JPH0773315A - Optical pattern recognizing and classifying device - Google Patents

Abstract

PURPOSE:To obtain a device which can easily and exactly extract a desired image pattern out of a lot of image patterns without using any matched filter by providing a pattern extracting means, feature amount detecting means and belonging category discriminating means for the extracted pattern. CONSTITUTION:This device is provided with a pattern extracting means 1 for extracting the desired pattern from an input image, means 2 (optical correlation arithmetic means) 2 for detecting the feature amount of the extracted pattern, and discriminating means 3 for discriminating the belonging catagory of the extracted pattern based on the carried feature amount. Then a representative pattern is extracted from the input image as an output image by performing Fourier transform while using the pattern extracting means 1, the position of any specified pixel is carried out, and the suitable size of the input image is segmented from that output. On the other hand, the quantity of light in the image provided by the Fourier transform of the extracted pattern is carried out in comparison with a reference pattern by the feature amount detecting means 2. Thus, since only the Fourier transformed image is plotted at a first space modulator 17 and no interference strip is plotted, no hologram or the like is required.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention extracts a specific object or character from an image including many objects or characters such as a landscape, and extracts the extracted object, character or image. The present invention relates to an optical pattern recognition classification device used for pattern recognition and retrieval, inference, association, and the like.

[0002]

2. Description of the Related Art A specific pattern to be extracted is extracted from an image (hereinafter referred to as an input image) that includes patterns of many objects and characters such as everyday scenery seen by humans, and recognition classification is performed. A matched filtering method is known as one of such methods.

In the matched filtering method, the Fourier transform image of the input image is formed on the back focal plane of the Fourier transform lens and is incident on the matched filter of the pattern to be extracted, which is arranged at the focal position, and transmits the transmitted light. Further Fourier transform by the Fourier transform lens, when the same pattern as the pattern to be extracted exists in the input image,
This is a method in which a correlation spot showing a high correlation appears on the back focal plane of the Fourier transform lens, and the position of the same pattern as the pattern to be extracted in the input image can be known by its position, and recognition classification can be performed. At this time, for the matched filter, a hologram in which the Fourier transform of the intensity distribution of the pattern to be extracted is recorded in the form of interference fringes with the reference light is generally used.

However, in this method, when a hologram is used as a filter, it is necessary to make a filter for each pattern to be extracted, and an allowable error in alignment of the filter is extremely small, that is, several μm.
There is such a problem that the positional deviation of (2) significantly deteriorates the performance.

Further, without using a hologram for the filter,
There is also a method using a spatial light modulator that can change the pattern to be extracted in real time, but the existing spatial light modulator does not have the ability to write interference fringes with a fine pitch equivalent to that of a hologram, and the performance of the device is considerably high. There is a problem of being inferior.

As another method, there is a template matching method used in a digital image processing device of a computer or the like. This template matching method is a method of moving a pattern to be extracted in an input image to find a most suitable place. As this method, there are a method for sequentially testing residuals, a method using a cross-correlation coefficient, and the like. However, in either method, basically, it is necessary to scan the pattern to be extracted in the input image, which takes time. This is a problem that becomes more prominent as the number of pixels increases.

Further, in any of the above methods, the pattern existing in the input image and the pattern present outside the input image are presented.
If the pattern to be extracted does not quite match, in other words, if the correlation is not sufficiently high, it is difficult to extract it from the input image, and even if there is only a slight difference in the shape of the pattern, for example, for each pattern, The matched filtering method has a drawback that a plurality of filters must be produced.

Therefore, in order to provide flexibility in pattern extraction, a method of extracting a pattern from an input image by using a special filter called SDF or MACE has been proposed, but in this case as well, it is desired to extract in advance. There is a problem that a complicated filter corresponding to the pattern must be designed and manufactured.

[0009]

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems in the prior art.
To provide an optical pattern recognition / classification device in which alignment of an optical system is relatively easy without using a matched filter that draws interference fringes such as holograms that must be manufactured for each pattern to be extracted. That is the purpose. Further, the present invention is not limited to a pattern that is the same as the pattern to be extracted from an image including many patterns of objects and characters without scanning the pattern to be extracted, and has a certain degree of correlation. It is an object of the present invention to provide an optical pattern recognition / classification apparatus which can easily extract patterns at the same time using a simple filter, and can perform finer recognition classification for each extracted pattern.

[0010]

The present invention has been made to solve the above technical problems. The optical pattern recognition classification device of the present invention inputs a pattern belonging to a desired large category composed of a set of one or more small categories including one or more patterns to be recognized and classified as an input image (hereinafter referred to as an input image). ), And an amount (hereinafter, referred to as a feature amount) representing each feature of the pattern (hereinafter, referred to as an extracted pattern) extracted by the pattern extraction unit is subjected to a correlation or similar operation. The feature amount detecting means for detecting using the optical system, and the determining means for determining the small category to which the extraction pattern belongs based on one or more feature amounts detected by the feature amount detecting means. The pattern extracting means displays a light source that emits a first coherent light beam and an input image with intensity transmittance or reflectance distribution. A first image display unit for giving an intensity distribution to the first coherent light beam, and a first Fourier transform lens for optically Fourier transforming the intensity distribution of the light beam transmitted or reflected by the first image display unit. A first Fourier transform image output means for outputting a Fourier transform image of a pattern (hereinafter, referred to as a representative pattern) having a common characteristic to all patterns included in the desired large category, and the first Fourier transform lens. A first spatial light modulator which is disposed on the Fourier transform surface of the first pattern, displays the Fourier transform image of the representative pattern as an intensity transmittance or reflectance distribution, and gives a further intensity distribution to the first light flux. A second Fourier transform lens for optically further Fourier transforming the intensity distribution of the light flux transmitted or reflected by the first spatial light modulator; and the second Fourier transform lens. An imaging means for imaging the output image obtained by the Fourier transform by the transform lens,
An image feedback unit for displaying the output image again as an input image of the first image display unit, and a pixel having an output value of a certain value or more in the output image is detected, and a pixel having a certain number or more of this pixel is detected. The position detecting means for detecting the position of the group, and the area cutting-out means for cutting out an area of an appropriate size from the input image on the basis of the output from the position detecting means, and the characteristic A quantity detecting means for emitting a second coherent light beam;
A second image display unit that displays one or a plurality of reference patterns with an intensity transmittance or a reflectance distribution to give an intensity distribution to the second coherent light flux, and a second image display unit that transmits the reference pattern. Alternatively, a third Fourier transform lens that optically Fourier transforms the intensity distribution of the reflected light flux,
Second Fourier transform image output means for outputting a Fourier transform image of the extracted pattern obtained by the pattern extracting means, and Fourier transform image of the extracted pattern arranged on the Fourier transform surface of the third Fourier transform lens. To
The intensity transmittance or reflectance distribution is displayed and displayed as the second
A second spatial light modulator which further gives an intensity distribution to the luminous flux of
A fourth Fourier transform lens for optically further Fourier transforming the intensity distribution of the light flux transmitted or reflected by the second spatial light modulator, and an image obtained by Fourier transforming by the fourth Fourier transform lens. And a light receiving unit that detects the amount of light corresponding to the one or more reference patterns.

Further, another optical pattern recognition / classification apparatus of the present invention includes a pattern including one or more patterns to be recognized / classified.
A pattern extracting unit that extracts a pattern belonging to a desired large category consisting of a set of two or more small categories from an input image (hereinafter referred to as an input image), and a pattern extracted by the pattern extracting unit (hereinafter An amount representing each characteristic of the extraction pattern) (hereinafter referred to as a characteristic amount) is detected by the characteristic amount detecting unit that detects the amount using an optical system that performs a correlation or a similar operation, and the amount detected by the characteristic amount detecting unit. The pattern extraction means is configured by a determination means for determining the small category to which the extraction pattern belongs based on one or more feature quantities.
A light source that emits a first coherent light beam, a first image display unit that displays an input image with intensity transmittance or reflectance distribution, and gives an intensity distribution to the first coherent light beam; The first Fourier transform lens for optically Fourier transforming the intensity distribution of the light flux transmitted or reflected by the image display means, and a pattern having common features to all the patterns included in the desired large category (hereinafter, a representative pattern 1) which outputs a Fourier transform image of
Of the Fourier transform image output means and the Fourier transform surface of the first Fourier transform lens, and the Fourier transform image of the representative pattern is displayed as an intensity transmittance or reflectance distribution to display the first Fourier transform image. A first spatial light modulator for further imparting an intensity distribution to the light flux; a second Fourier transform lens for optically further Fourier transforming the intensity distribution of the light flux transmitted or reflected by the first spatial light modulator; Image pickup means for picking up an output image obtained by Fourier transforming by a second Fourier transform lens, image returning means for displaying the output image again as an input image of the first image display means, and the output image A position detecting means for detecting a pixel having an output value of a certain value or more, and detecting the position of a group of a certain number or more of the pixel, and an output from the position detecting means as a reference. And a feature light source that emits a second coherent light beam, and one or a plurality of regions. Second reference image display means for displaying the reference pattern with intensity distribution or reflectance distribution to give an intensity distribution to the second coherent light beam, and each lens is located at the position of the one or more reference patterns. Is obtained by the pattern extracting means and the first Fourier transform lens array for optically Fourier transforming the intensity distribution of the light flux transmitted or reflected by the second image display means for each reference pattern. The second Fourier transform image output means for outputting the Fourier transform image of the extracted pattern and the image of the Fourier transform image of the extracted pattern are output to the first screen. D) An image processing device arranged in parallel at a position corresponding to each lens of the transform lens array, and the extraction arranged on the Fourier transform surface of each lens of the first Fourier transform lens array and created by the image processing device A third spatial light modulator for displaying a parallel image of a Fourier transform image of a pattern with an intensity transmittance or a reflectance distribution to further give an intensity distribution to the second light flux, and each lens includes one or a plurality of the spatial light modulators. A second Fourier transform lens array which is arranged corresponding to the position of the reference pattern and optically further Fourier transforms the intensity distribution of the light flux transmitted or reflected by the third spatial light modulator; Receiving the amount of light of the third output image obtained by the Fourier transform by each lens of the Fourier transform lens array of No. 1 in correspondence with the one or more reference patterns. And a light means.

In the apparatus of the present invention, at least one of the first or second Fourier transform image output means.
One is a light source that emits a coherent light beam, a third image display unit that displays the representative pattern or the extracted pattern as an intensity transmittance or reflectance distribution, and gives an intensity distribution to the light beam. A fifth Fourier transform lens for optically Fourier transforming the intensity distribution of the light flux transmitted or reflected from the third image display means, and the representative pattern obtained by the fifth Fourier transform lens,
Alternatively, it is preferably configured by an image pickup means for picking up a Fourier transform image of the extraction pattern.

In the apparatus of the present invention, it is preferable that at least one of the first, second, and third spatial light modulators is a first optical writing type spatial light modulator. .

Further, in the apparatus of the present invention, the first image display means includes a second optical writing type spatial light modulator and an input image on a writing surface of the second optical writing type spatial light modulator. It is preferable that the image forming lens system is configured to form an image.

Further, in the apparatus of the present invention, at least one of the first, second and third spatial light modulators.
Furthermore, it is preferable that the Fourier transform image of the representative pattern or the extracted pattern is displayed as an image binarized by an appropriate threshold value substantially in real time.

Further, in the apparatus of the present invention, at least one of the first and second optical writing type spatial light modulators performs an operation of binarizing an image written from the writing surface by a threshold value and displaying the image. It is preferable that

Further, in the above-mentioned apparatus of the present invention, the discriminating means acquires, for each small category, the feature quantities of the extraction patterns belonging to the small category a plurality of times, and the representative value of the set of the obtained outputs is used as a reference. Then, a membership function of the extraction pattern is created in advance for each component of the extraction pattern and the feature amount (hereinafter referred to as a feature parameter), and each small category is extracted from the feature amount of the extraction pattern extracted from the input pattern. The membership value is calculated for each characteristic parameter based on each of the membership functions, and the average value or minimum value of the obtained membership values is defined as the degree of belonging to each small category of the extraction pattern, It is preferable to use an arithmetic means that determines an extraction pattern from this degree.

Further, in the apparatus of the present invention, the discriminating means is a neural network which inputs each feature amount acquired for each of the extraction patterns and has an output indicating a small category of the extraction pattern. It is preferable that the extraction pattern be determined from.

[0019]

With the configuration of the optical pattern recognition and classification device of the present invention as described above, the first coherent light beam emitted from the light source is the first coherent light beam disposed on the front focal plane of the first Fourier transform lens. It is incident on the image display means. In the first image display means, since the input image is displayed with the intensity transmittance or reflectance distribution, the incident first coherent light flux is modulated into the intensity distribution of the input image and is transmitted or reflected as light. The first image display means is emitted. The modulated first coherent light flux passes through the first Fourier transform lens and forms a Fourier transform image of the input image on the rear focal plane.

On the other hand, a Fourier transform image of a pattern (hereinafter referred to as a representative pattern) having a common characteristic to all patterns belonging to a desired large category consisting of a set of one or more small categories including one or more patterns is obtained. It is output by the first Fourier transform image output means, and is displayed on the first spatial light modulator as an intensity transmittance or reflectance distribution. Since the first spatial light modulator is arranged on the rear focal plane of the first Fourier transform lens, the Fourier transform image of the input image is incident on the first spatial light modulator, and the Fourier transform of the representative pattern is performed. Further modulated by the intensity distribution of the image, the output of the product of the Fourier transform image of the input pattern and the Fourier transform image of the representative pattern is transmitted or reflected.

As described above, since only the Fourier transform image is drawn on the first spatial modulator and no interference fringes are drawn, a hologram for drawing the interference fringes used in a method such as the matched filter method or the like. Eliminates the need to use Therefore, the arrangement of the first spatial light modulator does not require strictness as compared with the matched filter. Moreover, when changing the representative pattern, it suffices to change the Fourier transform image of the representative pattern drawn on the first spatial light modulator, which can be easily performed by using a computer or the like. Furthermore, the representative pattern may be a simple one such as a circle or a square, and does not require a complicated one.

Thereafter, the first coherent light beam is again Fourier-transformed by the second Fourier transform lens disposed after the spatial light modulator by the rear focal length of itself,
An output image is formed at the front focus position of the second Fourier transform lens. This output image is the result of the cross-correlation operation between the input image and the representative pattern, and it can be said that a portion having a spatial frequency component similar to that of the representative pattern remains in the input image.

The output image is picked up by the image pickup means.
The captured output image is displayed again on the first image display unit as the input image by the image returning unit. The series of processes described above is repeated again. If the number of repetitions is controlled appropriately, only the patterns that correlate with the representative pattern in the input image remain and are output until the end.In the end, patterns that belong to the desired large category are extracted from the input image. The output image can be obtained. Further, since the above-described processing simultaneously performs parallel processing on the entire input image with one coherent light beam, there is no need for scanning time unlike template matching.

The finally captured output image is input to the position detecting means. The position detecting means detects a pixel having an output value equal to or higher than a certain level in the output image, and detects the position of a group in which the pixel is gathered at a certain number or more. Based on the information of this position, an area having an appropriate size corresponding to the size of the group of pixels is cut out from the original input image, and the extraction pattern is acquired. By doing this, one or more patterns belonging to a large category formed by a group of one or more small categories are extracted from an input image containing patterns of many shaped objects and characters. It is possible to extract not only a pattern having a high correlation with the representative pattern but also a pattern having a certain degree of correlation at the same time.

Further, the second coherent light beam emitted from the light source is incident on the second image display means arranged at the front focal position of the third Fourier transform lens. In the second image display means, since one or a plurality of reference patterns are displayed with an intensity transmittance or reflectance distribution, the incident second coherent light beam has an intensity distribution of one or a plurality of reference patterns. And is emitted to the second image display means as transmitted or reflected light. The modulated second coherent light flux passes through the third Fourier transform lens and forms a Fourier transform image of one or a plurality of reference patterns on the back focal plane.

On the other hand, the Fourier transform image of the extracted pattern extracted from the input image is output by the second Fourier transform image output means, and is displayed on the second spatial light modulator in the intensity transmittance or reflectance distribution. . Since the second spatial light modulator is arranged in the focal plane of the third Fourier transform lens, the Fourier transform image of one or a plurality of reference patterns is incident on the second spatial light modulator and the extracted pattern is obtained. Is further modulated by the intensity distribution of the Fourier transform image of 1), and the output of the product of the Fourier transform image of one or a plurality of reference patterns and the Fourier transform image of the extraction pattern is transmitted or reflected.

As described above, since the Fourier transform image is only drawn on this second spatial light modulator and no interference fringes are drawn, a hologram for drawing the interference fringes used in the method such as the matched filter method. It becomes unnecessary to use such as. Therefore, the arrangement of the second spatial light modulator does not need to be strict as compared with the matched filter.

After that, the second coherent light beam is again Fourier-transformed by the fourth Fourier transform lens disposed behind the second spatial light modulator by the back focal length of itself, and the second coherent light beam is subjected to Fourier transform again. The amount of light is detected by the light receiving means arranged corresponding to one or a plurality of reference patterns. The detected output increases as the degree of coincidence between the Fourier transform image of the reference pattern and the Fourier transform image of the extraction pattern on the second spatial light modulator increases, so that the cross-correlation calculation of the reference pattern and the extraction pattern is performed. It can be said that the result is output, and each component of the plurality of feature quantities of the extraction pattern can be acquired simultaneously and in parallel with a simple optical system. Further, since the comparison between one or a plurality of reference patterns and the extracted pattern is performed on the Fourier transform plane,
Even if the extraction pattern moves, position invariance of the output is ensured.

The feature amount detected by the above-described process of the feature amount detecting means is input to the discriminating means and it is determined which small category included in the large category the extraction pattern belongs to.

The second coherent light beam emitted from the light source is incident on the second image display means arranged at the front focal position of each lens of the first Fourier transform lens array. In the second image display means, since one or a plurality of reference patterns are displayed with a strong transmittance or reflectance distribution, the incident second coherent light flux is 1
Modulated to the intensity distribution of one or more reference patterns,
The second image display means is emitted as transmitted or reflected light.
The modulated second coherent light beam passes through the first Fourier transform lens array in which each lens is arranged corresponding to the position of one or a plurality of reference patterns, and one is generated at the rear focal position of each lens. Alternatively, Fourier transform images of a plurality of reference patterns are formed for each lens.

On the other hand, the Fourier transform image of the extraction pattern extracted from the input image is output by the second Fourier transform image output means, and in the image processing device, at a position corresponding to each lens of the first Fourier transform lens array, An image in which a plurality of Fourier transform images of the extraction pattern are arranged in parallel is created. After that, the parallel image of the Fourier transform image of the extraction pattern is displayed on the third spatial light modulator with the intensity transmittance or reflectance distribution at the position corresponding to each lens of the first Fourier transform lens array. Since the third spatial light modulator is arranged on the rear focal plane of each lens of the first Fourier transform lens array, the Fourier transform image of one or a plurality of reference patterns is obtained by the third spatial light modulator. Is further modulated by the intensity distribution of the Fourier transform image of the extraction pattern, and the output of the product of the Fourier transform image of one or more reference patterns and the Fourier transform image of the extraction pattern is transmitted or reflected.

As described above, since only the Fourier transform image is drawn and no interference fringes are drawn on the third spatial modulator, a hologram for drawing the interference fringes used in the method such as the matched filter method is used. Eliminates the need to use Therefore, the arrangement of the third spatial light modulator does not require strictness as compared with the matched filter.

After this, each lens of the second Fourier transform lens array is placed behind the third spatial light modulator by its own rear focal length, by which each second coherent light beam is again Fourier-transformed. The light amount of the converted image obtained at the front focal position of each lens of the second Fourier transform lens array is detected by the light receiving means arranged corresponding to one or a plurality of reference patterns. The detected output increases as the degree of coincidence between the Fourier transform image of the reference pattern and the Fourier transform image of the extraction pattern on the third spatial light modulator increases, so that the cross-correlation operation between the reference pattern and the extraction pattern is performed. It can be said that the result is output, and each component of the plurality of feature quantities of the extraction pattern can be acquired simultaneously and in parallel with a simple optical system. Further, since the comparison between one or a plurality of reference patterns and the extracted pattern is performed on the Fourier transform plane, the position invariance of the output is secured even if the extracted pattern moves.

The feature amount detected by the above-described process of the feature amount detecting means is input to the discriminating means, and it is determined which small category included in the large category the extraction pattern belongs to.

Further, in any one or more or all of the first or second Fourier transform image output means, the coherent light beam emitted from the light source has a representative pattern or an extracted pattern with a high transmittance, or The light is displayed by the reflectance distribution and is incident on the third image display means arranged on the front focal plane of the fifth Fourier transform lens.
The incident coherent light beam is modulated into the intensity distribution of the representative pattern or the extraction pattern, and is emitted as transmitted or reflected light. A Fourier transform image of the representative pattern or the extracted pattern is formed, and the pattern is imaged by the imaging means. By adopting such a means, it is possible to output the Fourier transform image of the representative pattern or the extracted pattern easily and optically, simply by displaying the representative pattern or the extracted pattern. Further, the representative pattern or the extraction pattern can be changed only by changing the image displayed on the third image processing means.

Further, by using an optical writing type for any one or more or all of the first, second or third spatial light modulators, the Fourier transform image output means is made to have a representative pattern, Alternatively, the Fourier transform image of the extracted pattern can be directly written in the spatial light modulator, and the representative pattern or the Fourier transform image of the extracted pattern can be optically displayed more quickly and easily.

Further, in the first image display means, an input image is directly formed on the writing surface of the spatial light modulator of the optical writing type by using the image forming lens system, so that an electric means such as a liquid crystal television is formed. It is possible to display an input image with intensity transmittance or reflectance distribution optically at high speed and easily without using an image feedback means, and the image feedback means also outputs the output image directly onto the writing surface of the optical writing type spatial light modulator. By forming an image, it can be realized easily at high speed.

Furthermore, a Fourier transform image of the representative pattern or the extracted pattern is substantially real-timed on any one or more or all of the first, second, and third spatial light modulators. Then, by displaying as an image binarized by an appropriate threshold value, it is possible to improve the prevention of contrast reduction of the extracted pattern in the output image or the reference pattern incident on the light receiving means.

Further, an operation of displaying any one or all of the first or second optical writing type spatial light modulators by binarizing the image written from the writing surface by a threshold value is displayed. Also by using such a method, it is possible to improve the reduction of the contrast of the extracted pattern in the output image or the reference pattern incident on the light receiving means.

Further, in the discriminating means used in the device of the present invention, by creating a membership function for each feature parameter for each extraction pattern, in the feature space having the dimension of the number of feature parameters, the degree to which it belongs is a member. It is possible to create a small category for each extraction pattern expressed by the ship function. In addition, by acquiring the characteristic parameter a plurality of times and using the representative value thereof as the reference for creating the membership function,
It is possible to create small categories that take into account fluctuations in output. After the creation, the membership value of each component is obtained from the feature amount obtained for each unknown extraction pattern, and the minimum or average of the obtained membership values is obtained to determine which small The degree to which the category belongs can be obtained, and the extraction pattern can be determined.

Further, a neural network is used as the discriminating means, each feature amount obtained by the feature amount detecting means is input to each node of the input layer of the neural network, and the output state of the node of the output layer of each neural network is input. By discriminating the small category to which the extracted pattern belongs, the recognition classification of the pattern can be performed easily and effectively. In the neural network, learning is performed to increase the weighting of the criterion of the characteristic parameter having a large contribution to the pattern discrimination, reduce the weighting of the criterion of the characteristic parameter having a small contribution, and further,
When there are a plurality of patterns having similar features, the part having the difference between the features is particularly emphasized in the determination, so that the determination standard can be easily created.

Next, the optical pattern recognition classification device of the present invention will be specifically described by way of examples, but the present invention is not limited thereto.

Embodiment 1 FIG. 1 is a schematic block diagram of an optical pattern recognition classification apparatus showing an embodiment of the present invention. In this embodiment, the pattern extracting unit 1 extracts a pattern belonging to a large category including some small categories from the input image, and the optical correlation calculating unit 2 and the discriminating unit 3 extract the extracted pattern belonging to the large category. Perform a detailed classification of which subcategory the pattern belongs to.

The coherent light beam 12 emitted from the laser 11 is collimated by the beam expander 13 so as to have an appropriate light beam diameter. The first light flux 12 a enters the first image display means 15. First image display means 15 arranged on the front focal plane of the first Fourier transform lens 16.
, The input image from which the patterns belonging to the large category are to be extracted is drawn with the transmittance distribution of the intensity pattern.
The light beam 12a is transmitted and the input image is read. The first light flux 12a passes through the first Fourier transform lens 16 and forms a Fourier transform image of the input image on the first spatial light modulator 17 arranged on the rear focal plane of the first Fourier transform lens 16.

On the other hand, a pattern (hereinafter referred to as a representative pattern) having a feature common to a desired large category to be extracted from the input image is previously stored in the computer 18, and the first Fourier transform image output means 19 represents the representative pattern. Fourier transform image of the first spatial light modulator 1 is output.
In FIG. 7, the intensity pattern of the Fourier transform image of the target pattern is displayed as a transmittance distribution. At this time, the computer 18
In, the Fourier transform image of the representative pattern may be stored in advance. Therefore, the Fourier transform image of the input image is modulated by the intensity pattern of the Fourier transform image of the representative pattern when passing through the first spatial light modulator 17,
The product of the corresponding spatial frequency components of the input image and the representative pattern is calculated. After that, the first light flux 12a passes through the second Fourier transform lens 20 and undergoes Fourier transform again, and is cross-correlated between the input image and the representative pattern on the screen 21 disposed on the front focal plane of the second Fourier transform lens 20. The result of the operation is output.

More specifically, the output image is an image in which the input image having a small correlation with the representative pattern is deleted.
The part of the input image that has a high degree of correlation with the representative pattern is
Contrast is output without deterioration, and conversely, the part with little correlation with the representative pattern is output with almost no contrast, and an image with a pattern similar to the representative pattern remains is output. The means 22 captures an image. After the output image is picked up by the image pickup means 22, it is displayed again as the input image on the first image display means 15, and if the series of processes described above is repeated an appropriate number of times,
A part of the input image having a low degree of correlation with the representative pattern disappears, and an image in which only the pattern having a higher degree of similarity with the representative pattern remains until the end is output. At this time, the number of repetitions is controlled to an appropriate number by the program of the computer 18.

Finally, the output image is taken into the computer 18 via the image pickup device 22, and the position of the pattern remaining in the output image is detected by the image processing device 23. Pixels having an output value of a certain value or more are detected in the output image, a group of pixels having a certain number or more is detected, and information on the position is output. An area having an appropriate size corresponding to the size of a group of pixels is cut out from the input image based on the position information, and becomes an extraction pattern. The image processing device 23 can be realized by using a general image processing board or the like that is used by connecting to a personal computer or the like.

By the above-described processing, one or a plurality of patterns having a correlation with the representative pattern is extracted and becomes an extracted pattern. That is, all the patterns belonging to the large category having the common feature are extracted from the input image as the extraction patterns.

In contrast to the device of the present invention as described above, in the conventional device used in the matched filtering method, the Fourier transform image of the input image is converted into the first Fourier transform lens 16 as shown in FIG. The second incident light is incident on the pattern matched filter 41 to be extracted which is formed on the rear focal plane and is arranged at this focal position, and the transmitted light is further Fourier transformed by the second Fourier transform lens 20. At this time, if the same pattern as the pattern to be extracted exists in the input image, a correlation spot showing a high correlation appears on the screen 21 arranged on the back focal plane of the second Fourier transform lens 20. The position of the same pattern as the pattern to be extracted in the input image can be detected from the position of this spot, and recognition and classification can be performed. A hologram is used as the matched filter 41. However, when a hologram is used as described above, it is necessary to create a separate filter for each pattern to be extracted, which causes a problem in alignment of the filter.

In the device of the present invention, the extraction pattern is classified into smaller subcategories in the subsequent processing. Here, FIG. 3 shows an example of the input image, FIG. 4 shows an example of the representative pattern of the input image of FIG. 3, and FIG. 5 shows an example of the output image of the input image shown in FIG.
FIG. 6 shows an example of the extraction pattern extracted by the analysis of the output image.

On the other hand, in FIG. 1, the second light beam 12b of the two light beams divided by the half mirror 14 is deflected by the mirror 24 and is incident on the second image display means 25. The reference pattern stored in the computer 18 is drawn in the intensity transmittance distribution on the second image display unit 25 arranged on the front focal plane of the third Fourier transform lens 26, and the second light beam 12b is transmitted to form the reference pattern. Read out. Second
The light beam 12b passes through the third Fourier transform lens 26,
A Fourier transform image of the reference pattern is formed on the second spatial light modulator 27 arranged on the back focal plane of the third Fourier transform lens 26. On the second spatial light modulator 27, the intensity pattern of one Fourier transform image of the extraction patterns extracted by the above-described processing output by the second Fourier transform image output means 28 is displayed as a transmittance distribution. Therefore, the Fourier transform image of the reference pattern is modulated by the Fourier transform image of the extraction pattern when passing through the second spatial light modulator 27, and the product of the corresponding spatial frequency components of the reference pattern and the extraction pattern is calculated. Done.

After that, the second light flux 12b is transmitted through the fourth Fourier transform lens 29, is again Fourier transformed, and is placed on the light receiving surface of the light receiving means 30 arranged on the front focal plane of the fourth Fourier transform lens 29. A pattern resulting from the cross-correlation operation between the reference pattern and the extracted pattern is created. To be precise, this pattern is obtained by further performing the calculation calculation of the extraction pattern and the correlation calculation of the reference pattern.
The light quantity of this pattern is detected by the light receiving means 30, and the output of the cross-correlation calculation is obtained. Hereinafter, the second image display means 25
The reference pattern to be displayed on the display is sequentially changed, and the above-described processing is performed to obtain the output of the cross-correlation calculation for the plurality of reference patterns.

These outputs are input to the identification / determination device 31 to determine which small category the extraction pattern belongs to. This determination is realized by previously examining what range of values the plurality of outputs of the extracted pattern with respect to each reference pattern show. As long as the ranges do not overlap among the output for all reference patterns among multiple small categories, the binary logic of whether or not the output value for each reference pattern of the input extraction pattern falls within the range previously examined. It can be done from a combination. Also, as a method of statistically evaluating the distance from each small category of a vector (called a feature vector) that has the characteristic parameter of the input extraction pattern as a component, Mahalanobis distance, urban distance, etc. have been proposed. It is also possible to make a determination using these distances. The identification / judgment device 31 can be realized with a judgment function by software using the computer 18, or can be realized with a similar function by hardware by an electronic circuit.

As the first and second image display means 15 and 25 used in the device of the present invention, a liquid crystal panel which is a general electric address type spatial light modulator can be used. Also, here, the first and second spatial light modulators 17,
A liquid crystal panel or the like which is a general electric address type spatial light modulator can be used for 27. Further, a CCD camera, an image pickup tube, or the like can be used as the image pickup device 22.

In addition, a circular mask or the like having an appropriate radius centered on the optical axis is arranged on the first spatial light modulator 17 to remove the representative pattern, the input image, or both images near the optical axis. You may go. As a result, low spatial frequency components that have almost any pattern in common are cut, and the pattern having a correlation with the representative pattern can be more efficiently extracted. Other than the use of a mask, there is also a method of removing the image near the optical axis when the Fourier transform image of the correlation pattern is output by the Fourier transform image output device 19.

Furthermore, when the binarized image of the intensity pattern of the Fourier transform image of the representative pattern is displayed on the first spatial light modulator 17, the Fourier transform image of the input image that matches the Fourier transform image of the representative pattern is displayed. Since the part is transmitted with almost no effect of the spatial light modulator, it is possible to prevent a decrease in contrast of the pattern having a correlation with the representative pattern remaining in the output image. In addition, the second spatial light modulator 2
If the binarized image of the Fourier transform image of the extraction pattern is displayed in 7, the portion of the Fourier transform image of the reference pattern that matches the Fourier transform image of the extraction pattern is transmitted with almost no effect of the spatial light modulator. Therefore, the light receiving means 3
It is possible to prevent a decrease in the light amount of the light flux that reaches 0.

As such a spatial light modulator, an electrically addressed type using a ferroelectric liquid crystal can be used. In particular, a spatial light modulator using a ferroelectric liquid crystal is suitable not only for binarizing an image, but also in principle has a high resolution and a high contrast. Further, if the binarization processing is performed using an image processing device or the like, the same can be done by using a general electric address type spatial light modulator.

As the light receiving means 30, an image pickup device such as a photodiode or CCD can be used. Furthermore, in the present embodiment, the half mirror 14 and the mirror 24 are used to divide the luminous flux into two, but the apparatus configuration is such that two lasers and two beam expanders are used to separately generate two luminous fluxes. But there is no problem.

Embodiment 2 FIG. 7 is a schematic block diagram showing another embodiment of the optical pattern recognition classification device of the present invention. In the second embodiment, the Fourier transform image of the representative pattern displayed on the first spatial light modulator 17 and the Fourier transform image of the extracted pattern displayed on the second spatial light modulator 27 are optically obtained, and a plurality of references are provided. The patterns are displayed in parallel on the second image display means 25, and each lens is arranged in a position corresponding to the reference pattern.
Also, it is characterized by using the second Fourier transform lens arrays 51 and 52. Also in this embodiment, the pattern extracting unit 1 extracts a pattern belonging to a large category including some small categories from the input image, and the optical correlation calculating unit 2 and the discriminating unit 3 extract the extracted pattern belonging to the large category. Perform a detailed classification of which sub-category the pattern belongs to.

The coherent light beam 12 emitted from the laser 11 is collimated by the beam expander 13 so as to have an appropriate light beam diameter, and the half mirrors 14a and 14a.
b, 14c and 14d respectively, and the light fluxes 12a and 1
It is divided into 2b, 12c and 12d. The light flux 12c enters the image display means 53. On the image display means 53 arranged on the front focal plane of the Fourier transform lens 54, the intensity pattern of the representative pattern stored in the computer 18 is drawn as the transmittance distribution, and the luminous flux 12c is transmitted to form the representative pattern. Read out. After that, the light beam 12c passes through the Fourier transform lens 54 and forms a Fourier transform image of a representative pattern on the screen 55 arranged on the rear focal plane. This image is picked up by the image pickup means 56. Here, as for the representative pattern displayed on the image display means 53, the representative pattern may be directly imaged by another imaging means and the image thereof may be displayed.

On the other hand, the luminous flux 12a is generated by the first image display means 1
It is incident on 5. On the first image display means 15 arranged on the front focal plane of the first Fourier transform lens 16, the input image that is the target of the extraction of the patterns belonging to the large category is drawn with the transmittance distribution of the intensity pattern, and the light flux 12a is generated. The input image is read out in a transparent manner. The light flux 12a passes through the first Fourier transform lens 16 and
A Fourier transform image of the input image is formed on the first spatial light modulator 17 arranged in the rear focal plane. First spatial light modulator 17
, The Fourier transform image of the representative pattern imaged by the image pickup means 56 is displayed with the transmittance distribution of the intensity pattern. Therefore, when the light flux 12a is transmitted, it is modulated by the intensity pattern of the Fourier transform image of the representative pattern. Then, the product of the corresponding spatial frequency components of the input image and the representative pattern is calculated. After this, the light flux 12a passes through the second Fourier transform lens 20 and undergoes Fourier transform again, and is subjected to cross-correlation calculation of the input image and the representative pattern on the screen 21 arranged on the front focal plane of the second Fourier transform lens 20. The result is output.

Specifically, the output image is an image in which the input image having a small correlation with the representative pattern is deleted.
The part of the input image that has a high degree of correlation with the representative pattern is
Contrast is output without deterioration, and conversely, the part with little correlation with the representative pattern is the one with almost no contrast, and an image with a pattern similar to the representative pattern remains is output. The image is taken by the imaging means 22. After the output image is picked up by the image pickup means 22, it is displayed again on the image display means 15 as the input image, and the series of processes described above is repeated an appropriate number of times. Disappear,
An image in which only the pattern having a higher degree of similarity with the representative pattern remains until the end is output. At this time, the number of repetitions is controlled to an appropriate number by the program of the computer 18.

Finally, the output image is taken into the computer 18 via the image pickup means 22, and the position of the pattern remaining in the output image is detected by the image processing device 23. Pixels having an output value of a certain value or more are detected in the output image, a group of pixels having a certain number or more is detected, and information on the position is output. An area having an appropriate size corresponding to the size of a group of pixels is cut out from the input image based on the position information, and becomes an extraction pattern. The image processing device 23 can be realized by using a general image processing board or the like that is used by connecting to a personal computer or the like.

By the above processing, one or a plurality of patterns having a correlation with the target pattern is extracted and becomes an extracted pattern. That is, all the patterns belonging to the large category having the common feature are extracted from the input image as the extraction patterns. Classification of the extraction pattern into smaller subcategories will be performed in the subsequent processing.

On the other hand, in FIG. 7, the light beam 12d enters the image display means 58. On the image display means 58 arranged on the front focal plane of the Fourier transform lens 59, one intensity pattern of the extraction patterns extracted by the above processing is drawn as a transmittance distribution, and the light flux 12d is transmitted. The extraction pattern is read. After that, the light flux 12d passes through the Fourier transform lens 59 and forms a Fourier transform image of the extracted pattern on the screen 60 arranged on the rear focal plane. This image is picked up by the image pickup device 61. The extracted Fourier transform image is input to the image processing device 62, and is converted by the Fourier transform lens array 51 into a plurality of images arranged in parallel at the positions corresponding to the respective lenses. The image processing device 62 can be realized by using a general image processing board or the like that is used by connecting to a personal computer or the like.

Further, the light beam 12b is deflected by the mirror 24 and enters the second image display means 25. On the second image display means 25 arranged on the front focal plane of each lens of the Fourier transform lens array 51, a plurality of reference patterns stored in the computer 18 are drawn in parallel in the intensity transmittance distribution,
The plurality of reference patterns are read out by transmitting the light flux 12b. The light flux 12b is transmitted through the Fourier transform lens array 51 in which each lens is arranged at a position corresponding to each reference pattern, and then onto the second spatial light modulator 27 arranged on the back focal plane of each lens. , Fourier transform images of a plurality of reference patterns can be formed at positions corresponding to the respective lenses. Since the parallel image of the Fourier transform image of the extracted pattern output from the image processing device 62 is displayed on the second spatial light modulator 27, each Fourier transform image of the plurality of reference patterns is the second spatial light modulator. When passing through the device 27, it is modulated by the Fourier transform image of the extracted pattern, and the product of the corresponding spatial frequency components of the reference pattern and the extracted pattern is calculated.

After that, the light beam 12b is again Fourier-transformed by the respective lenses of the Fourier transform lens array 52 in which the respective lenses are arranged corresponding to the positions of the plurality of reference patterns, respectively, and The light is incident on the light receiving means array 63 having the light receiving surface arranged at the front focus position, and a pattern resulting from the cross-correlation calculation of each reference pattern and the extracted pattern is formed on each light receiving surface. The light amounts of these patterns are photoelectrically converted by the respective light receiving means of the light receiving device array 63, and a plurality of outputs of the cross-correlation calculation of each reference pattern and the extracted pattern are simultaneously detected.

These outputs are input to the identification / determination device 31 to determine which small category the extraction pattern belongs to. This determination is realized by previously examining what range of values the plurality of outputs of the extracted pattern with respect to each reference pattern show. As long as the ranges do not overlap among the output for all reference patterns among multiple small categories, the binary logic of whether or not the output value for each reference pattern of the input extraction pattern falls within the range previously examined. It can be done from a combination. In addition, as a method of statistically evaluating the distance from each small category of a vector (called a feature vector) that has the characteristic parameter of the input extraction pattern as a component, the Mahalanobis distance or the city area distance is used for the determination. You can also The identification determination device 31 includes the computer 1
It is also possible to realize the determination function by software by using the above-mentioned 8, and the same function can be realized by hardware by an electronic circuit.

Image display means 15, 25, 5 used here
A liquid crystal panel or the like, which is a general electric address type spatial light modulator, can be used for 3, 58. Further, here, as the spatial light modulators 17 and 27, a liquid crystal panel or the like which is a general electric address type spatial light modulator can be used. In the spatial light modulators 17 and 27, the Fourier transform image of the representative pattern or the extracted pattern is written in the gray level or the binary image. When the binary image is written, the representative pattern in the output image or the light receiving device Array 62
The effect of suppressing a decrease in the contrast of the reference pattern incident on is obtained. In such a case, it is preferable to perform image binarization processing using an image processing device or the like.

Further, the image pickup devices 22, 56 and 61 are provided with C
A CD camera, an image pickup tube or the like can be used. In addition,
In this embodiment, the light flux 12 is transmitted to the half mirrors 14a and 14a.
Although it is divided into four by b and 14c, there is no problem even if an apparatus is provided in which four lasers and beam expanders are respectively prepared and four light beams are separately produced.

Embodiment 3 FIG. 8 is a schematic block diagram showing still another embodiment of the optical pattern recognition classification device of the present invention. The device of FIG. 8 is characterized in that an optical writing type represented by a liquid crystal light valve is used as the spatial light modulator. Also in the third embodiment, the pattern extracting unit 1 extracts a pattern belonging to a large category including some small categories from the input image, and the optical correlation calculating unit 2 and the discriminating unit 3 are extracted.
With this, detailed classification of which small category the extracted pattern belonging to the extracted large category belongs to is performed.

The coherent light beam 72 emitted from the semiconductor laser 71 is collimated by the collimator lens 73 and enters the image display means 53. The intensity pattern of the representative pattern stored in the computer 18 is drawn as the transmittance distribution on the image display means 53 arranged on the front focal plane of the Fourier transform lens 54, and the luminous flux 72 is transmitted to form the representative pattern. Read out. After that, the light flux 72 passes through the Fourier transform lens 54 and forms a Fourier transform image of a representative pattern on the writing surface of the optical writing type spatial light modulator 74 arranged on the rear focal plane. As a result, in the optical writing type spatial light modulator 74, for example, when a liquid crystal light valve is used, the resistance change of the photoconductor layer in the light valve causes the resistance change in the photoconductor layer according to the light intensity distribution of the incident Fourier transform image. An electric field is applied to the liquid crystal to change the alignment of the liquid crystal molecules, and a Fourier transform image of the representative pattern is written.

On the other hand, the coherent light beam 76 emitted from the semiconductor laser 75 is collimated by the collimator lens 77 and enters the image display means 15. An input image is drawn on the image display means 15 arranged on the front focal plane of the Fourier transform lens 16 in the transmittance distribution of the intensity pattern, and the light beam 76 is transmitted to read the input image. The modulated light beam 76 is reflected by the polarization beam splitter 78 and enters the Fourier transform lens 16. At this time, in order to maximize the reflectance at the polarization beam splitter 78, the light emitted from the image display means 15 may be in a linear polarization state (usually an s polarization state) corresponding to the total reflection of the polarization beam splitter. After passing through the Fourier transform lens 16, the Fourier transform image of the input image is incident on the reading surface of the optical writing type spatial light modulator 74 arranged on the rear focal plane and is reflected. At this time, the optical writing type space is used. Light modulator 7
As described above, the Fourier transform image of the representative pattern is written in 4 and the structure of the liquid crystal is changed. Therefore, the portion of the Fourier transform image of the input image corresponding to the portion of the Fourier transform image of the representative pattern is Writable spatial light modulator 74
The polarization state changes according to the pattern written in, and the reflected light again passes through the Fourier transform lens 16 and enters the polarization beam splitter 78 again. Here, the linearly polarized light changed according to the Fourier transform image of the target pattern written in the optical writing type spatial light modulator 74, and the component orthogonal to the state before the change (usually the p-polarized light component) is generated by the polarization beam splitter 78. Directly transmitted, Fourier transform lens 16
The image reaches the screen 21 arranged on the front focal plane of, and this image becomes the output image. Here, the Fourier transform lens 16
Is the Fourier transform lens 20 in the first or second embodiment.
The function of two lenses is realized by one lens.

This output image is the result of the cross-correlation operation between the input image and the representative pattern. Specifically, the output image is an image in which those having a small correlation with the representative pattern in the input image have been deleted, and the portion of the input image having a high degree of correlation with the representative pattern is output without decreasing the contrast. On the contrary, the portion having almost no correlation with the representative pattern is output with almost no contrast, and an image in which a pattern similar to the representative pattern remains is output, and this is imaged by the imaging means 22. . After the output image is picked up by the image pickup means 22, it is displayed again on the image display means 15 as the input image, and the series of processes described above is repeated an appropriate number of times. An image is output that disappears and only the pattern having a higher degree of similarity with the representative pattern remains until the end. At this time, the number of repetitions is controlled to an appropriate number by the program of the computer 18.

Finally, the output image is taken into the computer 18 through the image pickup means 22, and the position of the pattern remaining in the output image is detected by the image processing device 23. Pixels having an output value of a certain value or more are detected in the output image, a group of pixels having a certain number or more is detected, and information on the position is output. An area having an appropriate size corresponding to the size of a group of pixels is cut out from the input image based on the position information, and becomes an extraction pattern. The image processing device 23 can be realized by using a general image processing board or the like that is used by connecting to a personal computer or the like. By the above-mentioned processing, one or a plurality of patterns having a correlation with the target pattern is extracted and becomes an extracted pattern. That is, all the patterns belonging to the large category having the common feature are extracted from the input image as the extraction patterns. Classification of the extraction pattern into smaller subcategories will be performed in the subsequent processing.

Further, the coherent light beam 80 emitted from the semiconductor laser 79 is collimated by the collimator lens 81 and enters the image display means 58. On the image display means 58 arranged on the front focal plane of the Fourier transform lens 59, one intensity pattern of the extraction patterns extracted by the above processing is drawn as a transmittance distribution, and the light flux 80 is transmitted therethrough. The extraction pattern is read.
Thereafter, the light flux 80 passes through the Fourier transform lens 59, and the optical writing type spatial light modulator 8 arranged on the rear focal plane thereof.
The Fourier transform image of the extraction pattern is formed on the writing surface of No. 2, and the Fourier transform image of the extraction pattern is written in the optical writing type spatial light modulator 82 by the above operation.

On the other hand, the coherent light beam 84 emitted from the semiconductor laser 83 is collimated by the collimator lens 85 and enters the image display device 25. On the image display unit 25 arranged on the front focal plane of the Fourier transform lens 26, one or a plurality of reference patterns are drawn in the transmittance distribution of the intensity pattern, and the light flux 84 is transmitted to read the reference pattern. The modulated light beam 84 is reflected by the polarization beam splitter 86 and enters the Fourier transform lens 26. At this time, in order to maximize the reflectance at the polarization beam splitter 86, the light emitted from the image display device 25 may be in a linear polarization state (usually an s polarization state) corresponding to the total reflection of the polarization beam splitter. After passing through the Fourier transform lens 26, the Fourier transform image of one or a plurality of reference patterns is incident on the reading surface of the optical writing type spatial light modulator 82 arranged on the rear focal plane and is reflected at this time. As described above, since the Fourier transform image of the extraction pattern is written in the optical writing type spatial light modulator 82 and the structure of the liquid crystal is changed, the reference pattern corresponding to the portion of the Fourier transform image of the extraction pattern is changed. The polarization state of the portion of the Fourier transform image changes according to the pattern written in the optical writing type spatial light modulator 82, and the reflected light passes through the Fourier transform lens 26 again and enters the polarization beam splitter 86 again. .

Here, the linearly polarized light changed according to the Fourier transform image of the extraction pattern written in the optical writing type spatial light modulator 82, and the component orthogonal to the state before the change (usually p-polarized light component) is the polarized beam. The light is transmitted through the splitter 86 as it is to the light receiving means array 63 in which the light receiving surface is arranged at the front focal plane of the Fourier transform lens 26 at a position corresponding to one or a plurality of reference patterns drawn on the image display means 25. A pattern resulting from the cross-correlation of the extracted pattern and one or more reference patterns is formed on the light-receiving surface upon incidence. The light amount of these patterns is received by the light receiving means array 6
A plurality of outputs of the cross-correlation calculation of one or a plurality of reference patterns and the extracted patterns are simultaneously detected by performing photoelectric conversion or the like by each of the light receiving means of 3. Here, the Fourier transform lens 26 is used in Example 1 (FIG. 1) or Example 2.
The function of the Fourier transform lens 29 in FIG. 7 is also combined, and the function of two lenses is realized by one lens.

These outputs are input to the identification / determination device 31 to determine which small category the extraction pattern belongs to. This determination is realized by previously examining what range of values the plurality of outputs of the extracted pattern with respect to each reference pattern show. As long as the ranges do not overlap among the output for all reference patterns among multiple small categories, the binary logic of whether or not the output value for each reference pattern of the input extraction pattern falls within the range previously examined. It can be done from a combination. In addition, as a method of statistically evaluating the distance from each small category of a vector (called a feature vector) that has the characteristic parameter of the input extraction pattern as a component, the Mahalanobis distance or the city area distance is used for the determination. You can also The identification determination device 31 includes the computer 1
It is also possible to realize the determination function by software by using the above-mentioned 8, and the same function can be realized by hardware by an electronic circuit.

Image display means 15 and 2 used in the apparatus of FIG.
A liquid crystal panel or the like, which is a general electric address type spatial light modulator, can be used for 5, 53 and 58. A CCD camera, an image pickup tube, or the like can be used as the image pickup unit 22. Further, in the device of the third embodiment (FIG. 8), the optical writing type spatial light modulator is used, but as such a spatial light modulator, a liquid crystal panel combined with a photoconductive layer is generally used. , Those using nematic liquid crystal are on sale.

In the optical writing type spatial light modulators 74 and 82, the Fourier transform images of the representative pattern and the extracted pattern are written in the gray level or the binary image. The effect of suppressing the decrease in the contrast of the extracted pattern or the reference pattern incident on the light receiving device array 63 is obtained. As such a spatial light modulator, one using a ferroelectric liquid crystal can be used. In particular, a spatial light modulator using a ferroelectric liquid crystal is suitable not only for binarizing an image but also for high resolution and high contrast in principle.

Further, as an optical address type spatial light modulator, BSO (bismuth silicon oxide; Ba1) is used.
It is also possible to use a medium having a photo-induced refractive index change such as 2SiO20). In particular, in the case of BSO material, visible light of short wavelength with high sensitivity is used as writing light,
Writing and reading can be performed at the same time by using light in the wavelength range from red to infrared, which has low writing sensitivity, as the reading light. However, the BSO material can be used as a reflection type by using a reflection plate, but is usually used as a transmission type. In this case, it is necessary to change the incident direction of the reading light incident on the BSO material, but this is not an essential means of the present invention.

Fourth Embodiment In the fourth embodiment, the processing of the feature amount detecting means and the discriminating means is the same as in the above-mentioned first embodiment (FIG. 1), second embodiment (FIG. 7) and third embodiment (FIG. 8). Therefore, only the pattern extraction device will be described. FIG. 9 is a schematic configuration diagram of a part of a pattern extraction device of still another embodiment of the optical pattern recognition classification device of the present invention. The fourth embodiment is characterized in that the image feedback means is realized only by optical processing.

In FIG. 9, the coherent light beam 72 emitted from the semiconductor laser 71 is collimated by the collimator lens 73 and enters the image display means 53. The intensity pattern of the representative pattern stored in the computer 18 is drawn as the transmittance distribution on the image display means 53 arranged on the front focal plane of the Fourier transform lens 54, and the luminous flux 72 is transmitted to form the representative pattern. Read out. After that, the light flux 72 passes through the Fourier transform lens 54 and forms a Fourier transform image of a representative pattern on the writing surface of the optical writing type spatial light modulator 74 arranged on the rear focal plane. Thereby, the optical writing type spatial light modulator 74
Then, for example, when a liquid crystal light valve is used, the electric field is applied in the liquid crystal due to the resistance change of the photoconductor layer in the light valve according to the light intensity distribution of the incident Fourier transform image, and the liquid crystal molecular alignment changes. Then, the Fourier transform image of the representative pattern is written.

On the other hand, in the initial state, the mirror 91 is arranged as shown in FIG. 10, and the light flux from the input image 90 passes through the coupling lens 92 and is formed on the optical writing type spatial light modulator 93. The image of the input image 90 is formed. As a result, in the optical writing type spatial light modulator 93, for example, when a liquid crystal light valve is used, the resistance change of the photoconductor layer in the light valve causes the resistance change in the liquid crystal according to the light intensity distribution of the incident Fourier transform image. An electric field is applied to the liquid crystal to change the alignment of the liquid crystal molecules, and the image of the input image is written. The input image may be a natural image such as an object or a daily landscape.

Further, the coherent light beam 76 emitted from the semiconductor laser 75 is collimated by the collimator lens 77, passes through the polarization beam splitter 94, and enters the optical writing type spatial light modulator 93. At this time, in order to maximize the transmittance in the polarization beam splitter 94, the emitted light of the semiconductor laser 75 is in a linear polarization state (usually p polarization state) corresponding to the total transmission of the polarization beam splitter.
You can do this.

After passing through the polarization beam splitter 94,
A light beam 76 is emitted to the reading surface of the optical writing type spatial light modulator 93.
Is incident and reflected. At this time, since the input image is written in the optical writing type spatial light modulator 93 as described above and the structure of the liquid crystal is changed, the light flux that hits the image of the input image is changed. The polarization state of the portion changes according to the pattern written in the optical writing type spatial light modulator 93, and the reflected light is incident on the polarization beam splitter 94 again. Here, the linearly polarized light changed according to the image of the input image written in the optical writing type spatial light modulator 93, and the component orthogonal to the state before the change (usually s is the polarized light component) is reflected by the polarization beam splitter 94. , Half mirror 95, mirror 96,
The light enters the polarization beam splitter 78, is reflected as in the case of the polarization beam splitter 94, and enters the Fourier transform lens 16. Here, the half mirror 95 is used to capture an output image.

After passing through the Fourier transform lens 16, the Fourier transform image of the input image is incident on and reflected by the read surface of the optical writing type spatial light modulator 74 arranged on the rear focal plane. Since the Fourier transform image of the representative pattern is written in the mold spatial light modulator 74 and the structure of the liquid crystal is changed as described above, the Fourier transform image of the input image corresponding to the Fourier transform image of the representative pattern is obtained. The polarization state of the part changes according to the pattern written in the optical writing type spatial light modulator 74, and the reflected light again passes through the Fourier transform lens 16 and enters the polarization beam splitter 78 again.

Here, in the linearly polarized light changed according to the Fourier transform image of the target pattern written in the optical writing type spatial light modulator 74, the component orthogonal to the state before the change (usually the p-polarized component) is the polarized beam. The light passes through the splitter 78 as it is. Here, the Fourier transform lens 16 is the same as that of the first embodiment.
(FIG. 1) or the function of the Fourier transform lens 20 in the second embodiment (FIG. 7), and the function of two lenses is realized by one lens.

This transmitted image is an output image obtained by the cross-correlation operation of the input image and the representative pattern. Specifically, the output image is an image in which those having a small correlation with the representative pattern in the input image have been deleted, and the portion of the input image having a high degree of correlation with the representative pattern is output without decreasing the contrast. On the contrary, a portion having almost no correlation with the representative pattern is output with almost no contrast, and an image in which a pattern similar to the representative pattern remains is output. With the mirror 91 arranged as shown in FIG. 9, the output image is made incident on the optical writing type spatial light modulator 93 arranged at the front focus position of the Fourier transform lens 16, and the output image is written again as the input image. When the series of processes described above is repeated an appropriate number of times, the part of the input image having a low degree of correlation with the representative pattern disappears, and an image in which only the pattern having a higher degree of similarity with the representative pattern remains until the end is output.

The final output image is written in the optical writing type spatial light modulator 93, the image is read out in the same manner as described above, a part of the light flux is deflected by the half mirror 95, and is incident on the imaging lens 97. To do. With this imaging lens 97,
An output image is formed on the screen 21. This image is picked up by the image pickup means 22 and taken into the computer 18. After that, the position of the pattern remaining in the output image is detected by the image processing device 23, and an area is cut out from the input image based on the position information to form an extracted pattern. That is, all the patterns belonging to the large category having the common feature are extracted from the input image as the extraction patterns. Classification of the extraction pattern into smaller subcategories will be performed in the subsequent processing. Since the characteristic amount detecting means and the determining means after this are the same as those in the first to third embodiments, description thereof will be omitted.

As the image display means 53 used in the apparatus of FIG. 9, a liquid crystal panel which is a general electric address type spatial light modulator can be used. Further, the image pickup means 22
For this, a CCD camera, an image pickup tube, or the like can be used. Further, in the fourth embodiment, the optical writing type spatial light modulator is used. However, as such a spatial light modulator, a liquid crystal panel combined with a photoconductive layer is generally used, and a nematic liquid crystal is used. The ones that were there are on sale.

In the optical writing type spatial light modulators 74 and 93, the Fourier transform image of the representative pattern and the input image are written in the gray level or the binary image. The effect of suppressing the decrease in the contrast of the extracted pattern of the above is obtained. As such a spatial light modulator, one using a ferroelectric liquid crystal can be used. In particular, a spatial light modulator using a ferroelectric liquid crystal is suitable not only for binarizing an image but also for high resolution and high contrast in principle.

Further, as an optical address type spatial light modulator, BSO (bismuth silicon oxide; Ba1) is used.
It is also possible to use a medium having a photo-induced refractive index change such as 2SiO20). In particular, in the case of BSO material, visible light of short wavelength with high sensitivity is used as writing light,
Writing and reading can be performed at the same time by using light in the wavelength range from red to infrared, which has low writing sensitivity, as the reading light. However, the BSO material can be used as a reflection type by using a reflection plate, but is usually used as a transmission type. In this case, it is necessary to change the incident direction of the reading light incident on the BSO material, but this is not an essential means of the present invention.

Fifth Embodiment In the fifth embodiment, the processing of the characteristic amount detecting means and the discriminating means is the same as in the above-described first to third embodiments, and therefore only the pattern extracting device will be described. FIG. 11 is a schematic configuration diagram of a part of a pattern extraction device of still another embodiment of the optical pattern recognition classification device of the present invention. In the fifth embodiment,
Another feature is that the image feedback means is realized only by optical processing.

In FIG. 11, the coherent light beam 72 emitted from the semiconductor laser 71 is a collimator lens 73.
Is collimated by and enters the image display means 53.
On the image display means 53, the intensity pattern of the representative pattern stored in the computer 18 is drawn as a transmittance distribution, and the light flux 72 is transmitted and the representative pattern is read out. After that, the light flux 72 passes through the Fourier transform lens 54 and forms a Fourier transform image of a representative pattern on the writing surface of the optical writing type spatial light modulator 74 arranged on the focal plane thereof. Thereby, the optical writing type spatial light modulator 74
Then, for example, when a liquid crystal light valve is used, the electric field is applied in the liquid crystal due to the resistance change of the photoconductor layer in the light valve according to the light intensity distribution of the incident Fourier transform image, and the liquid crystal molecular alignment changes. Then, the Fourier transform image of the representative pattern is written.

On the other hand, the coherent light beam 76 emitted from the semiconductor laser 75 is collimated by the collimator lens 77 and enters the image display means 15. The input image is drawn on the image display unit 15 in the transmittance distribution of the intensity pattern, the light flux 76 is transmitted, and the input image is read.
The modulated light beam 76 passes through the polarization beam splitter 94, is reflected by the mirror 96, and is an optical writing type spatial light modulator 93 arranged on the front focal plane of the Fourier transform lens 16.
Incident on. At this time, in order to maximize the transmittance in the polarization beam splitter 94, the light emitted from the image display device 15 may be in a linear polarization state (usually a p polarization state) corresponding to the total transmission of the polarization beam splitter.

In the initial state, the semiconductor laser 10
1 emits light and is collimated by the collimator lens 102, and the half mirror 95 and the dove prism 103
After that, light is incident on the entire writing surface of the optical writing type spatial light modulator 93. As a result, in the case of a liquid crystal light valve, for example, an electric field is applied to all liquid crystals of the photoconductor layer in the light valve, the liquid crystal molecule arrangement is entirely changed, and writing is performed over the entire surface. At this time, instead of the semiconductor laser 101, an incoherent light source such as a metal halide lamp may be used. In addition to the full writing by light, the same effect can be obtained by changing the driving state of the optical writing type spatial light modulator 93 so that the state of the liquid crystal is changed even without writing light.

The light beam 76 that has entered the reading surface of the optical writing type spatial light modulator 93 is reflected and again enters the polarization beam splitter 94 through the mirror 96. At this time, the liquid crystal structure is formed over the entire surface. Since it has changed, the polarization state of the light beam incident on the readout surface of this modulator changes to the polarization (usually s-polarized light) orthogonal to that before the change, reflected by the polarization beam splitter 94, and the other polarization. It enters the beam splitter 78. That is, the input image displayed on the image display device reaches the polarization beam splitter 78 as it is.
Also in the polarization beam splitter 78, the light flux 76 is reflected in the same manner as described above, enters the Fourier transform lens 16, and is transmitted. After passing through the Fourier transform lens 16, the Fourier transform image of the input image is incident and reflected on the read surface of the optical writing type spatial light modulator 74 arranged on the rear focal plane.

At this time, the Fourier transform image of the representative pattern is written in the optical writing type spatial light modulator 74 as described above, and the structure of the liquid crystal is changed. The polarization state of the portion of the Fourier transform image of the input image which has hit changes according to the pattern written in the optical writing type spatial light modulator 74,
The reflected light passes through the Fourier transform lens 16 again and is incident on the polarization beam splitter 78 again. Here, the linearly polarized light changed according to the Fourier transform image of the representative pattern written in the optical writing type spatial light modulator 74, and the component orthogonal to the state before the change (usually the p-polarized light component) is transmitted through the polarization beam splitter 78. The optical writing type spatial light modulator 93 which is transmitted as it is, passes through the half mirror 95 and the Dove prism 103, and is arranged on the front focal plane of the Fourier transform lens 16 described above.
The output image is incident on the writing surface of. Here, the half mirror 95 is used to capture an output image, and the Dove prism is used to correct the inversion of the image vertically or horizontally.

At this time, if the emission of the semiconductor laser 101 is stopped, the light beam entering the photo-writing spatial light modulator 93 is only the above-mentioned output image, and the output image is written according to the above-mentioned principle. Here, when the entire surface writing is first performed only by controlling the driving method of the modulator without using light, the driving state is changed to the state of the liquid crystal in the portion where the writing light is incident. An output image is similarly written by changing to such a normal drive.

This output image is an image obtained by the cross-correlation operation between the input image and the representative pattern. Specifically, it is an image in which an input image having a small correlation with the representative pattern is deleted, and a portion having a high degree of correlation with the representative pattern in the input image is output without decreasing the contrast and In the portion having almost no correlation with the representative pattern, the image is output with almost no contrast, and an image in which a pattern similar to the representative pattern remains is output.

Thereafter, the light beam 76 modulated by the intensity pattern of the input image on the image display device 15 enters the optical writing type spatial light modulator 93 and is reflected. At this time, the optical writing type spatial light modulator 93. As described above, since the output image is written in and the structure of the liquid crystal is changed, the portion of the input image corresponding to the portion of the written output image is written in the optical writing type spatial light modulator 93. The polarization state changes according to the pattern and enters the polarization beam splitter 94 again via the mirror 96. Here, the linearly polarized light that has changed according to the output image written in the optical writing type spatial light modulator 93 and is orthogonal to that before change (usually the s-polarized light component) is reflected by the polarization beam splitter 94, and After that, it is reflected by the polarization beam splitter 78 as described above,
It is incident on the Fourier transform lens 16.

Like the output image, the pattern incident on the Fourier transform lens 16 is an image in which those having a small correlation with the representative pattern in the input image are removed, and the degree of correlation with the representative pattern in the input image. The part where the contrast is high is output without losing the contrast, and conversely, the part where the degree of correlation with the representative pattern is little is the image with almost no contrast and is the image in which a pattern similar to the representative pattern remains. .

Therefore, similarly, if the series of processes described above is repeated an appropriate number of times, the portion of the input image having a low degree of correlation with the representative pattern disappears, and only the pattern having a higher degree of similarity with the representative pattern ends. The remaining image is output.

The final output image is partially deflected by the half mirror 95, and the output image is formed on the screen 21 arranged on the front focal plane of the Fourier transform lens 16. This image is picked up by the image pickup device 22 and taken into the computer 18. After this, the image processing device 23
Thus, the position of the pattern remaining in the output image is detected, and the region is cut out from the input image based on the position information to form the extracted pattern. That is, all the patterns belonging to the large category having the common feature are extracted from the input image as the extraction patterns. Classification of the extraction pattern into smaller subcategories will be performed in the subsequent processing. Since the characteristic amount detecting means and the determining means after this are the same as those in the first to third embodiments, description thereof will be omitted.

Image display means 15 used in the apparatus of FIG.
A liquid crystal panel or the like, which is a general electric address type spatial light modulator, can be used for 53. A CCD camera, an image pickup tube, or the like can be used as the image pickup device 22. In addition, although the optical writing type spatial light modulator is used in the fifth embodiment, as such a spatial light modulator, a liquid crystal panel combined with a photoconductive layer is generally used.
Those using nematic liquid crystal are on sale.

In the optical writing type spatial light modulators 74 and 93, the Fourier transform image of the representative pattern and the output image are written in the gray level or the binary image. The effect of suppressing a decrease in the contrast of the extracted pattern or the like comes out. As such a spatial light modulator, one using a ferroelectric liquid crystal can be used. In particular, a spatial light modulator using a ferroelectric liquid crystal is suitable not only for binarizing an image but also for high resolution and high contrast in principle.

Further, as an optical address type spatial light modulator, BSO (bismuth silicon oxide; Ba1
It is also possible to use a medium having a photo-induced refractive index change such as 2SiO20). In particular, in the case of BSO material, visible light of short wavelength with high sensitivity is used as writing light,
Writing and reading can be performed at the same time by using light in the wavelength range from red to infrared, which has low writing sensitivity, as the reading light. However, the BSO material can be used as a reflection type by using a reflection plate, but is usually used as a transmission type. In this case, it is necessary to change the incident direction of the reading light incident on the BSO material, but this is not an essential means of the present invention.

Sixth Embodiment In the sixth embodiment, since the feature parameters are obtained by using the feature amount detecting means such as the optical correlation calculating device, the same as in the above-described first to fifth embodiments, the discrimination judgment is performed. Only the part of the device 31 will be described. In the sixth embodiment, a fuzzy logic arithmetic device is used as the identification determination device 31. This is done by creating a membership function based on a plurality of feature quantities of the extraction patterns belonging to the category obtained by the feature quantity detecting means, and then comparing the feature quantity of the unknown extraction pattern with the above membership function. A membership value is obtained for each characteristic parameter, and the extraction pattern is determined based on the obtained plurality of membership values.

First, the feature amount is acquired by the feature amount detecting means, but it is desirable to perform the extraction for a plurality of times for each extraction pattern in order to reduce the influence when the output of the feature amount detecting means is unstable. Therefore, the average value and the standard deviation amount, which are one of the representative values of the output values of the characteristic parameters, are calculated from the data of a plurality of times, and the membership function with respect to the output intensity is made into a trapezoidal shape. The membership value is 1 up to the range of the height, the membership value 0 is 3 times the standard deviation, and the portion corresponding to the hypotenuse of the trapezoid is from 1 to 3 times the standard deviation.

12 to 14 are membership function graphs of a small category to which the pattern A belongs, in which the outputs of the characteristic parameter 1 and the characteristic parameter 2 of a certain extracted pattern A are expressed as membership values. In this way, membership functions for each division pattern are created for all small categories and stored in the identification and classification device 31. Next, an unknown extraction pattern is shown instead of the pattern A, the output of each component of the characteristic parameter is obtained, and the output is compared with the membership function of the small category created in advance as described above.

For example, the membership functions of the small categories A, B, and C are shown in FIGS. 12, 13, and 14, respectively. At this time, the output of the characteristic parameter of the unknown extraction pattern is the reference pattern. It is assumed that the output is s point for 1 and the t point for reference pattern 2. Table 1
Is a graph showing to what extent these s and t points have membership values with respect to the membership function of each pattern (how much they belong to each small category).

[0114]

[Table 1]

AND from the membership values shown in Table 1
When the calculation is performed, the minimum membership value of each extracted pattern is taken, and the membership value for an unknown pattern is 1 in the small category A, 0 in the small category B, and 0 in the small category C. Therefore, it can be determined that the unknown extraction pattern is the small category A.

Further, the arithmetic mean value of the membership values in Table 1 can be used as a criterion. That is, in this case,
Since the small categories A, B, and C are 1, 0.5 and 0.2, respectively, for the unknown extraction pattern, the degree of each pattern can be estimated. However, when the arithmetic mean is used, an answer may be given even if there is no value of the characteristic amount of a certain characteristic parameter. Therefore, although the above-mentioned operation has less error, the answer is given by AND operation. It is meaningful as an auxiliary operation when it is difficult. That is, it is significantly different from the state when the average value is taken for some reason, 1
Membership value is 0 with only one feature parameter component
And, when it is 1 with other reference patterns, AND
You will not be able to give the correct answer by calculation. In this case, it is possible to use the arithmetic mean as the second candidate.

Although the membership value is trapezoidal in the sixth embodiment, any value may be used as long as it is a so-called convex fuzzy set in which the membership function has no depression. Further, such an arithmetic unit can be configured using a Neumann type computer or a hardware arithmetic unit.

Seventh Embodiment In the seventh embodiment, up to the point where the characteristic amount is obtained by using the characteristic amount detecting means such as the optical correlation calculating device,
Since it is the same as the above-described first to third embodiments, only the identification determination device 31 will be described.

Depending on the distribution of the small categories, the judgment by the discrimination judging device 31 may be logically judged as described in the first embodiment, but in a plurality of small categories, all the characteristic parameters can be taken. When the ranges stochastically overlap, it is effective to make a judgment using a hierarchical neural network represented by a single-layer or multi-layer perceptron. FIG. 15 shows an identification determination device 31 using a hierarchical neural network.
The block diagram of one Example is shown.

In such a hierarchical neural network, the feature amount for each feature parameter of the extraction pattern belonging to each small category is input to each node of the input layer, and the desired output indicating the small category at that time is presented. By performing the extraction pattern multiple times for all the extraction patterns, the connections in the neural network are organized so that the neural network outputs a small category suitable for the input extraction pattern. Categories that cannot be completely separated only by themselves can be determined. The feature amount of each feature parameter of the extracted pattern detected by the feature amount detecting means is input to each node of the input layer, and the unknown pattern is identified by the output state of the node of the output layer.

The identification judging device 31 includes a computer 18
The determination function may be realized by software by using, or the same function may be realized by hardware by an electronic circuit. Also, the learning algorithm is not particularly limited, and for example, in the case of a multi-layer perceptron structure having three or more layers, the well-known back propagation method based on the generalized delta rule can be used.

In addition to the above, the neural network is, for example, a network for creating a small category in a hypersphere in a feature space or a network for weighting a distance from a reference data point. There are functions, etc., and these may be used.

[0123]

With the configuration of the optical pattern recognition and classification apparatus of the present invention as described above, it is not necessary to use a matched filter in which interference fringes such as holograms, which must be produced for each pattern to be extracted, are drawn. Moreover, the alignment of the optical system becomes relatively easy because of this. Also, because the processing is performed optically on the entire image,
It is not necessary to scan the pattern to be extracted, which enables pattern extraction and recognition classification in a short time. Furthermore, not only the pattern that is exactly the same as the pattern to be extracted, but also a pattern having a certain degree of correlation can be easily extracted at the same time using a simple filter, and further finer classification can be performed thereafter. , A pattern can be extracted from an image such as a natural image containing many patterns of objects and characters, and can be recognized and classified.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram showing a configuration of an embodiment of an optical pattern recognition classification device of the present invention.

FIG. 2 is a schematic configuration diagram of an optical system of a conventional matched filtering method.

FIG. 3 is an explanatory diagram showing an example of an input image according to the present invention.

FIG. 4 is an explanatory diagram showing an example of a representative pattern according to the present invention.

FIG. 5 is an explanatory diagram showing an example of an output image according to the present invention.

FIG. 6 is an explanatory diagram showing an example of an extraction pattern according to the present invention.

FIG. 7 is a schematic configuration diagram showing the configuration of another embodiment of the optical pattern recognition classification device of the present invention.

FIG. 8 is a schematic configuration diagram showing the configuration of another embodiment of the optical pattern recognition classification device of the present invention.

FIG. 9 is a schematic block diagram showing a partial configuration of another embodiment of the optical pattern recognition classification device of the present invention.

FIG. 10 is a schematic configuration diagram showing a configuration of a part of the apparatus of FIG.

FIG. 11 is a schematic block diagram showing a partial configuration of another embodiment of the optical pattern recognition classification device of the present invention.

FIG. 12 is an explanatory diagram showing a membership function created based on the output of the characteristic parameter of the extraction pattern of one small category according to the present invention.

FIG. 13 is an explanatory diagram showing a membership function created based on the output of the characteristic parameter of the extraction pattern of another small category according to the present invention.

FIG. 14 is an explanatory diagram showing a membership function created based on the output of the characteristic parameter of the extraction pattern of another small category according to the present invention.

FIG. 15 is an explanatory diagram showing an example of a hierarchical neural network as an example of the identification determination device of the optical pattern recognition classification device of the present invention.

[Explanation of Codes] 1 ... Pattern extracting means 2 ... Optical correlation calculating means 3 ... Discriminating means 11 ... Laser 12, 12a, 12b, 12c, 12d, 72, 76,
80, 84 ... Luminous flux 13 ... Beam expander 14, 14a, 14b, 14c, 14d, 95 ... Half mirror 15, 25, 53, 58 ... Image display means 16, 20, 26, 29, 54, 59 ... Fourier transform lens 17, 27 ... Spatial light modulator 18 ... Computer 19, 28 ... Fourier transform image output means 21, 55, 60 ... Screen 22, 56, 61 ... Imaging means 23, 62 ... Image processing device 24, 91, 96 ... Mirror 30 ... light receiving means 31 ... identification determination device 41 ... matched filter 51, 52 ... Fourier transform lens array 63 ... light receiving means array 71, 75, 79, 83, 101 ... semiconductor laser 73, 77, 81, 85, 102 ... collimator lens 74, 82, 93 ... Optical writing type spatial light modulator 78, 86, 94 ... Polarizing beam splitter 90 ... Force image 92, 97 ... imaging lens 103 ... Dove prism

Claims (9)

[Claims] 1. A pattern belonging to a desired large category consisting of a set of one or more small categories including one or more recognition-desired patterns is extracted from an input image (hereinafter referred to as an input image). A pattern extraction unit and an amount (hereinafter referred to as a feature amount) representing each feature of the pattern extracted by the pattern extraction unit (hereinafter referred to as an extracted pattern) are detected using an optical system that performs a correlation or similar operation. The pattern extraction means, and a determination means that determines the small category to which the extraction pattern belongs based on one or more characteristic quantities detected by the feature amount detection means. , A first light source that emits a first coherent light beam, and an input image that is displayed with intensity transmittance or reflectance distribution. A first image display unit for giving an intensity distribution to a transparent light beam; a first Fourier transform lens for optically Fourier transforming the intensity distribution of the light beam transmitted or reflected by the first image display unit; Patterns that have common characteristics to all the patterns included in the large categories (hereinafter referred to as representative patterns)
A Fourier transform image output means for outputting a Fourier transform image of, and a Fourier transform image of the representative pattern, which is arranged on the Fourier transform surface of the first Fourier transform lens, as an intensity transmittance or a reflectance distribution. And a first spatial light modulator that further gives an intensity distribution to the first light flux, and an intensity distribution of the light flux transmitted or reflected by the first spatial light modulator is further optically Fourier transformed. A second Fourier transform lens, an imaging unit that captures an output image obtained by performing Fourier transform by the second Fourier transform lens, and the output image as an input image of the first image display unit. Image returning means for displaying, and position detecting means for detecting a pixel having an output value of a certain value or more in the output image and detecting a position of a group of a certain number or more of the pixel. An area cutting means for cutting an area of an appropriate size from the input image on the basis of the output from the position detecting means, and the feature amount detecting means is a second coherent device. A light source that emits a luminous flux, and a second image display unit that displays one or a plurality of reference patterns with an intensity transmittance or reflectance distribution to give an intensity distribution to the second coherent luminous flux. A third Fourier transform lens for optically Fourier transforming the intensity distribution of the light flux transmitted or reflected by the second image display means, and a second Fourier transform image of the extracted pattern obtained by the pattern extracting means. Fourier transform image output means of the third Fourier transform lens, and the Fourier transform image of the extracted pattern arranged on the Fourier transform surface of the third Fourier transform lens. Is a reflectance distribution for display, and a second spatial light modulator that further gives an intensity distribution to the second light flux, and an intensity distribution of the light flux transmitted or reflected by the second spatial light modulator A fourth Fourier transform lens for further performing a Fourier transform, and a light receiving unit for detecting the light quantity of the image obtained by the Fourier transform by the fourth Fourier transform lens in correspondence with the one or more reference patterns. And an optical pattern recognition classification device. 2. A pattern belonging to a desired large category consisting of a set of one or more small categories including one or more patterns to be recognized and classified is extracted from an input image (hereinafter referred to as an input image). A pattern extraction unit and an amount (hereinafter referred to as a feature amount) representing each feature of the pattern extracted by the pattern extraction unit (hereinafter referred to as an extracted pattern) are detected using an optical system that performs a correlation or similar operation. And a discriminating means for discriminating the small category to which the extracted pattern belongs based on one or more characteristic amounts detected by the characteristic amount detecting means. A first light source that emits a first coherent light beam, and an input image that is displayed with intensity transmittance or reflectance distribution, A first image display unit for giving an intensity distribution to the light beam; a first Fourier transform lens for optically Fourier transforming the intensity distribution of the light beam transmitted or reflected by the first image display unit; Patterns that have common characteristics to all patterns included in the category (hereinafter referred to as representative patterns)
A first Fourier transform image output means for outputting the Fourier transform image of the first Fourier transform image, and the Fourier transform image of the representative pattern arranged on the Fourier transform surface of the first Fourier transform lens into an intensity transmittance or a reflectance distribution. And a first spatial light modulator for giving an intensity distribution to the first light flux, and an intensity distribution of the light flux transmitted or reflected by the first spatial light modulator is further optically Fourier-transformed. A second Fourier transform lens, an image pickup means for picking up an output image obtained by performing a Fourier transform by the second Fourier transform lens, and the output image is displayed again as an input image of the first image display means Image returning means, and position detecting means for detecting a pixel having an output value of a certain value or more in the output image, and detecting the position of a group of a certain number or more of this pixel, And a region cutout unit that cuts out a region having an appropriate size from the input image with reference to the output from the position detection unit, and the feature amount detection unit is a second coherent unit. A light source for emitting a luminous flux, and a second image display means for displaying one or a plurality of reference patterns with an intensity transmittance or reflectance distribution to give an intensity distribution to the second coherent luminous flux. A lens is arranged corresponding to the position of the one or a plurality of reference patterns, and the intensity distribution of the light flux transmitted or reflected by the second image display means is optically Fourier-transformed for each reference pattern. Fourier transform lens array, second Fourier transform image output means for outputting a Fourier transform image of the extracted pattern obtained by the pattern extracting means, and the extracted pattern An image of the Fourier transform image, the first
An image processing device arranged in parallel at a position corresponding to each lens of the Fourier transform lens array of, and arranged on the Fourier transform surface of each lens of the first Fourier transform lens array, and created by the image processing device. A parallel spatial image of the Fourier transform image of the extraction pattern is displayed with an intensity transmittance or a reflectance distribution, and a third spatial light modulator for further imparting an intensity distribution to the second light flux, and each lens has one of the above or A second Fourier transform lens array which is arranged corresponding to the positions of a plurality of reference patterns, and which further optically Fourier transforms the intensity distribution of the light flux transmitted or reflected by the third spatial light modulator; The light quantity of the third output image obtained by Fourier transform by each lens of the Fourier transform lens array of 2 corresponds to the one or more reference patterns. Light receiving means for detecting for, and a, optical pattern recognition classifier, characterized in that. 3. At least one of the first or second Fourier transform image output means uses a light source that emits a coherent light beam, and the representative pattern or the extracted pattern, which is an intensity transmittance or reflectance distribution. Third image display means for displaying and giving an intensity distribution to the light flux, and a fifth Fourier transform lens for optically Fourier transforming the intensity distribution of the light flux transmitted or reflected from the third image display means. The optical system according to claim 1 or 2, further comprising: an imaging unit configured to capture a Fourier transform image of the representative pattern or the extracted pattern obtained by the fifth Fourier transform lens. Pattern recognition classifier. 4. The method according to claim 1, wherein at least one of the first, second, and third spatial light modulators is a first optical writing type spatial light modulator. 2. The optical pattern recognition classification device according to any one of 1. 5. The image forming lens, wherein the first image display means forms an input image on a writing surface of the second optical writing type spatial light modulator and the second optical writing type spatial light modulator. An optical pattern recognition classification device according to any one of claims 1 to 4, characterized in that it comprises a system. 6. A Fourier transform image of the representative pattern or the extracted pattern is provided on at least one of the first, second, and third spatial light modulators by an appropriate threshold value in substantially real time. The optical pattern recognition classification device according to any one of claims 1 to 3, wherein the optical pattern recognition classification device is displayed as a binarized image. 7. At least one of the first or second spatial light modulator of the optical writing type performs an operation of binarizing an image written from a writing surface by a threshold value and displaying the image. The optical pattern recognition classification device according to claim 4 or 5, characterized in that. 8. The discrimination means, for each small category,
The feature quantity of the extraction pattern belonging to this small category is acquired multiple times, and the representative value of the obtained output set is used as a reference,
A membership function of the extraction pattern is created in advance for each of the components of the extraction pattern and the feature amount (hereinafter, referred to as feature parameter), and the feature amount of the extraction pattern extracted from the input pattern is used for each small category. A membership value is calculated for each characteristic parameter based on each membership function, and the average value or minimum value of the obtained membership values is set as the degree of belonging to each small category of the extraction pattern. The optical pattern recognition classification device according to any one of claims 1 to 7, characterized in that the optical pattern recognition classification device comprises an arithmetic means for further determining an extraction pattern. 9. The discrimination means is a neural network which inputs each feature quantity acquired for each of the extraction patterns and has an output showing a small category of the extraction pattern, and discriminates the extraction pattern from this output. The optical pattern recognition classification device according to any one of claims 1 to 7, characterized in that.