Optical computing device and optical computing method

Abstract

The present invention achieves an optical computation device that makes it possible to perform high-speed optical computation while suppressing information loss from binarization. An optical computation device (1) comprises: a spatial light modulator (11) that generates signal light (L1) that represents a wavenumber space image (I1) by binary modulation of carrier light (L0) cell by cell; and a light modulation element group (13) that comprises one or more light modulation elements (13a1–13an) that act in sequence on the signal light (L1).

Description

[Technical Field]

[0001] The present invention relates to an optical arithmetic device including a group of optical modulation elements, and an optical arithmetic method using a group of optical modulation elements. [Background technology]

[0002] Optical modulation elements are known that have multiple cells and are designed to optically perform predetermined calculations by causing signal light transmitted through each cell to interfere with each other. Optical calculations using such optical modulation elements have the advantage of being faster and less power-consuming than electrical calculations using a processor. Furthermore, by sequentially operating multiple optical modulation elements on signal light, advanced optical calculations that are difficult to achieve with a single optical modulation element can be easily achieved.

[0003] Patent Document 1 discloses an optical neural network having an input layer, a hidden layer, and an output layer. The above-described optical modulation element can be used, for example, as the hidden layer of such an optical neural network. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] U.S. Patent No. 7,847,225 Summary of the Invention [Problem to be solved by the invention]

[0005] In an optical computing device, the signal light input to the optical modulation element group is usually generated by a spatial light modulator that modulates (intensity modulates or phase modulates) carrier light for each cell. For example, LCOS (Liquid Crystal On Silicon) is used as the spatial light modulator. LCOS modulates carrier light for each cell with multiple values ​​(usually 8 bits), so it can generate signal light with a large amount of information. However, the modulation speed of LCOS is about 60 Hz. Therefore, the modulation speed of LCOS becomes a bottleneck, and the high speed of optical computing by the optical modulation elements cannot be fully utilized.

[0006] Therefore, it has been considered to generate signal light to be input to a group of light modulation elements using a spatial light modulator that performs binary (1-bit) modulation on a carrier light for each cell. This is because a spatial light modulator that performs binary modulation on a carrier light for each cell can operate faster than an LCOS. Examples of such spatial light modulators include a DMD (Digital Mirror Device) and an FLCOS (Ferroelectric Liquid Crystal on Silicon). However, if an image to be supplied to the spatial light modulator is generated by binarizing an input image, a significant loss of information may occur.

[0007] One aspect of the present invention has been made in consideration of the above-mentioned problems, and its purpose is to realize an optical calculation device and an optical calculation method that are capable of performing high-speed optical calculations while suppressing loss of information due to binarization. [Means for solving the problem]

[0008] A wavenumber space image is generated by performing an inverse Fourier transform on an input image, a binary image is generated by binarizing the wavenumber space image, and a real space image is generated by performing a Fourier transform on the binary image. This can sometimes produce a real space image that more closely approximates the input image than a binary image obtained by binarizing the input image itself. This means that there are cases in which binarizing a wavenumber space image obtained by performing a Fourier transform on the input image results in less missing information than binarizing the input image itself. The present inventors focused on this point and came up with the present invention.

[0009] An optical computing device according to one aspect of the present invention includes a spatial light modulator that generates signal light representing a wavenumber spatial image by binary-modulating carrier light for each cell, and a group of optical modulation elements that includes at least one optical modulation element that sequentially acts on the signal light.

[0010] An optical computation method according to another aspect of the present invention includes the steps of generating signal light representing a wavenumber space image by binary modulating carrier light for each cell, and performing optical computation by sequentially applying at least one optical modulation element to the signal light. [Effects of the Invention]

[0011] According to one aspect of the present invention, high-speed optical calculation can be performed while suppressing loss of information due to binarization. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a side view showing the configuration of an optical arithmetic device according to an embodiment of the present invention; [Figure 2] 2 is a plan view showing a specific example of an optical modulation element included in the optical arithmetic unit shown in FIG. 1. FIG. [Figure 3] FIG. 3 is an enlarged perspective view of a part of the light modulation element shown in FIG. [Figure 4] 1. FIG. 4 is a side view showing a modified example of the optical arithmetic device shown in FIG. [Figure 5]5 is a plan view showing a specific example of an optical modulation element included in the optical arithmetic unit shown in FIG. 4. FIG. [Figure 6] FIG. 6 is a cross-sectional view of a cell that constitutes the light modulation element shown in FIG. DETAILED DESCRIPTION OF THE INVENTION

[0013] (Configuration of optical computing device) An optical arithmetic device 1 according to one embodiment of the present invention will be described with reference to Fig. 1. Fig. 1 is a side view showing the configuration of the optical arithmetic device 1.

[0014] As shown in FIG. 1, the optical calculation device 1 includes a spatial light modulator 11, a lens 12, a light modulation element group 13, and a driver .

[0015] The spatial light modulator 11 is configured to obtain signal light L1 representing a wavenumber spatial image I1 by binary-modulating the carrier light L0 for each cell. In this embodiment, a spatial light modulator that performs binary intensity modulation or binary phase modulation on the carrier light L0 for each cell is used as the spatial light modulator 11. An example of a spatial light modulator that performs binary intensity modulation on the carrier light L0 for each cell is a DMD (Digital Mirror Device). Also, an example of a spatial light modulator that performs binary phase modulation on the carrier light L0 for each cell is FLCOS (Ferroelectric Liquid Crystal On Silicon).

[0016] The lens 12 is configured to obtain signal light L2 representing a real space image I2 by Fourier transforming signal light L1 representing a wavenumber space image I1. In this embodiment, a lens having a positive focal length f is used as the lens 12. The lens 12 is positioned so that the principal plane of the lens 12 is parallel to the exit surface of the spatial light modulator 11 and so that the distance between the principal plane of the lens 12 and the exit surface of the spatial light modulator 11 coincides with the focal length f of the lens 12. Examples of lenses having a positive focal length f include convex lenses and Fresnel lenses. Alternatively, a transmissive light modulation element designed to function as a convex lens may be used as the lens 12.

[0017] The light modulation element group 13 is a collection of n light modulation elements 13a1 to 13an that sequentially act on the signal light L2. Here, n is any natural number equal to or greater than 1. FIG. 1 illustrates the case where n=4. Each light modulation element 13ai is an element having an optical calculation function, i.e., a function of converting the two-dimensional intensity distribution or two-dimensional phase distribution of the signal light in accordance with a predetermined conversion rule. Here, i is a natural number equal to or greater than 1 and equal to or less than n. The light modulation element group 13 is arranged so that the incident surface of each light modulation element 13ai is parallel to the principal plane of the lens 12 and so that the distance between the incident surface of the light modulation element 13a1 closest to the lens 12 and the principal plane of the lens 12 coincides with the focal length f of the lens 12. In this embodiment, a transmissive light modulation element is used as each light modulation element 13ai. Specific examples of the light modulation elements 13ai will be described later.

[0018] Signal light L2 representing a real-space image I2 is input to the light modulation element group 13. The signal light L2 passes through the first light modulation element 13a1, the second light modulation element 13a2, ..., and the n-th light modulation element 13an, in this order. As a result, in the light modulation element group 13, an operation f=fn...f2f1 is performed on the real-space image I2 represented by the signal light L2, combining an operation f1 by the first light modulation element 13a1, an operation f2 by the second light modulation element 13a2, ..., and an operation fn by the n-th light modulation element 13an. Therefore, the light modulation element group 13 obtains signal light L3 representing an image I3=f(I2), which is obtained by applying the operation f to the real-space image I2.

[0019] The driver 14 is configured to drive the spatial light modulator 11. The driver 14 has an inverse Fourier transform unit 14a and a binarization unit 14b. The inverse Fourier transform unit 14a performs an inverse Fourier transform on the input image I0 to obtain a wavenumber space image I1'. The number of gradations of the wavenumber space image I1' is the same as the number of gradations of the input image I0, e.g., 256. The binarization unit 14b binarizes the wavenumber space image I1' to obtain a binary wavenumber space image I1. The driver 14 drives the spatial light modulator 11 to generate signal light L1 representing the binary wavenumber space image I1.

[0020] Note that the lens 12 can be omitted. In this case, the signal light L1 representing the wavenumber space image I1 is input to the light modulation element group 13. In this case, the light modulation element group 13 performs an operation f=fn·fn-1·..., f2·f1 on the wavenumber space image I1 represented by the signal light L1, which is a combination of an operation f1 by the first light modulation element 13a1, an operation f2 by the second light modulation element 13a2, ..., an operation fn-1 by the (n-1)th light modulation element 13an-1, and an operation fn by the nth light modulation element 13an. Therefore, the light modulation element group 13 obtains a signal light L3 representing an image I3=f(I1) obtained by applying the operation f to the wavenumber space image I1.

[0021] In addition to the above-described configuration, the optical processing device 1 may also include a light source that generates carrier light L0. In addition to the above-described configuration, the optical processing device 1 may also include a two-dimensional image sensor that converts signal light L3 representing the image I3 into an electrical signal representing the image I3.

[0022] Furthermore, the distance between the exit surface of the spatial light modulator 11 and the principal plane of the lens 12 does not need to strictly match the focal length f of the lens 12. Similarly, the distance between the principal plane of the lens 12 and the entrance surface of the light modulation element 13a1 does not need to strictly match the focal length f of the lens 12. Even if these distances do not strictly match the focal length f of the lens 12, the signal light L2 incident on the light modulation element group 13 will contain a component representing the real space image I2. Therefore, it is possible to perform optical calculations equivalent to those when these distances strictly match the focal length f of the lens 12.

[0023] The light modulation elements 13a1-13an may be integrated. For example, n light diffraction layers formed in a structure that transmits signal light, such as a dry gel, may be used as the light modulation elements 13a1-13an. In this case, it is preferable to use a gel that shrinks while maintaining a similar shape when dehydrated and shrunk, such as a gel used in implosion fabrication. This makes it possible to easily manufacture the light modulation element group 13 in which the light modulation elements 13a1-13an are precisely arranged by drying the swollen gel on which the n light diffraction layers are formed.

[0024] (Example of a transmissive light modulation element) A specific example of the transmissive light modulation element 13ai will be described with reference to Fig. 2 and Fig. 3. Fig. 2 is a plan view of the light modulation element 13ai according to this example. Fig. 3 is an enlarged perspective view of a part of the light modulation element 13ai according to this example (the part surrounded by a dotted line in Fig. 2).

[0025] As shown in Fig. 2, the optical modulation element 13ai is composed of a plurality of microcells C, each having a phase modulation amount set independently of the other. When signal light is incident on the optical modulation element 13ai, the signal light phase-modulated by each microcell C interferes with each other, thereby performing a predetermined optical operation (conversion of a two-dimensional intensity distribution according to a predetermined conversion rule). The phase modulation amount of each microcell C may be variable or fixed, but is fixed in this specific example.

[0026] In this specification, the term "microcell" refers to a cell having a cell size of less than 10 μm, for example. The term "cell size" refers to the square root of the area of ​​the cell. For example, if the shape of the microcell C in plan view is a square, the cell size of the microcell C is the length of one side of the microcell C. The lower limit of the cell size of the microcell C is, for example, 1 nm.

[0027] 2 is composed of 200 × 200 microcells C arranged in a matrix. The planar shape of each microcell C is a 500 nm × 500 nm square, and the planar shape of the light modulation element 13ai is a 100 μm × 100 μm square.

[0028] (1) By independently setting the thickness of the microcell C for each cell, or (2) independently selecting the refractive index of the microcell C for each cell, the phase modulation amount of the signal light passing through each microcell C can be independently set for each cell. In this embodiment, method (1), which can be realized by nanoimprinting, is adopted. In this case, as shown in FIG. 3, each microcell C is formed by a rectangular pillar having a square base with each side length equal to the cell size. In this case, the phase modulation amount of the signal light passing through each microcell C is determined by the height of the pillar constituting the microcell C. In other words, the phase modulation amount of the signal light passing through a microcell C formed by a tall pillar is large, and the phase modulation amount of the signal light passing through a microcell C formed by a short pillar is small.

[0029] The setting of the phase modulation amount of each microcell C can be realized by, for example, machine learning. A model used in this machine learning can be, for example, a model that takes the two-dimensional intensity distribution of the signal light input to the optical modulation element 13ai as input and the two-dimensional intensity distribution of the signal light output from the optical modulation element 13ai as output, and that includes the phase modulation amount of each microcell C as a parameter. Here, the two-dimensional intensity distribution of the signal light input to the optical modulation element 13ai refers to a set of intensities of the signal light input to each microcell C constituting the optical modulation element 13ai. Furthermore, the two-dimensional intensity distribution of the signal light output from the optical modulation element 13ai refers to a set of intensities of the signal light input to each microcell C constituting the optical modulation element 13ai+1 arranged downstream of the optical modulation element 13ai, or a set of intensities of the signal light input to each cell constituting the two-dimensional image sensor arranged downstream of the optical modulation element 13ai.

[0030] (Modification of Optical Computing Device) In the optical arithmetic device 1, the light modulation element group 13 is composed of transmissive light modulation elements 13a1 to 13an, but the present invention is not limited to this. That is, the light modulation element group 13 may be composed of reflective light modulation elements 13b1 to 13bn.

[0031] 4 is a side view showing the configuration of an optical arithmetic device 1A in which the optical modulation element group 13 is composed of reflective optical modulation elements 13b1 to 13bn. In the optical arithmetic device 1A, signal light L2 is (1) modulated and reflected by optical modulation element 13b1, (2) modulated and reflected by optical modulation element 13b2, (3) modulated and reflected by optical modulation element 13b3, and (4) modulated and reflected by optical modulation element 13b4 to obtain signal light L3.

[0032] In the optical processing device 1A, the light modulation element 13b1 is disposed so that its center is located at the focal point of the lens 12. Therefore, in the optical processing device 1A, the intensity of the signal light L2 incident on the center of the light modulation element 13b1 is the same as the intensity of the signal light L2 incident on the center of the light modulation element 13a1 in the optical processing device 1. On the other hand, in the optical processing device 1A, the principal surface of the light modulation element 13b1 is not perpendicular to the optical axis of the lens 12. Therefore, in the optical processing device 1A, the intensity of the signal light L2 incident on a point other than the center of the light modulation element 13b1 is not the same as the intensity of the signal light L2 incident on the corresponding point of the light modulation element 13a in the optical processing device 1. Therefore, the intensity distribution of the signal light L2 incident on the light modulation element 13b1 does not represent the exact real space image I2, but rather represents an approximate real space image I2'. For this reason, the phase modulation amount of each cell constituting the light modulation elements 13b1 to 13bn is designed so as to perform a desired optical calculation when an approximate real space image I2' is incident on the light modulation element 13b1.

[0033] The light modulation elements 13b1 and 13b3 arranged on the same plane may be integrated. For example, the light modulation elements 13b1 and 13b3 may be embedded in a single substrate, or two regions of a single light modulation element may be used as the light modulation elements 13b1 and 13b3. The same applies to the light modulation elements 13b2 and 13b4 arranged on the same plane. Furthermore, four light diffraction layers formed in a structure that transmits signal light, such as a dry gel, may be used as the light modulation elements 13b1 to 13b4. In this case, it is preferable to use a gel that shrinks while maintaining a similar shape through dehydration shrinkage, such as a gel used in implosion fabrication. This allows for the easy manufacture of a light modulation element group 13 in which the light modulation elements 13b1 to 13b4 are precisely arranged by drying the swollen gel on which the four light diffraction layers are formed.

[0034] (Example of a reflective light modulation element) A specific example of the reflective light modulation element 13bi will be described with reference to Fig. 5 and Fig. 6. Fig. 5 is a plan view of the light modulation element 13bi according to this example. Fig. 6 is a cross-sectional view of a microcell C that constitutes the light modulation element 13bi according to this example.

[0035] 5, the optical modulation element 13bi is composed of a plurality of microcells C, each of which can set the amount of phase modulation independently of the other. When signal light is incident on the optical modulation element 13bi, the signal light phase-modulated by each microcell C interferes with each other, thereby performing a predetermined optical operation (conversion of a two-dimensional intensity distribution according to a predetermined conversion rule). The amount of phase modulation of each microcell C may be variable or fixed, but in this specific example, it is variable.

[0036] 5 is composed of 200 × 200 microcells C arranged in a matrix. The planar shape of each microcell C is a 500 nm × 500 nm square, and the planar shape of the light modulation element 13bi is a 100 μm × 100 μm square.

[0037] Each microcell C constituting the light modulation element 13bi can be composed of, for example, a polarizing plate C11, a reflecting plate C12, a first electrode C13, a magnetization free layer C14, an insulating layer C15, a magnetization fixed layer C16, and a second electrode C17, as shown in Figure 6.

[0038] The polarizer C11 and the reflector C12 are disposed to face each other. The first electrode C13, the magnetization free layer C14, the insulating layer C15, the magnetization fixed layer C16, and the second electrode C17 are stacked in this order and sandwiched between the polarizer C11 and the reflector C12. The stacking direction of the first electrode C13, the magnetization free layer C14, the insulating layer C15, the magnetization fixed layer C16, and the second electrode C17 is perpendicular to the stacking direction of the polarizer C11 and the reflector C12. Therefore, a first side surface of the magnetization free layer C14 is in surface contact with one major surface of the polarizer C11, and a second side surface of the magnetization free layer C14 facing the first side surface is in surface contact with one major surface of the reflector C12. The signal light L (1) enters the magnetization free layer C14 via the polarizer C11, (2) is reflected by the reflector C12, and (3) exits the magnetization free layer C14 via the polarizer C11.

[0039] The magnetization free layer C14 is made of, for example, a soft magnetic material (e.g., CoFeB) that is conductive and transparent. The magnetization fixed layer C16 is made of, for example, a hard magnetic material (e.g., Permalloy) that is conductive. The polarizer C11 is selected to selectively transmit a polarized light component whose polarization direction P is parallel to the magnetization direction M of the magnetization fixed layer C16. Figure 6 illustrates a case where the magnetization direction M and the polarization direction P are parallel to both the main surface of the polarizer C11 and the main surface of the magnetization fixed layer C16.

[0040] When a potential difference is applied between the first electrode C13 and the second electrode C17, a spin current (a flow of spin-polarized electrons) is injected from the magnetization fixed layer C16 through the insulating layer C15 into the magnetization free layer C14 due to the tunneling effect, and magnetization occurs in the magnetization free layer C14. Here, the magnetization generated in the magnetization free layer C14 is magnetization parallel to the magnetization direction M of the magnetization fixed layer C16, that is, magnetization parallel to the polarization direction P of the signal light incident on the magnetization free layer C14 via the polarizer C11. Therefore, the phase of the signal light is delayed by the transverse Kerr effect during propagation through the magnetization free layer C14.

[0041] Here, the amount of phase change of the signal light in cell C is determined according to the magnitude of magnetization generated in the magnetization free layer C14. Furthermore, the magnitude of magnetization generated in the magnetization free layer C14 is determined according to the magnitude of the spin current injected into the magnetization free layer C14. Furthermore, the magnitude of the spin current injected into the magnetization free layer C14 is determined according to the potential difference applied between the first electrode C13 and the second electrode C17. Therefore, the phase modulation of cell C can be controlled by controlling the potential difference applied between the first electrode C13 and the second electrode C17.

[0042] The setting of the phase modulation amount of each microcell C can be realized using, for example, machine learning. A model used in this machine learning can be, for example, a model that takes the two-dimensional intensity distribution of the signal light input to the optical modulation element 13bi as input and the two-dimensional intensity distribution of the signal light output from the optical modulation element 13bi as output, and that includes the phase modulation amount of each microcell C as a parameter. Here, the two-dimensional intensity distribution of the signal light input to the optical modulation element 13bi refers to a set of intensities of the signal light input to each microcell C constituting the optical modulation element 13bi. Furthermore, the two-dimensional intensity distribution of the signal light output from the optical modulation element 13bi refers to a set of intensities of the signal light input to each microcell C constituting the optical modulation element 13bi+1 arranged downstream of the optical modulation element 13bi, or a set of intensities of the signal light input to each cell constituting the two-dimensional image sensor arranged downstream of the optical modulation element 13bi.

[0043] (Summary 1) The optical computing device according to aspect 1 of the present invention comprises a spatial light modulator that generates signal light representing a wavenumber spatial image by binary-modulating carrier light for each cell, and a group of optical modulation elements that includes at least one optical modulation element that sequentially acts on the signal light.

[0044] According to the above configuration, a spatial light modulator that performs binary modulation on the carrier light for each cell is used, thereby enabling high-speed optical computation. Furthermore, according to the above configuration, the spatial light modulator generates signal light representing a wavenumber space image, so the image supplied to the spatial light modulator can be generated by binarizing the wavenumber space image. Therefore, compared to when the image supplied to the modulator is generated by binarizing a real space image, loss of information due to binarization can be reduced. Therefore, according to the above configuration, high-speed optical computation can be performed while reducing loss of information due to binarization.

[0045] In the optical computing device according to aspect 2 of the present invention, in addition to the configuration of aspect 1, a lens that performs a Fourier transform on the signal light is further provided, and the group of optical modulation elements acts on the signal light that represents a real-space image obtained by the lens.

[0046] According to the above configuration, high-speed optical calculations can be performed on signal light representing a real-space image while suppressing loss of information due to binarization.

[0047] In the optical computing device according to aspect 3 of the present invention, in addition to the configuration of aspect 2, the lens is a lens having a positive focal length, and the distance between the exit surface of the spatial light modulator and the principal plane of the lens, and the distance between the principal plane of the lens and the entrance surface of the light modulation element closest to the lens, each coincide with the focal length of the lens.

[0048] According to the above configuration, the Fourier transform of the signal light can be realized with a simple configuration.

[0049] An optical computing device according to a fourth aspect of the present invention employs, in addition to the configuration of any one of the first to third aspects, a configuration further comprising a driver for driving the spatial light modulator, the driver having an inverse Fourier transform unit for generating a wavenumber space image by performing an inverse Fourier transform on an input image, and a binarization unit for generating the wavenumber space image by binarizing the wavenumber space image.

[0050] According to the above configuration, it is possible to perform high-speed optical calculations on signal light representing a wavenumber space image (when a lens is not provided) obtained by inverse Fourier transforming an input image, or a real space image (when a lens is provided) obtained by Fourier transforming the wavenumber space image, while suppressing loss of information due to binarization.

[0051] In an optical computing device according to aspect 5 of the present invention, in addition to the configuration of any one of aspects 1 to 4, the spatial light modulator is a DMD (Digital Mirror Device) or FLCOS (Ferroelectric Liquid Crystal On Silicon).

[0052] According to the above configuration, binary modulation for each cell can be realized with a simple configuration.

[0053] In the optical arithmetic device according to aspect 6 of the present invention, in addition to the configuration of any one of aspects 1 to 5, a configuration is adopted in which each optical modulation element constituting the optical modulation element group is composed of a plurality of microcells whose phase modulation amounts can be set independently of each other.

[0054] According to the above configuration, optical computation for signal light can be realized with a simple configuration.

[0055] An optical arithmetic device according to a seventh aspect of the present invention employs the configuration of any one of the first to sixth aspects, in which each of the light modulation elements constituting the light modulation element group is a transmissive light modulation element.

[0056] According to the above configuration, signal light that more accurately represents a real space image can be input to the light modulation element group, compared to when the light modulation elements are configured using reflective light modulation elements.

[0057] An optical computation method according to aspect 8 of the present invention includes a step of generating signal light representing a wavenumber space image by binary modulating carrier light for each cell, and a step of performing optical computation by sequentially applying at least one optical modulation element to the signal light.

[0058] According to the above configuration, high-speed optical calculation can be performed while suppressing loss of information due to binarization.

[0059] (Summary 2) The optical computing device according to aspect 1 of the present invention comprises a spatial light modulator that generates signal light representing a wavenumber spatial image by binary-modulating carrier light for each cell, and a group of optical modulation elements that includes at least one optical modulation element that sequentially acts on the signal light.

[0060] According to the above configuration, a spatial light modulator that performs binary modulation on the carrier light for each cell is used, thereby enabling high-speed optical computation. Furthermore, according to the above configuration, the spatial light modulator generates signal light representing a wavenumber space image, so the image supplied to the spatial light modulator can be generated by binarizing the wavenumber space image. Therefore, compared to when the image supplied to the modulator is generated by binarizing a real space image, loss of information due to binarization can be reduced. Therefore, according to the above configuration, high-speed optical computation can be performed while reducing loss of information due to binarization.

[0061] In the optical computing device according to aspect 2 of the present invention, in addition to the configuration of aspect 1, a lens that performs a Fourier transform on the signal light is further provided, and the group of optical modulation elements acts on the signal light that represents a real-space image obtained by the lens.

[0062] According to the above configuration, high-speed optical calculations can be performed on signal light representing a real-space image while suppressing loss of information due to binarization.

[0063] In the optical computing device according to aspect 3 of the present invention, in addition to the configuration of aspect 2, the lens is a lens having a positive focal length, and the distance between the exit surface of the spatial light modulator and the principal plane of the lens, and the distance between the principal plane of the lens and the entrance surface of the light modulation element closest to the lens, each coincide with the focal length of the lens.

[0064] According to the above configuration, the Fourier transform of the signal light can be realized with a simple configuration.

[0065] An optical computing device according to a fourth aspect of the present invention employs, in addition to the configuration of any one of the first to third aspects, a configuration further comprising a driver for driving the spatial light modulator, the driver having an inverse Fourier transform unit for generating a wavenumber space image by performing an inverse Fourier transform on an input image, and a binarization unit for generating the wavenumber space image by binarizing the wavenumber space image.

[0066] According to the above configuration, it is possible to perform high-speed optical calculations on signal light representing a wavenumber space image (when a lens is not provided) obtained by inverse Fourier transforming an input image, or a real space image (when a lens is provided) obtained by Fourier transforming the wavenumber space image, while suppressing loss of information due to binarization.

[0067] In an optical computing device according to aspect 5 of the present invention, in addition to the configuration of any one of aspects 1 to 4, the spatial light modulator is a DMD (Digital Mirror Device) or FLCOS (Ferroelectric Liquid Crystal On Silicon).

[0068] According to the above configuration, binary modulation for each cell can be realized with a simple configuration.

[0069] In the optical arithmetic device according to aspect 6 of the present invention, in addition to the configuration of any one of aspects 1 to 5, each of the optical modulation elements constituting the optical modulation element group is a reflective optical modulation element and is composed of a plurality of microcells whose phase modulation amounts can be set independently of each other.

[0070] According to the above configuration, optical computation for signal light can be realized with a simple configuration.

[0071] In the optical arithmetic device according to aspect 7 of the present invention, in addition to the configuration of any one of aspects 1 to 5, each of the optical modulation elements constituting the optical modulation element group is a transmissive optical modulation element and is composed of a plurality of microcells whose phase modulation amounts are set independently of each other.

[0072] According to the above configuration, signal light that more accurately represents a real space image can be input to the light modulation element group, compared to when the light modulation elements are configured using reflective light modulation elements.

[0073] An optical computation method according to aspect 8 of the present invention includes a step of generating signal light representing a wavenumber space image by binary modulating carrier light for each cell, and a step of performing optical computation by sequentially applying at least one optical modulation element to the signal light.

[0074] According to the above configuration, high-speed optical calculation can be performed while suppressing loss of information due to binarization.

[0075] (Additional notes) The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in the above-described embodiments are also included in the technical scope of the present invention. [Explanation of symbols]

[0076] 1 Optical calculation device 11 Spatial Light Modulator 12 Lenses 13 Light modulation element group 13a1 to 13an Optical modulation element

Claims

1. a spatial light modulator that generates signal light representing a wavenumber spatial image by binary-modulating a carrier light for each cell; a lens that Fourier transforms the signal light into signal light representing a real space image; a light modulation element group consisting of at least one light modulation element that sequentially acts on signal light representing the real space image, An optical computing device characterized by:

2. a spatial light modulator that generates signal light representing a wavenumber spatial image by binary-modulating a carrier light for each cell; and an optical modulation element group including at least one optical modulation element that sequentially acts on the signal light and modulates the phase of the signal light while transmitting or reflecting the signal light. An optical computing device characterized by:

3. the lens is a lens having a positive focal length, a distance between the exit surface of the spatial light modulator and a principal plane of the lens, and a distance between the principal plane of the lens and an entrance surface of a light modulation element closest to the lens, each coincide with a focal length of the lens; 2. The optical computing device according to claim 1.

4. a driver for driving the spatial light modulator, the driver having an inverse Fourier transform unit for generating a wavenumber space image by performing an inverse Fourier transform on an input image, and a binarization unit for generating the wavenumber space image by binarizing the wavenumber space image; 4. The optical computing device according to claim 1, wherein the optical computing device is a multi-mode optical processor.

5. The spatial light modulator is a DMD (Digital Mirror Device) or a FLCOS (Ferroelectric Liquid Crystal On Silicon).

4. The optical computing device according to claim 1, wherein the optical computing device is a multi-mode optical processor.

6. Each of the light modulation elements constituting the light modulation element group is a reflective light modulation element and is composed of a plurality of microcells whose phase modulation amounts can be set independently of each other.

4. The optical computing device according to claim 1, wherein the optical computing device is a multi-mode optical processor.

7. Each of the light modulation elements constituting the light modulation element group is a transmissive light modulation element and is composed of a plurality of microcells in which the phase modulation amount is set independently of each other.

4. The optical computing device according to claim 1, wherein the optical computing device is a multi-mode optical processor.

8. generating signal light representing a wavenumber space image by binary modulating the carrier light for each cell; Fourier transforming the signal light into a signal light representing a real space image; and performing an optical operation by sequentially causing at least one light modulation element to act on signal light representing the real space image. An optical computing method comprising:

9. generating signal light representing a wavenumber space image by binary modulating the carrier light for each cell; and performing an optical operation by sequentially applying at least one optical modulation element to the signal light, In the step of performing the optical operation, the at least one optical modulation element modulates the phase of the signal light while transmitting or reflecting the signal light. An optical computing method comprising:

Images