JPS6146813B2 - - Google Patents

Description

Detailed Description of the Invention This invention has a good performance-to-price ratio, and uses an extremely simple optical system such as a projection optical system or a pinhole camera to perform operations on arbitrary two-variable binary logic functions in two-dimensional parallel fashion. The present invention relates to a parallel optical logic operation method that can be executed in parallel.

Currently, the realization of an optical computer that takes advantage of the high speed and parallelism of light is being awaited. However, in reality, a practical parallel optical logic operation method that can be used for this purpose has not yet been developed. Several optical logic calculation methods using special spatial modulation elements have been announced (e.g.
Applid Optics, Vol. 22, 1973, P. 2250), but these elements were very expensive and required complex optical systems to perform logical operations.
The present invention solves these drawbacks all at once, and
This paper provides an innovative parallel optical logic operation method that will bring the realization of optical computers one step closer to reality.

To explain the principle of the parallel optical logic operation method of the present invention according to FIG. 1, the encoded image line 7 of the input image A,
Let the encoded image 8 of the input image B be the superposition encoded input image 6. This is illuminated by point light sources 2, 3, 4, and 5 arranged in a square grid on the light source surface 1. At this time, images projected by the four point light sources are projected onto the screen 9 so as to be shifted from each other by half a pixel. When this projected image is viewed through a decoding mask 10 having windows arranged in a lattice pattern, the calculation results of an arbitrary two-variable binary logic function are obtained in parallel as light brightness signals for all pixels constituting the image. The type of logical function can be selected by a combination of the blinking states of the point light sources 2, 3, 4, and 5. The input signals of this logical operation method are two binary images (black and white images) A and B.
Assume that both of these pixels are composed of m×n pixels arranged in m columns vertically and n columns horizontally. Figure 2 is an example of this, where the input images A11 and B12 are both 4x4
This is a binary image composed of pixels. However, the white part represents "1" and the hatched part represents "0".

Logical operations by the parallel optical logical operation method of the present invention are performed completely in parallel for each pixel forming an image.
That is, in this method, it is assumed that independent logic gates exist for all pixels constituting a binary image, and that all of the logic gates operate simultaneously in response to one instruction and perform logic operations in parallel. Therefore,
The operation of logical operations can be considered by focusing on a particular pixel, and the same operation is performed in parallel on all other pixels.

Hereinafter, the logic gate of the (i, j)th pixel will be taken up and the encoding of the input image will be explained along with FIG. 2. The inputs of this logic gate are the (i, j)th image _aij of image A11 and the (i,j)th image aij of image B12.
The pixel b _ij . Since images A and B are both binary images, the two corresponding images a _ij and b _ij have a value of 0 (corresponding to black) or 1 (corresponding to white). As an example, if we pay attention to pixel (3, 3) in Figure 2,
The inputs are pixel a _ij 13 of image A11 and pixel b _ij 14 of image B12, where a _ij =0 and b _ij =1, respectively.
It is. These inputs a _ij and b _ij are subjected to special encoding corresponding to their values. This has two inputs a _i
It converts the brightness information of _j and b _ij into two-dimensional spatial position information, making it easy to utilize the characteristics of parallel processing using light, and is the core of this logical operation method.

Encoding is performed in two stages. The first step is to obtain encoded image 7 from image A11 and encoded image 8 from image B12 in FIG.
The second step is to superimpose the encoded input image 6.

In the first stage of encoding, the (i, j)th
As shown in FIG. 3, the pixel values are converted into codes in which half of the pixel sections are transparent and the other half are opaque.
Regarding the pixel a _ij of the image A, the code is changed so that only the upper half or the lower half of the (i, j)th pixel section transmits light depending on the value. This is shown in the row a _ij of FIG. 3, where the white areas represent transparency and the hatched areas represent opacity. In the case of a _ij 13, since it is 0, the upper half of the pixel block is encoded to be transparent, as shown in 15 in FIG. Pixel b of image B
The same goes for _ij , but like the row b _ij in Figure 3,
It is divided into right half and left half. Therefore, b _ij 14 in FIG. 2 is converted to code 16 whose right half is transparent. These encodings are performed for all pixels, and the results are recorded on a slide film or the like.

Next, the two encoded images 7 and 8 are superimposed so that the pixels at corresponding positions of the two images match. As a result, as shown in Fig. 4, a _ij , b _ij
Four types of codes can be created according to the combination of values. This is because only a quarter of the (i, j)th pixel section has a transparent window, and the position of this window within the pixel section depends on the combination of the values of a _ij and b _ij . Represents ivy information. For the (3,3)th pixel, a _ij =
0, b _ij =1, and by the first stage encoding, the second
It was converted into codes 15 and 16 in the figure, and when these are overlapped, code 17 having a transparent window in the upper right part of the pixel block is completed. As shown in FIG. 4, this corresponds to the case where a _ij =0 and b _ij =1.

The encoded input image 6 consists of point light sources 2, which are arranged in a square grid of four point light sources, as shown in FIG.
Illuminated at 3, 4, 5. Point light sources 2, 3, 4,
5 and 6 are diagrams showing the positional relationship between the encoded input image 6, the encoded input image 6, and the screen 9. FIG. 5 is a side view, and FIG. 6 is a perspective view.
However, in FIG. 6, a specific one of the encoded input image 6
Only one pixel section 18 is shown. Here, 1
The parts 9, 20, 21, and 22 are a _ij , respectively.
It shows possible window positions depending on the value of b _ij .

There are four point light sources, and if they emit light at the same time, the shadow of the (i, j)th pixel will be projected onto the screen 9 in four layers. The state of overlapping of these images can be determined by referring to FIG.
It is determined by the distance z ₁ between the light source surface 1 and the encoded input image 6, the distance z between the encoded input image 6 and the screen 9, and the size 2d of the pixel section on the encoded input image 6. In particular, when the following relationship is satisfied: z/z ₁ = d/L−d (however, d<L), pixel 18 on the encoded input image 6 on the screen 9 as shown in FIG.
Images 27, 28, caused by point light sources 2, 3, 4, 5,
29 and 30 are projected onto the screen 9 with a shift of just half a pixel in both the vertical and horizontal directions. As a result, when the light that passes through the four possible windows of the pixel sections on the encoded input image 6 reaches the screen 9, the possible sections that the light can reach are as shown in FIG. , there are nine. The second key point of this parallel optical logic operation method is to superimpose the projected images while maintaining this special positional relationship.

Of the nine sections on the screen 9, attention is paid to one section 31 in the center. Light is irradiated here in the following four cases. a _ij =1, b _ij =1
When the point light source 2 is on, a _ij = 1, b _ij =
0 and when the point light source 3 is on, a _ij =0, b _ij
=1 and when the point light source 4 is on, a _ij =0, b _i
When _j = 0, the point light source 5 is in the lighting state. These four cases are independent of each other. Therefore, by changing the combination of lighting states of the four point light sources 2, 3, 4, and 5, the values of a _ij and b _ij can be associated with the brightness of the central section 31 in a fixed relationship. .

FIG. 7 depicts the projected images seen on the screen 9 for four values of the input images a _ij and b _ij for the lighting states of various combinations of point light sources 2, 3, 4, and 5. It is. The four circles in the top row are the point light sources 2, 3, 4, and 5 lit, and the white circle is lit.
A black circle indicates that the light is off.

Furthermore, the three columns on the left represent the states of the input pixels a _ij and b _ij and their combined state. The numbers 0 and 1 below each projection image represent the brightness and darkness of the central section, with 0 being dark and 1 being bright. From FIG. 7, the brightness of the central section 31 is determined by the point light sources 2, 3, 4,
According to the lighting status of 5, all 16 types of 2 variables 2
It can be seen that a value logic function is realized. Therefore, as shown in FIG.
If 0 is placed on the screen 9, the output of the logic function will be obtained as information on the brightness of that window. The above logical operations are executed completely in parallel for all pixels making up the image.

Although the parallel optical logic operation method of the present invention was implemented using a projection optical system using multiple point light sources, an imaging system that produces an equivalent projected image on a screen surface, such as a pinhole camera having multiple pinholes, may also be used. It can also be implemented using

[Brief explanation of the drawing]

FIG. 1 is a conceptual diagram of the parallel optical logic operation method of the present invention. FIG. 2 is an explanatory diagram of encoding of an input image. FIG. 3 is a diagram showing the first stage encoding of an input image. FIG. 4 is a diagram showing the second stage encoding of the input image. FIG. 5 is a view of the projection optical system viewed from the side. FIG. 6 is a perspective view of the projection optical system. FIG. 7 is a diagram showing the relationship between the lighting state of the light source and the projected image on the screen. Explanation of symbols: 1...Light source surface, 2,3,4,5...Point light source, 6...Encoded input image, 7,8...Encoded image, 9...Screen, 10...Decoding mask, 11...
Input image A, 12... Input image B, 13... (3, 3)th pixel a _ij of input image A, 14... (3, 3) pixel b _ij of input image B, 15... th pixel of encoded image 7 (3, 3) pixel, 16...(3, 3) pixel of encoded image 8
3) Pixel, 17...(3,3) of the encoded input image
Pixel, 18...One pixel (i,
j), 19... window position of a _ij =1, b _ij =1, 20
...a _ij =1, b _ij =0 window position, 21...a _ij =
0, b _ij =1 window position, 22...a _ij =0, b _ij =
0 window position, 23...Divergent light from light source 2, 24
... Divergent light from light source 3, 25... Divergent light from light source 4, 26... Divergent light from light source 5, 27... 18 projected images by light source 2, 28... 18 projected images by light source 3, 29... Light source 4 18 projection images, 30...
18 projected images by the light source 5, 31...function output position.

Claims (1)

[Claims] 1 Using an encoding method that converts the shading information of pixels forming an image into position information of a window with a constant area and constant transmittance within the pixel section, the input image is used as an encoded input image, and its encoding is performed. From a single projection image of the input image or from multiple projection images that are regularly shifted and stacked two-dimensionally, the calculation results of all two-variable binary logic functions are processed in two-dimensional parallel manner as light contrast information. A parallel optical logic operation method characterized by the following: