JPH03287238A - Optical arithmetic unit - Google Patents

Abstract

PURPOSE:To eliminate the need for encoding of logical level by processing numerals which are represented as the amplitude and phase of a light wave optically by spatial inverse Fourier transformation and performing addition and subtraction, and reconverting the result into a spatial frequency by spatial Fourier transformation. CONSTITUTION:The light wave 9 is split by a half-mirror 11 into two directions, and one wave passes through an ND filter 12 and a beam expander 13 to has its diameter increased, and then passes through a phase distribution modulator 14. The phase distribution modulator 14 modulates the input light wave into 1, 0, and -1 and a numeral which is represented in binary or ternary notation is obtained by the modulation. The other split light wave is modulated through a similar process. Those light waves are put one over the other by a half-mirror 15 and processed through the spatial inverse Fourier transformation of a Fourier transforming lens 16, then the addition and subtraction are represented by superposing gratings which have some spatial frequencies, and the result is processed through the spatial Fourier transformation of a Fourier transforming lens 17. Consequently, the need for arithmetic of logical representation level is eliminated.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a high-speed arithmetic device using light, such as a control device.

[Conventional technology]

An example of the prior art will be explained with reference to FIG. Binary input image A
1 and B2 are composed of mXn pixels, and by the method shown in Figure 3 (, ), the codes corresponding to the binary levels O and 1 are expressed as light and dark, whether light passes through or not. Let it be an encoded input image 3 expressed as (b). This image is placed on the input surface 4 of the projection optical system, and the L
Illumination by ED6. The arrangement of the system is adjusted so that the projected images of the encoded input image 3 by the four LEDs 6 are projected on the screen 7 so that they are shifted by half a pixel from each other vertically and horizontally. This projected image is observed through a decoding mask 8 having square windows arranged in a square grid. As a result, the calculation results of the two-variable binary logic function are obtained in parallel as light brightness signals for all pixels constituting the image. 4 LEDs
By changing the combinations of the 6 blinking states as shown in FIG. 4, 16 types of logical operations become possible. This method can perform parallel logical operations due to optical parallelism.

For details on this, please refer to “Hikari Computer” 10 published by Ohmsha.
It is described on pages 2 to 111.

[Problem to be solved by the invention]

The problems with the prior art are as follows.

a. In order to perform negative expression for numerical operations using this method, encoding at the logic level is required.

Therefore, for example, complement representation is required.

60 An error occurs if only a part of the image is missing.

[Means to solve the problem]

The present invention is a method for expressing numerical values in binary or ternary values and assigning one spatial frequency to each digit of the numerical value expressed in that method, and a method for expressing this in the state of light waves. A means for adjusting the amplitude and phase of a light wave, a means for superimposing multiple light waves, a means for optically performing spatial inverse Fourier transform to convert the spatial frequency into a cosine or sine wave amplitude grating, and a means for spatially transforming the gratings into a cosine or sine wave amplitude grating. and a means for returning the frequency to the spatial frequency.

[Effect]

Numerical values are expressed in binary or ternary representation, and one frequency on the spatial frequency is assigned to each digit. For example, each digit is expressed in such a way that if a digit is 1, the assigned frequency exists, and if it is 0, it does not exist. Furthermore, 3
In the value representation, if there is a -1 digit, the assigned frequency exists, and the phase of the light wave is adjusted so that it is delayed compared to when it is 1. What performs numerical expression using this light wave is a means for expressing the amplitude and phase of the light wave.

In this way, when a numerical value expressed by the amplitude and phase of a light wave is optically converted by means of spatial inverse Fourier transform, each digit becomes a bright/dark pattern of a sine or cosine wave amplitude lattice. In this state, when the two numerical values expressed as described above are superimposed, the amplitude grids representing each digit strengthen or weaken each other, thereby performing addition or subtraction. Furthermore, when the result is converted back to a spatial frequency by means of spatial Fourier transform, the result returns to the original numerical expression that reflects the results of addition and subtraction.

〔Example〕

An embodiment of the present invention will be described based on FIGS. 1 and 5.

A light wave 9 emitted from a light source 10, which preferably provides coherent light, is split by a half mirror 11 into two directions. One side passes through the ND filter 12 and the beam expander 1
3, the diameter of the light wave is expanded, and the light wave passes through a phase distribution modulator 14. The phase distribution modulator 14 is composed of a phase delay material 30, an amplitude control material 31, and a pinhole 32, as shown in the principle diagram shown in FIG. 0. -1. This modulation gives a numerical value expressed in binary or ternary values. This is done for each digit of the given number.

The phase retardation material 30 can be realized by an electro-optic material or the like,
The amplitude control material 31 can be realized by a filter or the like.

! In addition, the light wave split into the other side undergoes a similar process and is modulated. When these light waves are superimposed by a half mirror 15 and subjected to spatial inverse Fourier transform by a Fourier transform lens 16, each digit of the numerical value becomes an amplitude lattice as illustrated on the right side of the lower half of FIG. These are, when the original digits are 1 and -1 respectively,
Since the phases of the gratings differ by 180°, they cancel each other out, and their spatial frequency components disappear. Also, the original digits are O and 1, 0 and -1, 1+1. In the case of a combination of -1+-1, each spatial frequency component is conserved for one session. 5- Therefore, addition and subtraction are realized by superimposing lattices with a certain spatial frequency.

Furthermore, this result is subjected to spatial Fourier transformation using the Fourier transformation lens 17, and when the DC component appearing at the center of the light wave and the other spatial frequency component are superimposed on the light receiving element play 25, they are converted into binary or ternary values. A numerical value is obtained, and this is the result of addition or subtraction.

Among them, according to the phase distribution control means of the embodiment described above, information on 6 pits can be obtained. In other words, one pit can be represented by a pair of materials located at target positions sandwiching the optical axis.

[Effect of the invention] According to the present invention, 0, 1. Since the representation of -1 can be realized optically, calculations at the level of logical representation are unnecessary. Furthermore, since numerical values are expressed as a collection of amplitude grids, even if part of the grid is missing, the original numerical values can be reproduced.

[Brief explanation of drawings]

FIG. 1 is an overall diagram of a device according to an embodiment of the present invention, and FIG. 2 is a principle diagram of a conventional optical calculation device.
3. 1 and 4 are diagrams showing the input and output logic expressions of a conventional optical arithmetic device, and FIG. 5 shows the relationship between the numerical expression by the phase distribution modulator of the present invention and the spatial inverse Fourier transform. FIG.

Claims (1)

[Claims] (1) It has a light source and a plurality of phase distribution control means for controlling the phase distribution of light waves spread over a plane from the light source, and a spatial cosine or sine wave-like grating is provided by each of the plurality of phase distribution control means. The light is characterized in that it has a means for assigning a code to the image according to a spatial frequency, a means for synthesizing the light waves emitted from the phase distribution control means and spatial (inverse) Fourier transform, and a means for performing spatial Fourier transform again. Computing device.