DE2354661C3 - Device for the optical identification of geometrical figures - Google Patents

Description

60

The invention relates to a device for the optical identification of items contained in an input structure geometric figures according to the preamble of claim 1.

Such a device has become known from Γ-OS 20 45 781, for example. The various holograms superimposed on one another in the filter of this known device are each the Fourier transform of a discrete geometric figure, modulated by means of the respective reference bundle. These filters are therefore matched filters with the disadvantage that a special hologram is required for each geometrical figure to be identified, i.e. at least N holograms for N geometrical figures to be distinguished, in the specific example described in the older document even 2N holograms. The superposition of so many holograms creates difficulties when a great many geometrical figures are to be distinguished from one another. Already with 50 2: u superimposed holograms according to z. B. 50 zn differentiating geometric figures, it is difficult to produce a filter superimposed from these many holograms, which has the necessary selectivity to reliably differentiate the figures to be identified from each other. One way out of this would be to accommodate only some of the required holograms in a first filter and the remaining holograms in further filters. However, this would require several filters to be exposed to the light bundle influenced by the input structure either simultaneously or one after the other. With simultaneous evaluation by several filiers, however, the effort then increases considerably, while with evaluations taking place one after the other, the evaluation speed is necessarily reduced. A further disadvantage of these matched filters is that a certain set of holograms superimposed on one another in the filter can only be used for a population of discrete geometric figures that was predetermined at the time the filter was manufactured. If it turns out during use of the device that a further geometric figure should also be identifiable, a completely new filter must be manufactured.

An arrangement for identifying geometric figures has become known from DT-OS 19 45 085, in which point holograms superimposed on a plate are used as pupil multipliers, multiplied by the light bundle influenced by the input structure to a corresponding number of adapted ones Direct filter. These are located on a carrier that is separate from the carrier of the point holograms side by side, so that no superposition of the matched filters is necessary. However, it is here required, a number of point holograms corresponding to the number of matched filters to overlay. In addition, the space requirement is due to the side-by-side arrangement of the matched filters as well as the visual effort for further guidance of the bundle considerably if it is a large one Number of geometrical figures to be identified.

Furthermore, from the publication IBM Technical Disclosure Bulletin Vol. 14, No. 8, Jan. 1972, page 2503 has become known from a certain geometrical figure, ι. B. a stylized number "8" to transfer only figure elements to a filter. However, these filters are not superimposed holograms, let alone point holograms. According to this older document, because of the spatially separated arrangement of the adapted filters for the figure excerpts, a considerable amount of space and effort is necessary for the separate guidance of the bundles to the various filters. The use of the figure known from this last-mentioned prior publication

lü

IS

. Extracts as image templates for the production of the other superimposed holograms based on the model ί DT-OS 20 45 781, however, was not an obvious choice, since with a

There are no advantages associated with mere transfer

er. The number of ί to be superimposed

rt logramme would vary according to the number of

Excerpts J ^e geometrical figure only vervielfa-

. which in no way improves the selectivity

could be. This way therefore had to be considered Appear promising.

The invention has the task of avoiding the disadvantages of the known arrangements m

A device to provide a hen for identifying geometrical ^U figures, while maintaining a small and multiple type, and a fast mode of operation allows for increased selectivity, increased J ₁ jjje certainty with which to

■ Identifying geometric figures can be distinguished from one another _$ - _m ^ _unc j, _{although aucn} ^ ann, _{if very}

Many such figures should be distinguished, although it is not even necessary that they be all of one another the figures to be distinguished are fixed from the start.

The invention solves the problem posed primarily by the measures specified in claim 1. Further advantageous embodiments within the scope of the task at hand are provided by the subordinate claims marked.

In the following description, exemplary embodiments of the invention are based on the drawings explained. It shows

Fig. 1 is a schematic representation of an optical System for producing a point hologram used in the device for optical identification

Fig.2 is a schematic representation of the optical System for use in an identification device using a multi-point hologram

3 shows an example of the Walsh-Hadamard matrix,

pig 4 the way in which the Walsh-Hadamard matrix is in a Walsh function is transposed, where a development formula of the Walsh function in two-dimensional Room is also shown,

Fig. 5 is a view of spatial filters that use the Represents Walsh function in two-dimensional space,

6 shows a schematic representation of the optical Another embodiment system comprised in an identification device using a composite multi-point hologram used, is provided.

1 shows an embodiment of an optical system for producing a multifaceted point hologram, that used in the identification device

A He-Ne laser (wavelength 6328 Å) is used as source 1 for the coherent light. The one from the coherent light source 1 outgoing light beam passes through a shutter ta and a semi-permeable Mirror 2 before it is transformed into another parallel light beam by lenses 3 and 4, which falls on a small opening 6. Under the condition that the laser beam is controlled in such a way that If its diameter is smaller than that of the small opening, the lenses 3 and 4 can be omitted. The Light that has passed through the small opening 6 is Fourier transformed by a lens 7 to an intensity of 6s distribution on a photosensitive layer carrying surface 9 of a carrier, for. B. a drying lath is arranged perpendicular to the optical axis to generate. This intensity distribution represents that Fourier transformed spectrum of the small aperture 6 On the other hand, the light reflected by the semi-transparent mirror 2 is incident on a reflective one Mirror 10 and serves as a reference light. The reference light is through the reflecting mirror 10 on the Surface 9 of the dry plate and is directed into a parallel light beam, the reference beam through a Lens pair 11 and 12 converted. -It then falls at an angle θ with respect to the normal to the surface 9 of the drying plate. A mask 8 is in contact with the surface 9 of the drying plate, its permeability determines the transmission efficiency of the optical filter. After the dry plate in this way has been exposed, the mask 8 is opposed to another with a different transfer effect exchanged. The dry plate is then exposed a second time, the angle of incidence of the Reference bundle is changed. This described procedure is carried out several times in succession, The frequency depends on the number of filter functions required.

An arrangement for varying the angle of incidence of the reference beam is now referred to on F i g. t described. As soon as the semi-transparent rotatable mirror 2 is rotated by a drive 13, until it assumes the somewhat inclined position designated by 2 ', the semitransparent mirror 2 becomes the light reflect in the direction of the dashed line. The reference light, its exit angle in this way has been changed is then reflected back through a reflecting mirror 10 '. This one became like this shifted to direct the reflected reference light onto the dry plate and then into a parallel one Reference beam to be transformed by the lenses 11 'and 12' before it is incident on the drying plate. on in this way the angle of incidence of the reference beam on the dry plate is determined by the rotation of the semitransparent mirror 2 and the displacement of the reflecting mirror 10 and the pair of lenses 11 and 12 changed.

After developing a dry plate exposed in this way, a multiple point hologram superimposed from several point holograms is obtained on the dry plate. If, in the production process of such a point hologram, a scattering plate 5 is arranged at the small opening 6, a point hologram recorded with diffuse light is obtained which, compared to a conventional point hologram, has an improved overdetermination and an almost uniform scattering efficiency over its entire surface and also the recording with light allows a reduced intensity difference, which results in an improved linearity of the recording. . The transmission coefficient of a multi-point hologram thus produced could be expressed in the following way. If we take the transmission coefficient of the small opening 6 with φ the plane reference wave with

A exp ( -jkx sin θ I) and the amplitude transmission function of the mask 8 with

1.2,

.N)

then the light intensity distribution obtained on the surface of the dry plate can be l (x)

laughed after exposing the dry plate N times will be through

/ (χ) = Σ {Wf (x) lN (A ² + \ F {t (£)} \ ² )

ι = I

Light G (x) transmitted through the filter 19 is described by

4- / 4F {«(£)} exp (// ex sin«,)

+ AF {t (£)} exp (- // ex sin (9,)]}.

τ (χ) of the obtained point

The transmission coefficient
lolograms then reads:

r (x) =

'(X) IN (A ² +1 F {t (S)

+ AF {t ({)} exp (// ex sin «,)

+ AF {t (£)} exp ( -jkx sin «,.)]}. (2)

where | is the coordinate on the surface of the small opening, χ the coordinate on the dry plate, θ the angle of incidence of the reference beam on the dry plate, k = 2π / λ, where λ is the wavelength of the light. r _b is the given transmission factor, β is a constant and F is an operator of the Fourier transform. For a better overview, the formulas (1) and (2) are written in such a way that they only correspond to one-dimensional structures.

Fig.2 shows an optical system as it is used for the identification of two-dimensional geometric Figures using a multi-point hologram is used. That of the coherent Light source 14 outgoing light is through a pair of lenses 15 and 16 in a further parallel Converted light bundle and illuminates a two-dimensional input structure 17. A lens 18 leads a Fourier transformation of the two-dimensional input structure 17, the Fourier transformed The spectrum of the input structure is formed in the back focal plane of the lens 18, where a filter 19 is located is located, the superimposed from each other, each modified by a mask upstream of the recording Point holograms of the type described above, that is to say consists of a multiple point hologram. Of the light emanating from the filter 19, only the light diffracted into the first order is transmitted Lenses (201, 202, ..., 20N) immediately behind the Filters 19 are arranged, combined in the corresponding focal planes of these lenses and subsequently detected by the assigned photoelectric converters (211, 212,..., 21 / V). Everyone this transducer is arranged so that its optical axis corresponds to that of each exposure Production of a point hologram used light is aligned. Which into an electrical signal converted intensity of the diffracted light is used as a quantity to determine the properties of the Information can be found in the input structure. The analysis of the structure is carried out by subsequent electronic Circuits (not shown) are performed.

One denotes the transmission coefficient of the input structure of the optical system, as shown in FIG. 2, with S (I) (ξ is the coordinate of the surface of the input structure), then the light filling the filter 19 corresponds to the Fourier transform of Sß; and is represented by F (S (^) J. The -τ (χ) χ

A plurality of lens transducer groups are arranged behind the filter 19 so that the angle between each of their optical axes and the main optical axis of this device corresponds to the angle of incidence of each reference beam which was used in each exposure to produce a point hologram. In this way, each photoelectric converter detects the intensity of the light diffracted in the first order. If the amplitude of the first order diffracted light emitted by the multi-point hologram at the angle Θ; is denoted by G (x) , and if the photoelectric converters integrate all of the incident light, then the intensity of the received light /, - can be expressed as follows

■ x ·

H ² A ² JW? (X) | F {S (i)} | ² dx. (3)

Equation 3 describes the light information of the Fourier-transformed spectrum of the input structure, which is detected by the mask, whose transmission function of the light intensity is described by W? (X) , by a photoelectric converter arranged at the angle Θ. It can thus be seen that the point hologram produced in this way acts as a filter, the intensity-transmittance function of which for the diffracted light is described by W, ² (x) . The arrangement of the masks used in the production of the point hologram determines the mode of operation of the resulting optical filter. A multi-point hologram is used as a filter, which is used for the orthogonal transformation of the information in the input structure.

Typical examples of such a filter have a filter that is made up of a variety of filters for development composed of the Walsh function. The orthogonal transformation can be made easier and more effective Way to be used for information processing. Various known methods such as Fourier transform and Walsh-Hadamard transformation can be used for information processing will. In contrast to the Fourier transform, which requires multiplication in an analogous way needs to be performed, the use of an orthogonally transforming filter allows the processing the input information in a digital way only by addition and subtraction. Because of this, the Walsh-Hadamard transformation, which is a higher

5S processing speed enables, especially observed. The aforementioned orthogonal transformation is a method that was originally intended for use with Optical readers have been developed that allow finding the properties of the input structure

do exclusively with electronic data processing systems has been made. To make such a complicated processing in easier and faster To achieve this way through the use of optical elements, only one serves in the present case Multipoint hologram.

In the following, the method with a Walsh-Hadamard transformation is used as an example of the orthogonal transformation described.

+ η Ν + H _n N = 2 " H _n = 2 2 + H _n - H _n η = 1,2, .. "i "2"

The Walsh-Hadamard transform of an input structure represents the information of the input structure through a matrix, one of which is already known Walsh-Hadamard matrix (Fig. 3). To make such an orthogonal transformation in the plane of the Fourier transformed spectrum of the To perform two-dimensional input structure, a Hadamard matrix is first made with 4 Walsh functions transposed, see the functional illustration in FIG. 4, in which the angular coordinate of the spectral plane is used as the abscissa. In this way, an optical filter with a Walsh function is obtained. In general, the Walsh-Hadamard matrix is defined by

3 and 4 represent the Walsh functions {n = 2). The real optical filters - these are representations of the Walsh functions in two-dimensional space - are formed by a group of sector-like openings Φι, ..., Φι , as shown in Fig.5. The Walsh function is a two-valued function that takes either the value 1 or 0. The light can pass through the white sections shown in FIG. 5, while it cannot penetrate the hatched sections. It is evident that the multi-point hologram (which has been fabricated in the manner described above using the masks having the openings shown in Figure 5) acts like a plurality of Walshen type optical filters.

The production of a multi-point hologram is explained in detail. First, the mask Φ \ is arranged in the plane of the mask 8 (FIG. 1), and a dry plate is exposed, the reference beam being incident at an angle θι. The mask Φ \ is then replaced by a mask Φ ₂ , and the angle of incidence of the reference beam assumes the value Θ2, whereupon the dry plate is exposed a second time. In this way, the dry plate is exposed repeatedly, precisely as often as there are masks, until a multi-point hologram is obtained. If a multi-point hologram produced in this way is arranged as a filter 19 in the optical system (FIG. 2) and an input structure 17 is introduced at the predetermined point, the Fourier-transformed spectrum of the input structure is obtained on the surface of the multi-point hologram serving as filter 19 . The / th photoelectric converter whose optical axis is at the angle θ ₍ detects the imaging intensity, which is expressed as

/ _w . = CJ 0,. | F {S (i)) | ² dY

where / = 1, 2, ..., 7 and C are a constant. The imaging intensity represents a quantized characteristic Represents the size obtained by the Walsh-Hadamard transform of the input structure would. It should be noted that the diameter of the round opening of the masks used in the optical system (Fig. 2) is so large that the entire Fourier transformed spectrum of the input structure could be covered. It is known that the method of two-dimensional figure identification, using such an S Walshen type optical filter has the advantage of a to allow a safe method of searching for the characteristic properties. One is both from a parallel shift of the input structure in its plane, as well as an enlargement or Reduction of the input structure independently. Besides the Walsh-Hadamard filters described above, various other types of optical filters are known. These are filters for Fourier expansion and a modified Walsh-Hadamard filter suitable for orthogonal transformation. As mentioned earlier, any of these filters can be used as the optical filter in the invention.

From these considerations it can be seen that the optical filter, which consists of a multi-point hologram is constructed, is a filter which, although it consists only of a single optical element, has the effect of a variety of conventional optical filters.

The discussion that follows is a modified method of identifying geometric figures that uses a composite multi-point hologram as an optical spatial filter.
A composite multi-point hologram consists of a number of individual point holograms which are arranged next to one another on a dry plate. Each of these point holograms is a multiple point hologram that is recorded on a small area (e.g. 1 cm ² ) of the dry plate. Proceeding in this way, a single composite multi-point hologram allows an even larger amount of input structure information to be processed compared to a single multi-point hologram. Two manufacturing methods for such a composite multi-point hologram are possible. In the production according to the first method, the optical axis of the light beam emerging from the small opening is displaced in that the small opening is displaced step by step in its plane. The shift is carried out along a predetermined constant distance along the two Cartesian coordinate axes (ξ, η) , so that individual multi-point holograms are formed at the points illuminated by the light beam on the dry plate. In contrast to this, in the second method, the small opening and the mask are held in place while the drying plate is shifted step by step in its plane, so that a large number of multi-point holograms are formed at the locations on the drying plate which are aligned with the mask. The second method is more advantageous than the first because it can be performed with a simpler optical system.

6t) A composite multi-point hologram has the ability to simultaneously and in parallel a larger amount of two-dimensional input structure information to be processed as a single multi-point hologram. It has in particular

('5 found that the composite multi-point hologram a further increase in the selectivity for the identification of geometric figures compared to a single multiple

dot hologram. It is assumed that an input structure is broken down into a multitude of basic structures (hereinafter referred to as picture element), so that the Fourier-transformed spectrum of each picture element on an associated hologram a single multi-point hologram is formed from which, among other things, the composite Multipoint hologram is constructed and processed in the manner already mentioned. The explanation lies in the fact that the given information is detected in an input structure in the form of bits corresponding to the number of picture elements. This procedure is referenced below to the embodiment of FIG. 6 explained

A known method of making a multiple image can be used to produce a To break down the input structure into a multitude of picture elements. The multiple mapping method is a Procedure in which the light is first split into several beams by a semi-transparent mirror and in which an optical component such as a multiple microlens is then an arrangement of many small openings or a point hologram is used. A method will now be described by way of example in which a multiple microlens is used. An identification device for two-dimensional geometric Figures using a multi-point composite hologram according to the present invention has a optical system as shown in FIG. 6 is shown. The light emanating from a coherent light source 22 is expanded and transformed by a pair of lenses 23 and 24 into a parallel light beam. It falls on one Multiple microlens 25, shown as containing 9 separate microlenses closely spaced are arranged in a common plane. A variety of point light sources (or secondary Light sources), the number of which is equal to that of the microlenses used, is provided by the Microlenses formed bundled light bundles in their focal planes. The diverging bundles of light These secondary light sources are converted into parallel bundles of light that structure the input 27 illuminate. A lens 28 and a multiple microlens 29 are arranged so that the Images of the input structure 27 are formed on the surface 30 of a mask. That way will a large number of images of the input structure, the number of which equals the number of microlenses, formed on the mask surface 30, the location of these images with the plurality of secondary Light sources matches. A custom filter made up of several types of custom filters is selected, an image of the input structure is provided on the mask surface 30 at each point, so that images are extracted from the input structure. For example, these can be customized Filters can be those which allow only a desired portion of light to pass through the picture elements, but not the remainder Light. Then the images become the input structure Fourier-transformed by a multiple microlens 31 so that the Fourier-transformed spectra of the corresponding picture elements c are formed in the back focal plane of the multiple microlenses 31. The multiple microlens 31 has a plurality of microlenses, the number, arrangement and size of which are identical to those of the multiple microlines 25. A composite multi-point hologram serves as filter 32 and is so disordered that each im composite multiple point hologram containing single multiple point hologram with a Fourier transform Spectrum of each corresponding picture element is aligned. The pictures are at the end of the optical system in the same way as the two-dimensional identification device Figures using a single multi-point hologram detected. The pictures are through a large number of lens groups (33al, 33a2, ...,

.o 33aN; 3361, 3362, 336N; 33cl, 33c2 33cN) and

a variety of photoelectric conversion groups

(34al, 34a2 34a / V; 3461,3462, ..., MbN; 34cl, 34ώ,

.... 34cN; proven. N denotes a multi-point hologram which has been exposed N times, and the letters a, b, ... relate to the corresponding position of the corresponding picture elements on the hologram surface. The properties of the input structure are taken as electrical signals. Due to this method, the number of bits that can be processed from the mapping information is increased, since the information of the input structure is processed after it has been broken down into picture elements is used, thereby increasing the sharpness of separation between similar structures.

In summary, the following properties and advantages of the invention can be stated. A multi-point hologram that acts as a variety of optical filters with desired properties acts can be easily done with a variety of masks that have the same permeability effect as that desired optical filters and placed in contact with a dry plate will.

Processing the information of the two-dimensional Input structure can be used to distinguish between very similar figures despite very good selectivity by means of a small and economical device that has such a simple optical System as shown in Figs. 2 or 6, includes the features and advantages mentioned above of a multiple point hologram, a composite multiple point hologram has the additional advantage on, a larger amount of information in two-dimensional input structures at the same time in to be able to filter in parallel. In particular, when there is a large number of input structure information in the In the form of a group of picture elements, a larger number of bits can be detected In this way, the structural properties can additionally be increased selectivity with a higher Velocity can be taken, compared mn the method in which a single multiple point hologram is used.

The multiple point hologram and the composite point hologram can be a point hologram recorded with diffuse light. Wcgci do its high overdetermination and even ^one Slreuwirkung over the entire surface is the mi diffused light measured point hologram al every point a diffracted intensity proportion "of the intensity of the incoming light, fts wodurc it is possible to remove the structural properties with less error.

The inventive device for figure cutter certification that a ViclfachDunktholoErnmm or egg

<r

Containing multi-point composite hologram as an optical filter results in a much simpler one and faster way through such complex processing operations as previously using a electronic data processing system has been carried out in conjunction with an optical laser are. This is due to the fact that the device is simultaneously processing the information in the input structure in the performs two-dimensional space. The previously known light filter processes were very granular

This is complicated, since a large number of individual filtering operations had to be carried out in order to obtain a large number of Sizes, or did not have such a good selectivity. You can see that the present Invention overcomes this disadvantage and an easy-to-use, small device for character identification provides that, because of the improved selectivity, will greatly contribute to the practical application from optical readers to expand.

For this purpose 4 sheets of drawings

Claims (5)

λsi patent claims: 1. Device for the optical identification of geometrical data contained in an input structure Figures in which a filter is arranged in a light beam influenced by the input structure, that on one and the same surface of a carrier superimposed several consisting of holograms Contains filter functions that differ from one another both due to when recording set reference bundle inclinations with regard to the diffraction angle of through the holograms diffracted sub-bundles as well as with regard to the filter operations effected on these sub-bundles differs, and furthermore the light of the diffracted partial bundles on each one of several photoelectric converters falls to the electrical circuits to identify the Figures are connected, characterized that the various holograms superimposed on one another in the filter (19; 32) are point holograms are, during the recording of a photosensitive layer carrying surface (9) of the A mask (8) that is different for each reference bundle inclination is connected directly upstream of the carrier is that in the plane of the point hologram its local intensity distribution immediately before the impact on the photosensitive layer corresponding to that present in the mask (8) Function modified that is a function for performing an orthogonal transformation of the Fourier-transformed Spectrum of the input structure is. 2. Apparatus according to claim 1, characterized in that the point hologram with a diffuse Light is recorded point hologram. 3. Apparatus according to claim 1 or 2, characterized in that the mask (8) is sector-shaped forms openings. 4. Device according to one of the preceding claims, characterized in that several filters thus constructed are arranged next to one another on a carrier (32), each of which in one of several separately running, but only mutually enforcing in the input structure Bundle channels is arranged (F i g. 6). 5. Apparatus according to claim 4, characterized in that in each individual bundle channel in a plane that is optically conjugate to the input structure, one matched filter (30) for each one certain element of a geometric figure is arranged and in this a Fourier transformation supplying level each a point hologram filter is arranged.