provides an optical pattern recognition device that can perform

[Conventional technology]

The conventional optical pattern LWg, especially in the method using analog optics, is to input light containing a signal image into a massotid filter that records phase conjugate information of a reference image, and then Fourier transform it. It was common practice to measure the correlation peak between the human force image and the (No. 8 image).

Furthermore, in recent years, phase conjugate wave elements that can optically create phase conjugate information in quasi-real time are used to create phase conjugate images of Class 21I optical information, and align them so that their optical axes coincide. This makes it possible to easily create the sum or difference information of two input optical information (A, E, Chu, P, Ye-1M.

Cone 1 Fubizan Optical Engineering,
Vol, 27. No, 5. May385-39H
1988) (A, E.

Chiou P, Yeh, M, Khoshne
visan, Opt, Eng, 27(5), 385
391 (1988)), it became possible to easily obtain the difference between input information using this information without using precise optical components.

[Problem to be solved by the invention]

However, in conventional methods, in order to obtain the correlation of input information, the obtained sum information or difference information must be printed on photographic film in order to convert the obtained sum information or difference information into light intensity information, or , it was not possible to perform high-speed optical pattern recognition because the signal had to be converted into an electrical signal using a camera or the like and then re-inputted into the optical system using a liquid crystal display or the like.

[Means to solve the problem]

The optical pattern recognition device of the present invention includes at least two optical information manual means for converting two-dimensional image information from an external electric system or optical system such as a CCD camera or an imaging optical system into coherent optical image information, and means for obtaining sum information or difference information of the coherent optical image information by forming the phase conjugate 11''tIil of the coherent optical image information and superimposing the phase conjugate 1ff information of the coherent optical image information,
After the above-mentioned sum information or difference M fill is Fourier-transformed by a Fourier-transform lens, the Fourier-transformed sum information or difference information is sequentially transferred to a recording mode and a recording mode, or an erasing mode and an erasing mode, or a recording mode. mode and erase mode, the optical axis is superimposed on a spatial light modulator that can input light so that the optical axis coincides with the optical axis, and the data is recorded, read out again using coherent light, Fourier transformed by a Fourier transform lens, and then recorded by a photodetector. By using a configuration having a means for outputting the cross-correlation or autocorrelation peak of each input information by detecting the output, printing can be performed without development processing or electrical system intervention. The above-mentioned problems have been solved.

[Function] In the optical pattern recognition device of the present invention, at least one of the at least two optical information manual means is a spatial light modulator capable of electrical input such as a liquid crystal display, and the optical information inputted by the spatial light modulator is and b, it is possible to convert into sum information or difference information of a+b or a-b by forming and superimposing phase conjugate information.

Then, after Fourier transforming a+b or a-b using a Fourier transform lens, only the sum information A+B, which was Fourier transformed using an optical switch, is used to create a liquid crystal element that enables optical input of information using the switching characteristics of the photoconductive film. (Hereinafter, this will be referred to as an optical writing type crystalline spatial light modulator), B I
I! Information is recorded in a spatial light modulator such as a device (hereinafter referred to as FROM) that enables optical input of information by utilizing the photoconductive properties of the S i OH0 single crystal and the Bonokels effect. Next, the Fourier-transformed difference information /IB is superimposed on the Fourier-transformed sum information A+B so that the optical axes coincide with each other, and input to the above-mentioned spatial light modulator. At this time, if the difference information is input manually in the same mode as the sum information, the above spatial light modulator contains lA + Bl'' + l8 to l'', that is, 2 (A
`` + B 〉 is recorded, and if the difference information is manually input in a mode different from the sum information, l A + B l is stored in the spatial light modulator.
-l A-B The real part of l'' or 2A''B is recorded. Therefore, if the input information recorded in the spatial light modulator is read out again and Fourier transformed using a lens, the sum of the autocorrelation or the cross-correlation peak of the input information a and b can be obtained independently. It is possible to identify how similar the input optical information 1i1a and b are by the presence or absence or strength of the correlation peak.

〔Example〕

DESCRIPTION OF THE PREFERRED EMBODIMENTS The optical pattern recognition device of the present invention will be described below using examples with reference to the drawings. FIG. 1 is a basic configuration diagram of an optical pattern recognition device of the present invention.

In FIG. 1, l is an input light source system, 2 is an optical information manual means,
3 is a phase conjugate information generation system, 4 is a first multiplexing system, 5 is an optical switch, 6 is a first Fourier transform system, 7 is a second multiplexing system, 8 is a spatial light modulation system capable of inputting light 3g , 9 is a reading light source system, IO is a second Fourier transform system, 11 is a photodetection system, 1
2a and +2b are external electrical systems or optical systems. Further, in Fig. 1, the solid line with an arrow indicates the flow of optical information, where a is the optical information input from the first optical information input means, al is the phase conjugate information of a, and A+ is the Fourier transform of al. It is. The input light source system 1 expands or shapes coherent light obtained from an argon laser, a helium-neon laser, a semiconductor laser, or the like, and demultiplexes it into each of the optical information input means 2 at an appropriate intensity. The optical information input means 2 uses a spatial luminous intensity 4ml for inputting information using an electrical means such as a liquid crystal display or an electrochrominok display, or an optical means such as an optical writing type liquid crystal spatial light modulator or FROM. By using a spatial light modulator that inputs information, it is possible to manually input optical information from an external electrical system or optical system 12a. In this case, it is necessary to use at least two optical information input means 2, n (n is a natural number) for boat fishing. The information recorded in the optical information manual means 2 is read out by the light from the manual light source system 1, enters the phase conjugate information generation system 3, and is converted into 0 phase conjugate information. Optical information is electrically combined with any two optical information by the first acoustic wave system 4, and reaches the optical switch 5. At this time, one of the optical switches 5 has the sum information of the above arbitrary two optical information,
Difference information is input to the other. Next, the optical switch where the sum information is input is in a state in which the light passes through it, and the other optical switch is in a state in which no light passes through, and only the sum information is Fourier-transformed in the first Fourier transform system 6, and then transferred to the second multiplexing system. 7 to record on a spatial optical scanning device 8 capable of optical input. Next, the optical switch where the difference information is manually entered is in a state where light passes through, and the other optical switch is in a state where light does not pass through, and only the difference information is Fourier-transformed in the first Fourier transform system 6 and transferred to the second multiplexing system. 7 to a spatial light intensity alit device 8 capable of inputting light. At this time, the second multiplexing system is an optical system in which optical components are arranged so that the above-mentioned sum information and difference information can be recorded on the spatial light modulator 8 which can be optically inputted so that the optical axes thereof overlap. The Fourier-transformed sum information and difference information superimposed on the spatial light modulation 2S8 that can be optically inputted are read out by light from the readout optical system 9, and then optically converted by the second Fourier transformation system 10. The light is Fourier transformed and detected by the photodetection system 11. The output detected at this time is the spatial luminous intensity that allows optical input in the same mode as the above sum information and difference information! If recorded in the lil unit 8, it becomes an autocorrelation of the two arbitrary pieces of information, and if the sum information and the difference information are recorded in mutually opposite modes, it becomes a cross-correlation of the two pieces of arbitrary information. In this case, the same mode means that when light enters the spatial light modulator 7 that can input light, both the Fourier-transformed sum information and the Fourier-transformed difference information are in the recording mode, or the Fourier-transformed sum information is also Fourier-transformed. This means that the detected difference information is also recorded in the erase mode, and the reverse mode means that the spatial light modulator 7 that can input light
When light is incident on , the Fourier-transformed sum information is recorded in the recording mode and the Fourier-transformed difference information is recorded in the erasure mode, or the Fourier-transformed sum information is recorded in the erasure mode, and the Fourier-transformed difference information is recorded in the erasure mode. It means a recording mode in which information is recorded in recording mode. Moreover, the phase conjugate information generation system 3 is B i +! S i Ozm single crystal and B
a Photorefractive materials such as T I Oy single crystals (Photo
This is an optical system that can perform a self-pumping method or a degenerate four-wave mixing method using a refractive material. Note that the output signal obtained from the photodetection system 11 enters the external electrical system or optical system +2b again and is processed.

Next, Example 2.3 will be described. FIG. 2 is a block diagram of an optical pattern recognition device using an i3-type spatial light modulator. In Fig. 2, 13 is an argon laser;
4 is a first beam expander, 15 is a first half mirror, 116 is a second half mirror, 17 is a first liquid crystal display, 18 is a second liquid crystal display, 19 is a first
, 20 is a second mirror, 21 is a first condenser lens, 22 is a second condenser lens, 23 is a BaTiO3 single crystal, 24 is a first liquid crystal light shuffler, and 25 is a second liquid crystal light 26 is a first Fourier transform lens, 27 is a second Fourier transform lens, 28 is a third mirror, 29 is a third half mirror 130 is a fourth half mirror 131
is a helium neon laser, 32 is a second beam expander, 33 is a third crystal light, 34 is a PROM,
35 is a third Fourier transform lens, and 36 is a CCD.

13, 14, 15, and 16 are input light source system 1 in Fig. 1.
, 17 and 18 are the optical information input means 2 in FIG.
19゜20.21.23 is the phase conjugate information generation system 3 in Fig. 1, 15.16 is the first multiplexing system 4 in Fig. 1, and 24.25 is the optical switch 8J in Fig. 1. 5,
26.27 is the first Fourier transform system 6 in FIG. 1, 28°29 is the second multiplexing system 7 in FIG.
30 to the spatial light modulator 8 capable of optical input in FIG.
31, 32, and 33 correspond to the readout light source system 9 in FIG. 1, and 36 corresponds to the photodetection system 11 in FIG. 1, respectively. As the first liquid crystal display and the second liquid crystal display 17, 18, simple matrix type black and white liquid crystal displays used in ordinary commercial televisions were used. If the display characteristics of human information do not change due to argon laser irradiation, T P
An active matrix black-and-white liquid crystal display using a high-speed FTLM transistor may also be used. Also, BaTi0. The angle of incidence of light on the single crystal 23 is BaTi0. , the angles were -15 degrees and -19 degrees with respect to the a-axis of the single crystal. In addition, the first liquid crystal display 1
The size of the input image to the liquid crystal display 7 and the second liquid crystal display 18 was approximately 10 m1 in diameter.

Furthermore, the beam diameter of the argon laser 13, which is the writing light, is approximately 12 m1. The beam diameter of the helium neon laser 31 serving as the readout light was approximately 15 mm in diameter.

Further, the output of the argon laser 13 is 100 rnW, and the output of the helium neon laser 31 is 5 mW.

The wavelength of the argon laser 13 used was 514 nm, and the wavelength of the helium neon laser 31 was 633 nmri. Further, the first liquid crystal display 17, the second liquid crystal ice display 18, the first condensing lens 21, and the second condensing lens 22 are arranged at positions conjugate to the second half mirror 16. Further, the difference between the optical path length from 16 to 23 via 19 and the optical path length from 16 to 22 to 23 was made not to be longer than the coherent length of the argon laser 13. Also, the first Fourier transform lens 26 and the second
The Fourier transform lenses 27 were arranged so that 17 and 18 were at their front focal points, and 34 was arranged at the back focal point of 26 and 27. At this time, if the size of the input image to 34 is not appropriate, it is necessary to place an imaging magnifying lens between 29 and 34 to magnify the Fourier transformed image formed by 26 or 27 before inputting it to 34. 3
4 and 36 are placed at the front and back focal positions of 35, respectively, and if the size of the human-powered image on 36 is not appropriate, 3 is placed again.
It is necessary to place an image magnifying lens between 5 and 36 to magnify the Fourier transformed image formed by 35 and input it to 36. Furthermore, a DC high voltage power supply is used for FROM34.
．． A DC voltage of 6 to 2.0 was applied. Also,
The input image used was a Gothic alphabet capital letter or a clock movement. The input method is to first input the alphabet or one of the clock movements stored in the computer as a reference image to the first liquid crystal display 17 via a video signal, or to write it on paper as a signal image. Any clock movement placed on the uppercase letters of the Gothic alphabet or the heltoconhair is captured using a CCD camera so that it has the same size and orientation as the reference image input to the first liquid crystal display 17. 2 LCD display 1
Entered 8. The 514 nm laser beam emitted from 13 is expanded into a beam by 14, and the beam is expanded by 15 to approximately 5 nm.
It transmits 0%, is split by about 50% by 16, and enters 17 and 18. The laser beam incident on 17 is loaded with the information of the reference image of 17 and is inverted by 1.1 at 19 to become 21G.

Therefore, while being condensed, it is reflected at 2o and is not focused at 23e.
23 generates a phase l projection of the reference image of I7 by self-pumping action, and the phase conjugate image of this reference image passes through 21.19 and 17 to 16? This is reached. Similarly, 18 input statues 2
It reaches 23 after going through 2 and 20, and 4 due to the self-pumping action of 23.
is converted into a phase conjugate image and again 20.22 and 184
Eventually it will reach 16. In this way, the phase conjugate images of the reference image and the input image, which are 1.l at 16, are combined at 16 and 15 and then reach 24 and 25. First, only 24 is set to transparent state B, and 25 is set to non-transparent state. Only the sum information of reference 1n information and human power information is Fourier transformed at 26, and 29. Ko0
-t- recorded in 34 through. At this time, a negative voltage is applied to 34 on the human power side, and the output side is grounded. Alternatively, it is assumed that the input surface side is grounded and the output surface side is applied with a positive voltage. Hereinafter, this state will be referred to as the recording mode. If 0th order 24 is set to an opaque state and 25 is set to a transparent state, only the difference information between the reference information and the manual information is Fourier transformed at 27 and recorded at 34 through 29.30. be done. At this time, it is assumed that both the human power side and the output side of 34 are grounded, or a positive voltage is applied to the human power side, and the output side is grounded. Hereinafter, this state will be referred to as erasure mode. At this time, the sum information and difference information are 34
The optical axes are recorded on the top to match. Next, the information recorded in 34 is read out via 32, 33, and 30 using 633 nm light from 31, and the information recorded in 34 is Fourier-transformed in 35.
6, a cross-correlation peak between the reference information and the human information is detected. Furthermore, if the sum information is recorded in erase mode and the difference information is recorded in 34 in record mode, the output obtained in 36 becomes the autocorrelation peak of the reference information and input information. In order to obtain the autocorrelation peak as described above, it is necessary to irradiate the FROM 34 with light having a uniform intensity distribution that does not contain any information and record δ before erasing the sum information. In this way, we were able to semi-optically obtain a correlated image of the alpha henot capital or the clock movement, and we were also able to perform optical pattern recognition at video rate.

Moreover, an optical writing type spatial light modulator, which is a reflection type spatial light modulator, can be used instead of the FROM, which is the spatial light i-tuning 134 in FIG. Optical writing type liquid crystal spatial light modulators using liquid crystal are well known as optical writing type spatial light modulators, and there are types using TN liquid crystal as the liquid crystal material and types using strong electrostatic liquid crystal. Those using a TN liquid crystal can record gradations, but the response speed is slow, and those using a strong electric liquid crystal have a fast response speed, but have the disadvantage that they can only record 2(a). type of light 3
The read-in type liquid crystal spatial light modulator also has significantly better resolution than FROM and a sufficiently large contrast value, so it is extremely effective to use the optical pattern recognition device of the present invention. FIG. 2 shows a configuration diagram of an optical pattern recognition device using an optical writing type ferroelectric liquid crystal spatial light modulator. 37 is an optically written ferroelectric liquid crystal spatial light modulator. The only difference in FIG. 3 from FIG. 2 is that the insertion position of the readout optical system is on the detector side of the spatial light modulator. However, as mentioned above, 37 is 50~
Because it has an extremely high resolution of 1501 P/mm, it is possible to record optical information with a high spatial frequency, which makes it possible to improve the accuracy of optical pattern recognition and at the same time reduce the size of the device. . However, since the optical writing type ferroelectric liquid crystal spatial light modulator used in the present invention can only perform binary recording, the operating threshold voltage for the writing light intensity was set so as to obtain the highest pattern recognition rate. Such an operation is not necessary when using a spatial light modulator using TN liquid crystal.

Figure 4 shows Ba as a photovoltaic crystal as a phase conjugate generation system.
FIG. 2 is a configuration diagram of an optical pattern recognition device using a degenerate four-wave mixing method using TfOi.

38 is the fifth half mirror, 139 is the fourth mirror, 4° is the fifth mirror, 41 is the sixth mirror, 42 is the seventh mirror, 43 is the sixth half mirror, and 144 is the half wavelength It is a board. The degenerate four-wave mixing method is the fourth method because it allows the convergence of human-powered light.
As shown in FIGS. I7 and 1.8, even when the intensity of optical information input manually becomes extremely weak, phase conjugate information with large intensity can be obtained. Figure 4 shows "C19,20"
, 2, 22.23.3B, 39.40, 41.42.
43.44 is the IJI phase conjugate information generation system.

The systems other than the phase conjugate information generation system in FIG. 4 have the same configuration as in FIG. 3. 15.42 in FIG. 4 produces probe light, and 38, 39, 44.40.41 produce pump light. The probe light created in this way interferes with the input light information from FIG. 4 17 and the reference light information from FIG. 4 18 to record a hologram at 23 in FIG. 4, and the pump light is recorded. The hologram is reproduced to create phase conjugate information of the manual light information and reference light information.The reflectance of 38 is 50%, and 44 is incoherent between the probe light, input light information, reference light information, and pump light. This is to rotate the direction of the linearly polarized light of the pump light by 180 degrees. When the optical pattern recognition device configured as shown in Fig. 4 was operated in the same way as the optical pattern recognition device shown in Fig. 3, it was possible to obtain a highly sensitive optical pattern even when only input light information of extremely weak intensity can be used. Mini R
I was able to do h.

As the photorefractive crystal used in the optical pattern recognition device of the present invention described above, BaTi0. In addition to S t,
Ba, −, Nb0i (0<x<11 and LiNb0
1 and B i lzS i Ox. It was possible to use crystals such as Further, a solid-state laser excited by a semiconductor laser may be used as the input light a13, and a semiconductor laser may be used as the readout light source 31. In addition, a spatial light modulator that can input light arbitrarily - care must be taken to ensure that the manual Fourier-transformed sum information and difference information are accurately superimposed and recorded without magnification, image rotation, or lateral shift. .

〔Effect of the invention〕

As described above, the optical pattern E29 & device of the present invention uses at least two optical information systems that convert two-dimensional image information from an external electrical system or optical system such as a CCD camera or an imaging optical system into coherent optical image information. and means for forming phase conjugate information of each of these coherent optical image information and superimposing the phase conjugate in fFJ to obtain sum information or difference information of the coherent optical image information, After Fourier transforming the iut+ using a Fourier transform lens, the Fourier transformed sum information or difference information is sequentially optically transmitted through an optical switch in a recording mode and a recording mode, or an erasing mode and an erasing mode, or a recording mode and an erasing mode. The two-dimensional image information is recorded by superimposing it on a spatial light modulator that can be input so that the optical axes match, and is read out again using coherent light. After being Fourier transformed by a Fourier transform lens, the two-dimensional image information is transmitted to a photodetector, etc. By having a structure that has means for detecting cross-correlation or autocorrelation peaks and converting them into electrical signals, optical pattern recognition can be performed at high speed, with high precision, or with high sensitivity, and it is possible to recognize characters or figures. The effects on systems such as manuscript copying devices, Robonoto systems, shape measuring devices, and voice recognition devices are extremely large.

[Brief explanation of drawings]

FIG. 1 is a basic configuration diagram of an optical pattern recognition device of the present invention, FIG. 2 is a configuration diagram of an optical pattern recognition device using a transmission type spatial light modulator, and FIG. 3 is a configuration diagram of an optical pattern recognition device using a transmission type spatial light modulator. FIG. 4 is a block diagram of an optical pattern recognition device using the optical variable j112m, and FIG. 2 ・ ・ 3 ・ ・ 4 ・ ・ 5 ・ ・ 6 ・ ・ 7 ・ 8 ・ ・ 9 ・ 10 ・ ・ 11 ・ 2a 13 ・ ・ 14 ・ 15 ・ ・ Input light source system/light information input means phase Conjugate information generation system, first multiplexing system, optical switch, first Fourier transform system, second multiplexing system, spatial light modulator capable of optical input, reading light source system, second Fourier transform system, Photodetection system 12b...external electrical system or optical system, first argon laser beam expander, first half mirror 16, 17, 18, 19, 20, 21, 22, 23・ ・ 24 ・ 25 ・ 26 ・ 27 ・ 28 ・ 29 ・ 30 ・ 31 ・ 32 ・ 33 ・ 34 ・ 35 ・ ・ 2nd half mirror ・ 1st liquid crystal display Second liquid crystal display, first mirror, second mirror, first condensing lens, second condensing lens, B a T i 03 jA4!・First liquid crystal optical shutter ・Second LCD optical shutter ・First Fourier transform lens ・Second Fourier transform lens ・Third mirror ・Third half mirror ・Fourth half mirror Helium neon laser ・Second Fourier transform lens 2 beam expander, 3rd liquid crystal optical shaft, PROM, 3rd Fourier transformation 1 negative lens 36, 37, 38, 39, 40, 41, 42, 3 44, CD Light pollution Built-in ferroelectric liquid crystal spatial light modulator・Fifth half mirror・Fourth mirror・Fifth mirror Sixth mirror・Seventh mirror Sixth half mirror More than half wave plate 1- Output 1 person Seiko Electronic Industry Co., Ltd.

Claims (1)

[Scope of Claims] At least two optical information input means for converting two-dimensional image information from an external electric system or optical system such as a CCD camera or an imaging optical system into coherent optical image information, and coherent light of each of these optical information input means. It has means for obtaining sum information or difference information of the coherent optical image information by forming phase conjugate information of the image information and superimposing the phase conjugate information, and Fourier transforms the sum information or difference information using a Fourier transform lens. After that, the Fourier-transformed sum information or difference information is sequentially inputted into a spatial light modulator capable of optical input via an optical switch in a recording mode and said mode, or an erasing mode and erasing mode, or a recording mode and erasing mode. The two-dimensional image information is recorded overlappingly so that the optical axes are aligned, read out again using coherent light, Fourier transformed using a Fourier transform lens, and then the cross-correlation or autocorrelation peak of the two-dimensional image information is detected by a photodetector etc. An optical pattern recognition device having means for converting a signal into an electrical signal.