JPH01201632A - Method and device for optical parallel processing - Google Patents

Abstract

PURPOSE:To utilize the parallel property and the high speed property of light and to execute an optical parallel processing which can execute a multistage input and a multistage processing, and also, which is programmable by replacing the binary state of digital information with two polarizations intersecting orthogonally with each other, separating them so that there is no overlap in accordance with the polarization and selecting its light. CONSTITUTION:A light beam from a light source 1-1 becomes a linearly polarized light by a polarizer 1-2. The linearly polarized light is varied in the polarizing direction by an input signal 1-3 in a space modulator 1-3'. That is, in the binary state, at the time of one state, the polarized light rotates by 90 deg. and becomes a linearly polarized light in the vertical direction in parallel to the paper surface, and at the time of the other state, the polarized light is not varied and remains vertical to the paper surface, as it is. The light beam passing through the space modulator 1-3' is separated spatially in accordance with the polarized light by a double-refraction plate 1-4. In such a way, the light beam separated spatially in accordance with the state of the input signal 1-3 is selected to a carnel 1-5. Therefore, the result of processing can be obtained at the speed of light, and the processing speed is high. Also, parallel image information consisting of a large quantity of picture elements can be processed as it is in the state that the parallel property is high.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a method and apparatus for optically processing digital information in a binary state, and in particular to an optical parallel processing method suitable for image processing and various calculations. The present invention relates to a processing method and apparatus.

[Conventional technology]

FIG. 3 shows an example of a conventional optical parallel processing method.

This is an example of image processing, and as shown in FIG. 3(a), it is composed of a point light source array 3-1, an encoded image 3-2, a screen 3-3, and a decoding mask 3-4. A binary image (referred to as A and B) is encoded in advance according to the value of each pixel, and is input as an encoded image 3-2. Point light source array 3-1 is turned on, encoded image 3-2 is projected onto screen 3-3, and a calculation result is obtained through decoding mask 3-4. In this method, the point light source array 3-1
By changing the pattern, the type of calculation is changed. This correspondence is shown in FIG. 3(b). For details, please refer to Jun Tanida "Parallel Optical Logic Operations" (Optics, Vol. 15, No. 5, published by the Optics Conference of the Japan Society of Applied Physics, 1986, 1).
(October).

[Problem to be solved by the invention]

Conventional optical parallel processing methods have the following problems.

■ Multi-stage processing is not possible. Since the computation is completed by projecting the decoding mask onto the closely attached screen, the optical signal projected onto the -degree screen cannot be used as input light to the next stage.

■ Difficult to encode/input. When attempting to perform calculations between multiple images, all images must be combined and encoded as input, but the complexity rapidly increases as the number of images increases. This becomes a load that cannot be ignored during input processing.

■ It is impossible to process all calculations using light. Since the operation is realized by logical sum (OR) using light, once OR is realized, logical product (AND) cannot be executed.

Therefore, the result of the OR operation projected onto the -degree screen is encoded as the input signal again, and the result is again ORed.
(Since AND cannot be performed, the positive and negative logic is reversed and OR is implemented. However.

It is meaningless unless the -th OR has already been completed and the result saved. This is because OR becomes AND. ) all operations are possible. This involves the process of collecting processing results from light, encoding them, inputting them again, and projecting them onto a screen, which is a very complicated process, and the high-speed nature of light becomes ineffective.

An object of the present invention is to provide an optical parallel processing method and apparatus that can utilize the parallelism and high speed of light, are capable of multi-stage input and multi-stage processing, and are programmable.

[Means to solve the problem]

In the optical parallel processing method of the present invention, the binary state of digital information is replaced with two orthogonal polarized lights, and the replaced lights are separated according to the polarization so that they do not spatially overlap. The separated light is selected in accordance with the processing.

The optical parallel processing device of the present invention includes a light source, a polarizer that polarizes light from the light source, and polarized light of the polarizer.

a spatial modulator that replaces light with two orthogonal polarizations corresponding to the binary state of digital information;
It has an optical element that separates two polarized lights so that they do not spatially overlap, and a polarization selection element that selects the optical output of the optical element in accordance with the processing. Using this as the basic configuration, furthermore, eight stages (N is an integer of 2 or more) of pairs of the spatial modulator and optical element are arranged in series, and 2 m positions (m = 1 .2.-N-1) enters the next stage spatial modulator and replaces each with two orthogonal polarized lights, and each output light of the spatial modulator is spatially transformed again according to the polarization. 2 B+ so that there is no overlap
Separate into one position. Furthermore, a λ/4 plate (λ is the wavelength of the incident light) and a polarizer may be installed after the optical element.

[For production]

The basic idea of the present invention is to replace the binary state of input digital information with two polarized lights, spatially separate them according to the polarizations, and select the separated lights. The correspondence between this input information and the replacement with polarized light is very simple,
Easy to input.

In addition, in the present invention, operations are performed by performing polarization encoding, repeating operations to spatially separate encoded light, and selecting light according to the processing content (this selection is called a kernel). conduct. Therefore, multi-stage processing is extremely easy, and all calculations can be performed using light, making use of the advantages of light. Furthermore, in the present invention, by installing a λ/4 plate polarizer after the optical element for spatial separation, the polarized light is arranged well and the correspondence between the spatial arrangement and logic becomes easy to understand.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows the simplest embodiment of the present invention, in which a light source 1-
1, a polarizer 1-2, an input signal 1-3, a transmissive spatial modulator 1-3', a birefringent plate 1-4, and a kernel 1-5.

The light from the light source 1-1 becomes linearly polarized light by the polarizer 1-2. In FIG. 1, the polarization is perpendicular to the plane of the paper, and the linearly polarized light, indicated by "O" in the figure, undergoes a change in polarization direction by the input signal 1-3 in the spatial modulator 1-3'. In other words, in one of the binary states (referred to as A), the polarization is 90°.
It rotates and becomes linearly polarized light in the vertical direction parallel to the plane of the paper (dashed arrow in the figure), and in the other state (designated as X), the polarization does not change and remains perpendicular to the plane of the paper. Spatial modulator 1-3
The light that has passed through ' is spatially separated by the birefringent plate 1-4 according to the polarization. In FIG. 1, polarized light indicated by a circle travels straight, and polarized light indicated by an arrow moves upward. In this way, spatially separated lights (one light is only on one side) are selected for the kernels 1-5 depending on the state of the input signals 1-3.

FIG. 2 shows an embodiment in which the input shown in FIG.
, an input signal 2-4, a reflective spatial modulator 2-4', a birefringent plate 2-5, and a kernel 2-6. The light from the light source 2-1 becomes linearly polarized light by the polarizer 2-2, passes through the half mirror 2-3, and enters the spatial modulator 2-4'. The reflected light from the spatial modulator 2-4' has its polarization direction unchanged or rotated by 90 degrees depending on the binary state (A or τ) of the input signal 2-4. This reflected light is reflected by the half mirror 2-3, enters the birefringent plate 2-5, and is spatially separated. The light emitted from the birefringent plate 2-5 is converted into the kernel 2.
The selection with -6 is the same as in FIG.

FIG. 4 shows the positional relationship between the light beam and the polarized light when the light emitted from the birefringent plate 1-4.2-5 in FIGS. 1 and 2 is viewed from the front. That is, in state X, the position is horizontally polarized and the position is lower, and in state A, the position is vertically polarized and upper.

FIG. 5 shows an example of how to configure the kernel 1-5, 2-6, which includes a polarizer 5-1, a polarization rotation element 5-2, an analyzer 5-3, and an instruction signal 5-4.
It is made up of. The polarizer 5-1 and the analyzer 5-3 are orthogonal to each other, and are tilted 45 degrees in opposite directions from the horizontal (at the same time, 45 degrees from the vertical), as shown below the dotted line.
(tilted) linearly polarized light is passed through each. Note that a configuration without the polarizer 5-1 is also possible. In this case, the analyzer 5-3 is set to pass horizontally or vertically polarized light. FIG. 6 shows the detailed configuration of the polarization rotator 5-2, which is constructed by arranging two polarization rotation element pieces so as to correspond to two spatially separated light beams. Each includes a medium having an electro-optical effect, such as a crystal or liquid crystal 6-1, and a transparent electrode 6-2 to which an electric field of a calculation signal 5-4 is applied.

The thickness of the medium 6-1 is such that the polarization rotation due to the electric field is Oo or 90
Set it so that it can be controlled at °.

The light spatially separated into two by the birefringence plate 1-4 and 2-5 in Figures 1 and 2 passes through the polarizer 5-1 and enters each month of the polarization rotation element 5-2. However, when an electric field is applied to the electrode 6-2 by the calculation signal 5-4 and the polarized light is rotated by 90 degrees, the incident light passes through the analyzer 5-3, and when the polarization rotation is 0 degrees, the light is transmitted through the analyzer 5-3. -3 cannot be passed through. Logical operations can be performed by selecting this spatially separated light. Figure 7 shows this.The "O" mark indicates "selection" (on), which allows the analyzer 5-3 to pass through, and the "・" mark indicates "non-selection", which does not pass through the analyzer 5-3 (off). This represents the case where 7
-1 is logic "1", 7-2 is logic "0", 7-3 is logic X, and 7-4 is logic A.

FIG. 8 shows an embodiment in which calculations between two side images are performed using two transmissive spatial modulators, including a light source 8-1 and a polarizer 8-1.
2. First input signal 8-3, first transmissive spatial modulator 8
-3', a birefringent plate 8-4, a second input signal g8-5, a second transmissive spatial modulator 8-5', a birefringent plate 8-6, and a kernel 8-7.

The light from the light source 8-1 becomes linearly polarized light by the polarizer 8-2. In the figure, the polarization is perpendicular to the plane of the paper and is indicated by an "○" mark. The polarization of the linearly polarized light is changed by the first spatial modulator 8-3' depending on the state of the input signal 8-3. That is, input signal 8-3
When in one of the binary states (A), the polarized light is rotated by 90' and becomes linearly polarized light in the vertical direction parallel to the plane of the paper (indicated by the dashed arrow in the figure), and the other state is In the state (X), the polarization does not change and remains perpendicular to the plane of the paper.

The light that has passed through the spatial modulator 8-3' is spatially separated by the birefringent plate 8-4 according to the polarization. In the figure, polarized light indicated by a circle travels straight, and polarized light indicated by an arrow moves upward.

The output light of the birefringent plate 8-4 is further transmitted to the second spatial modulator 8.
-5' changes the polarization in the state of the input signal 8-5.

Here, when the input signal 8-5 is in one of the binary states (referred to as B), the polarized light is rotated by 90°. In the other state (say 100), the polarization does not change. The emitted light from the spatial modulator 8-5' is separated spatially (front and rear in the figure, left and right when viewed from the front) according to the polarization by a birefringent plate 8-6. That is, the birefringent plate 8-6 allows polarized light perpendicular to the plane of the paper to travel straight, and separates polarized light parallel to the plane of the paper in the front-rear direction (to the right when viewed from the front). From the lights separated in this way, a predetermined light is selected by a kernel 8-7 set according to the calculation, and a logical calculation can be performed.

FIG. 9 shows an embodiment in which calculations are performed between two side images using two reflective spatial modulators, including a light source 9-1, a polarizer 9-2,
Half mirror 9-3, first input signal 9-4, first spatial modulator 9-4', birefringent plate 9-5, half mirror 9
-6, second input signal 9-7, second spatial modulator 9-7
', a birefringent plate 9-8, and a kernel 9-9. The process up to obtaining separated light in accordance with the first input signal 9-4 with birefringent plate 9-5 is the same as obtaining separated light with birefringent plate 2-5 in FIG. The output light from the birefringent plate 9-5 passes through a half mirror 9-6 and is sent to a second spatial modulator 9-7.
', and in accordance with the binary state of the second input signal 9-7, when in one state (assumed B), the polarization is rotated by 90°,
In the other state (double), the reflected output light remains unchanged in polarization.

The reflected output light is reflected by the half mirror 9-6, and is passed through the birefringent plate 9.
-8, and spatially depending on the polarization state (in the figure, front and back,
separated into left and right (when viewed from the front).

The birefringent plate 9-8 allows polarized light perpendicular to the plane of the paper to travel straight,
Use a device that separates polarized light parallel to the plane of the paper in the front-back direction (right side when viewed from the front). From the lights separated in this way, a predetermined light is selected by a kernel 9-9 set according to the calculation, and a logical calculation can be performed.

FIG. 10 shows the spatial arrangement of polarized light when the state of the light emitted from the birefringent plate 8-6 in FIG. 8 is viewed from the front with respect to the light beam. A logical operation corresponding to this is shown in FIG.

The lower left of FIG. 10 is the horizontally polarized light, which in FIG. 8 is the polarized light perpendicular to the plane of the paper. Since the first input pixel a was in state W, the polarized light perpendicular to the plane of the paper (output of 8-2) goes straight through the birefringent plate 8-4 without changing its polarization, and the second input pixel a Since it was in state B, the polarization did not change and the birefringent plate 8-6
This is the case when you go straight ahead. This is the logical operation A- in Figure 12.
Corresponds to B.

The upper left of FIG. 10 is the horizontally polarized light, which in FIG. 8 is the polarized light perpendicular to the plane of the paper. Since the first input pixel a was in state A, the polarized light (output of 8-2) perpendicular to the plane of the paper changes and becomes vertically polarized light, which moves upward at the birefringent plate 8-4. , since the second input pixel was in state B, the polarized light changed and became horizontally polarized light (perpendicular to the plane of the paper in FIG. 8) and went straight through the birefringent plate 8-6. This corresponds to logical operation AB in FIG.

The lower right of FIG. 10 is vertically polarized light, which in FIG. 8 is parallel to the plane of the paper. Since the first input pixel a was in state Since it was in state B, the polarized light changed and became vertically polarized light (parallel to the plane of the paper in FIG. 8) and moved to the right by the birefringent plate 8-6. This corresponds to logical operation AB in FIG.

The upper right corner of FIG. 10 is horizontally polarized light, which in FIG. 8 is horizontal to the plane of the paper. Since the first input pixel a was in state A, the polarized light (output of 8-2) perpendicular to the plane of the paper changes and becomes vertically polarized light, which moves upward at the birefringent plate 8-4. , since the second input pixel was in the state (3), the polarization did not change and became a perpendicular (horizontal to the plane of the paper in FIG. 8) polarization and moved to the right by the birefringent plate 8-6. This corresponds to logical operation AB in FIG.

FIG. 11 shows the kernel 8-7 in FIG. 8, which is used for each pixel in order to selectively extract the four spatially separated lights of the birefringent plate 8-6 shown in FIG. It has a matrix shape.

This can be constructed in the same manner as shown in FIG. 5, and can also be realized by arranging four polarization rotation element pieces as in FIG. 6.

The logical operation shown in FIG. 12 also appears in operations between two side images, but the correspondence between the spatial arrangement and the logic is not easy to understand. The optical element 14 shown in FIG. 14 is next to the birefringent plate 8-4.8-6 in FIG. 8 or the birefringent plate 9-5.9-8 in FIG. 9.
-1. Inserting 14-2 makes it easier to understand the correspondence between spatial arrangement and logic. FIG. 13 shows the correspondence.

(1) in Figure 14 is a λ/4 plate (λ is the wavelength of the incident light) 1
4-1 and (linear) polarization selection element 14-2,
W Itψ has been done. The λ/4 plate 14-1 rotates the incident light (which is either horizontally polarized or vertically polarized in FIG. 10) clockwise or counterclockwise by the plate 14-1. becomes circularly polarized light. Next, the polarization selection element 14-2 converts the light into linearly polarized light (perpendicular to the plane of the paper in the figure), and the light is spatially separated and the polarized light becomes the same output light. 1st
(2) in Figure 4 shows the (linear) polarization selection element 14-3 and λ/
It is composed of two plates (λ is the wavelength of the incident plate) 14-4. The light incident on the polarization selection element 14-3 (which is one of the two types of polarization, horizontal polarization or vertical polarization in FIG. 10) is polarized by 45 degrees from the vertical direction ( It becomes linearly polarized light at an angle of 45° from the horizontal direction. In this way, the light is spatially separated and the polarized light becomes the same output light.

To set the direction in any direction as in (1), the λ/2 plate 14-
4 may be used to rotate the polarization. In addition, the λ/2 plate 14-4
It doesn't have to be there.

FIG. 15 (1) shows the correspondence between kernels and logical operations when the elements shown in FIG. 14 are used. FIG. 15(2) shows the truth table of the logical operation.

FIG. 16 shows an example in which processing is performed between transmission-type normal images by expanding on FIG. modulator 16-3',
Birefringent plate 16-4, second input signal 16-5, second spatial modulator 16-5', birefringent device 16-6, third input signal 16-7, third spatial modulator 16 -7', birefringent plate 16-8. It is composed of a kernel 16-9. In FIG. 16, the birefringent plates 8 to 16-6 in FIG.
It is the same as up to -6. The emitted light from the birefringent plate 16-6 enters the third spatial modulator 16-7', and the third input signal 1
Depending on the binary state of 6-7 (C, C) polarization is 90
'Rotate or no rotation. Next, the light is spatially separated by a birefringent plate 16-8 according to the polarization state. In the figure, four rays are separated into eight rays on the front and back sides.

A predetermined one of the separated light beams is selected by a kernel 16-9 set according to the calculation, and a logical calculation is executed.

FIG. 17 shows an embodiment in which processing is carried out between reflective normal images by developing FIG. 9, and includes light sources 17-1. Polarizer 17-2, half mirror 17-3, first input signal 17-4, first spatial modulator 17-4', birefringent plate 17-5, half mirror 17-6, second input signal 17-7, second spatial modulator 17-7', birefringent plate 17-8, half mirror 17-
9, third input signal 17-10, birefringent plate 17-11,
Kernel 17-12, (17-11 and 17-12 are the first
(16-8 and 16-9 in FIG. 6 are omitted because they are completely the same). In FIG. 17, the parts up to birefringent plate 17-8 are the same as those up to birefringent plate 16-8 in FIG. The output of the birefringent plate 17-8 is a half mirror 17-
9 and enters the third spatial modulator 17-10',
Depending on the binary state of the third input signal 17-10, the polarized light is reflected without rotation or with rotation by 90 degrees. This light is reflected by the half mirror 14-9 and enters the birefringent plate 14-11. The following is the same as in FIG. 16, and logical operations can be executed.

FIG. 18 shows the relationship between polarization and spatial arrangement when the output light of the birefringent plate 16-8 of FIG. 16 is viewed from the front of the light beam. The state of the first input is A, A, the state of the second input is B, B, the state of the third input is C9, and the first pixel a, the second pixel b, and the third pixel C It's Toshide. For example, at the bottom left, horizontally polarized light comes when a=τ, b=B, and c=c (corresponding to A-B-C-).

Fig. 19 shows the optical element shown in Fig. 14 and the birefringent plate 1 shown in Fig. 16.
This shows the relationship between the spatial arrangement and logic when placed after 5-4.15-6.15-8. FIG. 20 shows an example of the correspondence between logical operations and kernels that execute them.

FIG. 21 shows an example of the configuration of a logic operation system according to the present invention. In 2\, as an example of the calculation formula, the calculation formula for calculation between neighboring pixels with a distance of 1 is A AND (A, ORA.

ORA, ORA, ) eAi is an input obtained by shifting input A. For A, move A one pixel upward, and for A3, move A one pixel upward.
move one pixel to the right, A5 moves A one pixel downward, A7
uses these four inputs with A shifted one pixel to the left. Both were reflective types.

Each kernel is the same as shown in FIG. 9 and outputs the logic A-A□ in response to the input A□.

Figure 22 shows the kernel settings. When a two-stage operation is performed and the signals are combined by a half mirror, an OR (logical sum) is performed and an output corresponding to FIG. 20 is obtained.

〔Effect of the invention〕

In the optical parallel processing according to the present invention, light is transmitted through a spatial modulator (SLM).
) and optical elements, and propagates to the kernel. Processing results can be obtained at the speed of light, and processing speed is extremely fast even when processing in multiple stages. In addition, since parallel image information consisting of a large number of pixels can be given at the same time as the light beam, highly parallel processing can be performed. Comparing the processing power of electrical processing and optical parallel processing, optical processing is more advantageous when processing large amounts of highly parallel information (particularly image-related information). In electricity, even highly parallel information must be converted into serial data and processed as extremely high-speed data. This has become a bottleneck for increasing processing power.

Furthermore, the optical parallel processing according to the present invention includes parallel two-dimensional digital filtering operations, in addition to parallel inter-pixel logic operations.
It can also be used for image processing such as logical operations between parallel neighboring pixels. It has high utility value because it can perform all the processing expressed by logical expressions. By configuring addition and subtraction, it can also be used as a calculator. It can also be used to calculate differences between moving images, and when used for data compression, it can also be applied to the fields of image transmission and transmission processing.

[Brief explanation of the drawing]

1 and 2 are diagrams showing the simplest embodiment of the present invention, FIG. 3 is a diagram showing a conventional configuration example, and FIG. 4 is the output light beam of the birefringent plate in FIGS. 1 and 2. FIG. 5 and FIG. 6 are diagrams showing examples of the structure of kernels. FIG. 7 is a diagram showing logical correspondence with the kernels of the embodiments of FIGS. 1 and 2. FIG. 8 is a diagram showing an example of calculating between two side images, which is a development of the transmission type shown in FIG.
Figure 10 is a diagram showing the polarization and spatial positional relationship of the light rays in Figures 8 and 9, and Figure 11 is the same. 12 and 13 are diagrams showing the correspondence between the spatial arrangement of light rays and logic; FIG. 14 is a diagram showing an example of an optical element inserted after the birefringence plate; FIG. 15 is a diagram showing an example of the structure of the kernel; The figure shows the logical correspondence between the kernels in FIGS. 8 and 9 and the relationship between the logical formula and the truth table when the optical element shown in FIG. 14 is used, and FIG.
The figure shows an embodiment that performs calculations between three images, which is a further development of the transmission type shown in Fig. 8, and Fig. 17 shows an example that performs calculations between two side images, which is a further development of the reflection type shown in Fig. 9. Figure showing an example, 1st
Figure 8 is a diagram showing the polarization of light rays and spatial positional relationships in Figures 16 and 17, Figure 19 is a diagram showing spatial arrangement and logical correspondence, Figure 20 is a diagram showing logical formulas, examples, and kernels; 2
FIG. 1 is a diagram showing an example of the configuration of an arithmetic system according to the present invention, and FIG. 22 is a diagram showing a kernel used in FIG. 21. 1-1...Light source, 2...Polarizer, 1-3...
Input, 1-4... Birefringence plate, 1-5... Kernel, 2-1... Light source, 2-2... Polarizer, 2-2...
...Half mirror 12-4...Input, 2-5...Birefringent plate, 2-6...Kernel. Figure 1 N Disease Figure 2 Receiving layer Figure 3 (cL) (1>) / Figure 6 Figure 7 Figure 8 Figure 9 Figure 12 Figure 13 Figure 14 Brass brocade A B AOR13^AN
D 8%5i AXORB AANDB
AANDB 0th Figure 15 (2) m91 A bar 118 Mourning PL AOFIII Rei 5 @ AANDB Transport A IQY B Haruri Shame ■ Arori Ao＊E
Kyopa 1 electric kiki T Kikan T
Security A~OR8 Hanri^NDB@5% AXORB
# Sharp AANDT, $31τAND 9
810 (ku, b, c) = Yukoteihiyuki 1. Long 4. α葡: Ku=A. Toni 1-b-year-old B, b=o: b-a. Cs2: c2C, c20: C-te.し=Ob=+ b=o b−I C・Qic−+ 榛51! ! (A EQY B) ORCA
B^NDC) Figure 21 Ward Shooting Heha 7no! 1; ^^NDA, iUshi8N[
)3

Claims (4)

[Claims] (1) Replace the binary states of digital information with two orthogonal polarized lights, separate the replaced lights so that they do not spatially overlap according to the polarization, and select the separated lights. An optical parallel processing method characterized by: (2) a light source, a polarizer that polarizes the light from the light source, a spatial modulator that replaces the polarized light of the polarizer with two orthogonal polarized lights corresponding to the binary state of digital information; The present invention includes an optical element that separates the output light of the spatial modulator according to two polarized lights so that there is no spatial overlap, and a polarization selection element that selects the optical output of the optical element in accordance with the processing. Features of optical parallel processing device. (3) The above-mentioned spatial modulator and optical element pair are arranged in N stages (N is an integer of 2 or more) in series, and are arranged at 2^m positions (m = 1, 2, ...N-1)
The respective optical outputs are input to the next spatial modulator and replaced with two orthogonal polarized lights, and each output light of the spatial modulator is divided into two polarized lights according to the polarization so that there is no spatial overlap. ^m
Claim (2) characterized in that the device is separated into ＾+＾1 devices.
The optical parallel processing device described above. (4) The optical parallel processing device according to claims (2) and (3), characterized in that a λ/4 plate (λ is the wavelength of incident light) and a polarizer are installed after the optical element.