DE19549074A1 - Method and device for the high-resolution determination of distances in the focused image of a lens-pupil system and of metric object parameters, in particular for the purpose of the precise and intelligent control of automata - Google Patents

Abstract

A process and device are disclosed for determining spatial and/or chronological intervals in the focused image or focused image sequences produced by a lens-aperture diaphragm system and/or for determining spatial or time-related object parameters (10), spectral and/or local frequency-specific object parameters, in particular for controlling automated machines and monitoring production processes. The incident electromagnetic radiation is focused by the lens-diaphragm system (10), location-specific modulation of the propagation direction, intensity, wavelength, polarisation and/or time modulation frequency of the electromagnetic rays (11) is effected by a 3-D modulation device (12) behind the lens-diaphragm system (10), the modulated electromagnetic rays are detected by a detector device (13) situated behind the 3-D modulation device (12), and the spatial and/or chronological intervals (14) in the electromagnetic rays in the 3-D modulation device (12) are ascertained by determining the modulation and/or by calculating the spatial, chronological, spectral and/or local frequency-specific object parameters.

Description

The present invention relates to a method for ge exact determination of spatial and / or temporal Ab would be in the focused image of a lens-pupil system mes and / or of spatial and / or temporal object pairs parameters such as B. speed or depth, in particular for the precise and intelligent control of Automatic machine comprising the following step:

• (a) Focusing the incident electromagnetic Rays through a lens-pupil system, as well as a front direction to perform the same.

Geometric-optical imaging cameras are known which image on photographic films or on CCD arrays. Both methods deliver spatial 3D information in flä 2D shape in the image plane. Photographic cameras are only in optical resolution Parameters and by the grain size of the film material be limits, but only storable, but not to Suitable for real-time image processing. CCD cameras are capable of relatively quick image processing, in resolution but limited by the pixel size of the CCD arrays. Holographic imaging and recording Facilities are capable of 3D spatial information tion to deliver and store; however, this only works in the coherent light. Also, the picture is always taken a 2-beam method (object and reference beam) is required does. And finally, holographic systems are not suitable for image processing in real time. Stereo cameras are technically more complex and require an offset of at least 2 CCD arrays for 3D information receive.

In the field of optical sensor technology is in the grid Optical spatial frequency filtering of the CORREVIT / CORREFOT measurement technik (CORREVIT = trademark of LEICA GmbH Wetz lar) a signal preprocessing with regard to length, Speed and distance measurements at relative to Sensor moving objects (road, industrial belt goods u. a.) realized by lattices between a geome tric imaging optics and downstream photodiodes in  the optical beam path as signal modulators (Heitmann, K., Schneider, E., Eisenkopf, H. Device for non-contact measurement, DE-OS 21 44 487; Leitz, L., Heitmann, K., Schneider, E., institution for Image correlation DE-AS 21 56 617). Here, however, grids only as 1D / 2D beam splitter, but not as dif fractive 1D / 2D / 3D grids are used.

In the institute for optosensorics Dr. Lauinger / Wetzlar were theoretical model calculations based on a 3D grids are based on the "inverted retina of man Lichen eye "and carried out with the human See well-known subjective phenomena (aperture effects, so-called Stiles-Crawford effects I and II, Bezold-Brücke-Phä nouns and a.) seen in relation (Lauinger, N. A new in interpretation of the Stiles-Crawford Effects in Human Vi sion. Journal of Biological Physics 19: 167-188, 1994; Lauinger, N. The realtionship between brightness, hue and saturation when the inverted human retina is interpreted as a cellular diffractive 3D chip. SPIE Proceedings Vol. 2588, October 1995, 208-232). Partial aspects became exemplary te 4D-lattice optically calculated (Carbon, M., Using diffrac tion theory of human vision for design of color vision devices. SPIE Proceedings Vol. 2353, 1994, 550-560). Ef effects of a diffractive 3D grid were created for the inco renz coherence conversion of optical radiation described (Lauinger, N., Method and device for producing coherent electromagnetic waves from into an optical Imaging system incident incoherent rays. DE-OS 37 40 533 A1).  

Despite extensive efforts, especially with regard to Automation and robotics, so far there is no technology high-resolution robust and real-time capable 3D motion determination based on non-contact sensor data collection. The central importance of optical flow fields for visual movement determination is known (Carmesin, H.-O. & Ch. Herwig, Method of Generation maximum information gain for output signals, esp special control signals, from machines, patent application AZ 19503606.9, Structure from Motion by Linear Transforma tion. Proc. Workshop "Cognitive Robotics", B. Krieg- Brückner, Ch. Herwig Edts., ZKW report 3/95, Center for Cognitive Science University of Bremen, March 1995). The main features of neural networks are parallelism and adaptivity (Carmesin, H.O. theory neural adaptation, Köster, Berlin 1994; Carmesin, H.-O. Neurophysics of adaptation. Physics Essays 8 (1), 38-51, 1995). Although neural networks are for self-justification high-resolution image processing are particularly suitable, Neural adaptation theory has so far not been based on Be motion determination still on the evaluation of diffraction re flex sequences used in imaging sensors.

The same problem lies outside of the visible one Wavelength spectrum.

The invention has for its object a genus method and a generic device for To make available with which practically in real time the exact determination of spatial and / or temporal Distances in the focused image of a lens-pupil system  stemes and / or of spatial and / or temporal object parameters such as B. speed or depth light becomes.

With regard to the method, the task is fiction according to solved by the following steps:

• (b) Site-specific modulation of the direction of propagation and / or intensity and / or wavelength and / or polar station and / or the temporal modulation frequency of the electroma gnetic rays by a modulation device behind the lens pupil len system in or near the focal plane of the lens pu pill system.
• (c) Detection of the modulated electromagnetic rays by a detector device, which is behind the Modulation device is located.
• (d) Calculation of the spatial and / or temporal distances of the electromagnetic rays in the modulation direction by determining the modulation and / or calculation space Licher and / or temporal object parameters.

It can be provided that the method follow the steps include:

(b-1) Site-specific modulation of the electromagnetic Blasting through a modulation device with a mod  filter in the focal plane of the lens-pupil system stems, which is an incident electromagnetic Beam through location-specific filter elements indexed with i elements modulated, the filter elements alternating relatively rich in solvability of equations systems for performing step (d) modulating or are not selected modulating.

(c-1) focusing within the detector device through a second lens-pupil system.

(c-2) Measurement of the modulated electromagnetic rays in the focal plane of the detector device by an imaging detector arrangement of detector elements indexed with j, into which the filter elements with index i are projected from an associated index set I _j .

(d-1) Calculate the exact position of one by Parame ter determined intensity distribution in the beam path indirectly in front of the modulation device in the archetype of a jth detector element from the measurements in the jth detector gate element, the calculation being carried out in such a way that the number of measurement results the number of to be determined Parameter exceeds due to the modulation measurements and a system of equations that is as independent as possible occurs.

It can also be provided that the method includes the following step:

(d-1-1) Perform the calculation on an edge element parameterized by slope m and y intercept b in  the x-y plane of the modulation device according to y = mx + b.

In addition, the use of j-th detector elements ( 19 ) can be provided together with adjacent detector elements.

The use of linear filter elements can also be used elements should be provided.

In a particular embodiment of the invention it should be provided that the method follows the following steps includes:

(b-1) Site-specific modulation through modulation facility in the form of a site-specific facility Redirection with deflection elements and, if necessary, also to the module tion of intensity, wavelength, polarization and modula frequency of light rays in or near the Fo kus plane of the lens-pupil system.

(d-1) Calculation of the place of origin of a considered electromagnetic beam in the modulation device based on the signals detected in detector elements, and the resulting spatial and / or temporal Ab would be in the focused image of the lens-pupil system and / or spatial and / or temporal object parameters.

In a further embodiment it can be provided that the process comprises the following steps:

(b-1) Site-specific modulation of the electromagnetic  Radiation through a modulation device in the form of a curved 3D diffraction grating in or near the focus plane of the lens-pupil system, with the diffracted Rays corresponding to the reflection on inclined network planes oriented of the curved 3D diffraction grating and / or according to the lattice aperture conditions in their intensity modulated and / or according to the Bragg condition modulated in the wavelength and / or according to the Anisotropies of the diffraction grating are polarized and / or ent speaking of a possible imprinted temporal Schwin are modulated in time, the site-specific Curvature and the other diffraction grating properties in the In terms of site-specific modulation with regard to on the solvability of systems of equations for implementation from step (d) are sufficiently varied.

(d-1) Calculation of the place of origin of a considered electromagnetic beam in the modulation device based on the signals detected in detector elements and the resulting spatial and / or temporal Ab would be in the focused image of the lens-pupil system and / or spatial and / or temporal object parameters.

The method can further comprise the following steps:

(c ′) Differentiation of the run into a detector element the electromagnetic rays into those that come from a 0. Diffraction order, and those that come from a higher Beu order, based on the specific wavelength the diffracted electromagnetic rays according to the Bragg condition and / or with the help of direction-specific  Detector elements.

(d ′) Calculation of the place of origin of a considered electromagnetic beam in the modulation device based on the signals detected in detector elements, and the resulting spatial and / or temporal Ab would be in the focused image of the lens-pupil system and / or spatial and / or temporal object parameters.

Conveniently, the process comprises the following steps te:

(c ′ ′) application of a temporal high-pass filter to the signals detected by a detector element, the transmitted frequencies are to be adjusted adaptively, that the significantly increased in picture sequences since speed of the higher diffraction orders is, and thus differentiation in a detector element incoming electromagnetic rays into those that are not deflected or from a 0th diffraction order originate, and those that are redirected or from a higher diffraction order.

(d ′ ′) calculation of the place of origin of a considered electromagnetic beam in the modulation device based on the signals detected in detector elements, and the resulting spatial and / or temporal Ab would be in the focused image of the lens-pupil system and / or spatial and / or temporal object parameters.

Furthermore, the method according to the invention can follow the  the steps include:

(b-1) Site-specific modulation of the electromagnetic Radiation through a modulation device in the form of a 3D non-curved gratings in or near the Fo kusplane of the lens-pupil system, the diffracted Rays corresponding to the reflection on inclined network planes oriented of the uncurved 3D diffraction grating and / or according to the lattice aperture conditions in their Intensity modulated and / or according to the Bragg-Be condition modulated in the wavelength and / or accordingly polarizes the anisotropies of the diffraction grating and / or according to a possible imprinted time Vibration are modulated in time, with the diffraction grid properties with regard to the site-specific mod lation with regard to the solvability of equation sy sufficiently to perform step (d) are varied.

(d-1) Calculation of the place of origin of a considered electromagnetic beam in the modulation device based on the signals detected in detector elements and the resulting spatial and / or temporal Ab would be in the focused image of the lens-pupil system and / or spatial and / or temporal object parameters.

The method can also carry out the steps (c) and (d) in the Fraunhofer far field.

It can also carry out steps (c) and (d) in Talbot or Fresnel planes in the near field of 3D diffraction  include grids.

The following steps can be provided:

• - Registration of the depth of focus in Talbot and / or Fresnel planes in the near field behind a 3D diffraction grid with a depth-sensitive detector element.
• - Calculate the difference between two neighboring Ent removals from objects.

The method preferably comprises the following steps:

(a′-1) Preliminary definition of distinctive 3D features or Patterns in the form of picture elements, together with one drawn position picture element as the origin of a lo cal coordinate system for each pattern or for each 3D feature.

(a′-2) carrying out a so-called learning phase, each 3D feature being brought into every 3D position and 3D orientation, or every pattern being brought into every position and orientation,
then determining the detector element j onto which the position picture element is projected,
subsequently the associated signal pattern, including signal pattern of the modulations, of the detector elements in the vicinity of the jth detector element is determined,
and then the assignment of the j-th detector element together with the signal pattern of the surrounding detector elements to the pattern or 3D feature, position of the position image element and orientation is stored in an associative neural network.

(a′-3) Application of the learned and saved assignment on the measured jth detector element together with signal mu ster of the surrounding detector elements to patterns or 3D feature, position of the position picture element and orien through an associative neural network.

The use of patterns can be provided in Form of picture elements based on intensity patterns in the focal plane of the lens-pupil system and / or Bewe supply patterns and / or frequency patterns and / or color or Wavelength patterns and / or polarization patterns and / or Intensity maxima and / or 3D shapes.

In addition, the use of new can also be provided ronal networks using the Hopfield rule or a perceptron learning rule or a back propagation learning learn rule and the couplings between two neurons and / or couplings between three neurons and / or Kopp lungs between four neurons and / or couplings between a maximum number of neurons that add complexity the signal pattern is adapted, and those in complicated Learning tasks using a perceptron learning algorithm Couplings with a maximum number of neurons, which on the complexity of the signal pattern is adjusted, so that a solution of the learning task according to a Kon  evidence of vergence is ensured.

The assignment can be stored through an electronic or optical memory.

In a particular embodiment of the invention the execution of the calculation steps is provided and / or neural networks before or after the detector level through a correlator-optical structure.

With regard to the device, the task is fiction according to solved by a modulation device on site specific modulation of the direction of propagation and / or Intensity and / or wavelength and / or polarization and / or modulation frequency of the electromagnetic Rays that are behind the lens-pupil system in or close to the focal plane of the lens-pupil system located.

It can be provided that the modulation device a modulation filter in the focal plane of the lens pu pills system, which deflection elements in the form of Contains mirror elements.

In a particular embodiment of the invention comprises the modulation device has a curved 3D diffraction grid ter, the incident electromagnetic radiation in ge diffracted electromagnetic radiation is implemented, the ge diffracted rays according to the reflection on inclined Network planes of the curved 3D diffraction grating oriented and / or according to the lattice aperture conditions in ih  rer intensity modulated and / or according to the Bragg-Be condition modulated in the wavelength and / or correspond polarized according to the anisotropies of the diffraction grating and / or according to a possible imprinted Zeitli Chen vibration are modulated in time.

Alternatively, the device can be an uncurved 3D diffraction gratings include, the diffracted rays ent speaking of the reflection on inclined network planes of the ge curved 3D diffraction grating oriented and / or correspond according to the lattice aperture conditions in their intensity modulated and / or according to the Bragg condition in the Wavelength modulated and / or according to the anisotropy pien of the diffraction grating polarized and / or accordingly a possible imprinted temporal oscillation time are modulated.

It can also be provided that the 3D diffraction grating one of the 231 possible periodic 3D crystal lattices represents.

On the other hand, it can also be provided that the 3D Beu grid is periodic.

Furthermore, the 3D diffraction grating can have multiple layers of Include 2D grids.

On the other hand, it can also be provided that the 3D Beu Grid acousto-optical cells.

It can also be provided that the 3D diffraction grating  Wigner crystals included.

On the other hand, it can also be provided that the 3D Beu Grid polymer latex comprises.

It can also be provided that the 3D diffraction grating holographically produced layers.

Furthermore, the detector device can be spectrally sensitive show speed.

The detector device preferably has directional temperature sensitivity to.

In addition, the detector device preferably contains min at least a polarization filter.

In a special embodiment can also be provided be that additionally in the imaging optical beam path other gratings are attached to the 3D diffraction grating, which increase the degree of coherence.

Furthermore, it can also be provided that the detector direction contains a CCD array.

In addition, the detector elements preferably contain Waveguide.

In a preferred embodiment it can be provided that the lens-pupil system and the modulation device device for electromagnetic waves, including X-ray  rays, designed outside the visible light are. This is particularly advantageous for a picture / sound / language che processing in the acousto-optical message area (Videophone and others) and also the 3D measurement of star and Galaxy distances by designed for X-rays grid vision systems.

Furthermore, it can also be provided that the modulation facility vibrates at carrier frequency.

The device advantageously has a neural one Network on.

In addition, a correlator-optical structure can go in front or in front ter of the detector level to carry out the calculations be provided.

The correlator-optical structure can be 4f optics include.

Finally, in a special embodiment, the Invention can be provided that in the modulation device device filter elements and / or deflection elements and / or git elements using a random number generator relative are arranged in a varied manner.

The invention is based on the surprising finding that in the image plane with the length resolution about the Light wavelength focusable electromagnetic beam approximately with the accuracy of the light wavelength Modulate site-specifically. The thus modulated elec  tromagnetic rays can later be in a detector device with lower length resolution can be detected, and the locations in the image plane can be based on the location-specific fishing modulation can be calculated. For example neh electromagnetic rays diffracted on a grating according to the reflection condition one by the Kri stall orientation certain direction and one of the Bragg condition corresponding color.

The advantages achieved with the invention are shown here playful for the visible wavelength spectrum in the set out the following points:

• (1) At present, the accuracy of the determination of distances of a brightness distribution in a focal plane is determined by the dimensions of the pixels of an imaging sensor, these are approximately 20 micrometers. When using a special embodiment of the device according to the invention, ie with a modulation device in the form of a 3D diffraction grating, this accuracy is limited by the wavelength of the light and the grating spacing. When choosing an optimal grating, these are 0.4 micrometers. This length resolution can therefore be increased by a factor of 50 and the corresponding area resolution by factor 2500.
  Such an increase in accuracy is important because the human eye has 10⁸ receptors in the retina with a corresponding length resolution capacity of about 1 micron each, because correspondingly culturally created environments in the home, leisure, traffic and work world rely on visual systems such as that of human beings, and since future intelligent mobile automats and robots will need comparable visual systems to be safe, reliable and economically operable.
• (2) The inventive method can spatial and / or temporal object parameters, such as speed speeds or depths, quickly and with little effort be evaluated optically.
• (3) By the method according to the invention for each applications for corresponding specific optical Devices and their dimensions universal algorithms in the form of learning neural networks on intended appropriate place can be applied without a new one Algorithm would have to be developed. This will make it more common wise, significant development costs of avoided beforehand.
• (4) The diffraction theory is large Di for the limit case punch from the grid (far field) very developed; they fin det extensive application in optics and material research. In contrast, physics is for the near realm Fields relatively complex and application-specific; ent speaking the theory is even more viable, it are not yet any applications related to movement mood known, whereas the learnable used here neural networks applicable due to their flexibility and are particularly adequate. It is generally the central one Importance of river fields in motion determination  already known since Helmholtz, also applies to distances From five meters the movement determination from river fields as advantageous compared to the movement determination by Bino Specularity, in general, offers visual determination against spatial and / or temporal object parameters over other modalities the advantage of high resolution, ho range, high speed and more natural Radiation sources.
• (5) A light field generally contains both amplitude and also phase information. Most technical visual systems only ver the amplitude information applies, however, in the method according to the invention with the help of the 3D diffraction grating also the included one 3D phase information can be used, even with Use incoherent light such as natural Lichen radiation sources is emitted. With natural visual systems finds an advantageous evaluation of Phase information presumably with the human eye Using the 3D diffraction grating of the inverted retina instead of. The inverted retina may be one qualitative leap in the development history of the Se hens, which then took place for the first time at the anthropods, and for 3D object parameter determination and thus for that Survival was immediately relevant.
• (6) The neural networks have been in the past year tenth experienced a rapid development. The essential The advantages are parallelism and adaptivity. Thereby neural networks particularly suitable for self-adjustment high-resolution image processing in real time,  even with complex evaluations. Here is the technical Development so far far behind the natural pre picture back. According to a particular embodiment of the The inventive method, it is possible to neurona networks at a point where they can can also develop full potential at parallel high-resolution reconstruction of object para meters of rich, possibly modulated and moving diffraction patterns.

Further features and advantages of the invention result from the claims and from the description below, in the exemplary embodiments for the visible wavelength gene spectrum based on the schematic drawings in a are explained.

It shows:

Fig. 1 shows schematically a device according to the prior invention;

Fig. 2 shows a specific embodiment of the device according to the invention;

Fig. 3 shows schematically the imaging and resolution situation for a specific embodiment of the device according to the invention;

FIG. 4a shows a further embodiment of the inventive device SEN;

FIG. 4b shows a detail of Fig. 4a;

FIG. 5 shows the situation when using a curved 3D diffraction grating in accordance with another exporting approximate shape of the device according to the invention; and

Fig. 6 schematically shows the situation when performing a learning phase according to an embodiment of the inventive method.

Fig. 1 shows schematically a device according to the prior invention. This includes a lens-pupil system 10 , which focuses the incident light (here symbolized by a light beam 11 ), a modulation device 12 for location-specific modulation of the light that radiates in the focal plane of the lens-pupil system 10 and is located in front of a detector device 13 located behind it. The light beam 11 strikes at a location 14 (origin) in the modulation device 12 and is modulated in the present case with regard to its direction of propagation.

Fig. 2 shows a specific embodiment of the device according to the Invention. A modulation device 12 is in the form of a device for site-specific deflection 22 with deflection elements 23 in the form of mirror elements. Since the inclination of the deflection elements 23 is location-dependent, the direction of the reflected light beams 26 , 27 is location-dependent. Consequently, the reflected light rays 26 , 27 in a detector device 13 with an imaging detector arrangement 17 of detector elements 19 in specific pixels which correspond to the direction and thus the deflection element 23 at the point of origin. Consequently, distances can be detected in this way with the detector device 13 , which could not be resolved without the modulation device 12 due to the dimensions of the detector elements 19 , since without a deflection, the light beams 11 , as indicated by the light beams 24 and 25 , on one and the same detector element 19 would hit.

In a special embodiment it can also be provided that semi-transparent micro mirrors are used as deflection elements 23 , which also color the transmitted light beams 24 and 25 red. For each red local brightness maximum in a detector element 19 , the associated local brightness maximum of the non-red, ie reflected light beams 26 and 27 is determined with increased accuracy. From their position in the detector arrangement 17 one determines the place of origin in the modulation device 22 .

Furthermore, it can be provided that for each red local brightness maximum in a detector element 19, the associated local brightness maximum at the following point in time and the speed of the local brightness maximum is determined therefrom. From this speed one estimates the increased speed of the redirected light rays. A high-pass filter is set accordingly and the deflected light beams 26 and 27 are detected with increased accuracy. The position of origin in the modulation device 12 is then determined from their position in the detector arrangement 17 .

With the present invention, a crate can be Accuracy through bit multiplication per module tion, as theoretically shown in the following: A light beam that only in without modulation device light / dark to determine spatial and temporal Distances and object parameters contribute in the frame information theory with one bit. By the modu The three colors red, green and blue, each with egg nem additional bit, the number of bits is increased by three device. With a calculation, which these bits also in the certain spatial and temporal distances and object parameters, you get an increase in Ge accuracy by three bits, d. H. almost a decimal place. This connection should be based on a concrete execution example.

The device according to the invention shown in FIG. 3 has the following parameters:
Lens-pupil system 10 when using a deflecting device 22 , 23 : focal length: 17 mm; Panel diameter: 4 mm. Resulting aperture cone opening angle: 2/17. Distance 29 between a modulation device 12 and a detector device 13 : 20 μ. Lateral deflection of the deflected beams on the detector device 13 : 20 µ. Resulting Airy disk diameter 28 : 5 µ. Size of a detector element 19 : 20 µ.

This leads to a doubling of the number of detector elements 19 reached in length and to a quadrupling of the number of detector elements 19 reached in the plane and thus to a quadrupling of the number of bits. With knowledge of the relevant technical data of a specific application, a significant improvement in the dimensions can be expected.

Fig. 4a shows a further embodiment of the device according to the present invention. Behind a lens-pills system 10 is a modulation device 12 with a modulation filter 15 in the focal plane. A detector device 13, which in turn is arranged behind the modulation device 12 , comprises a second lens pupil system 16 for focusing the modulated light beams and an imaging detector arrangement 17 of detector elements 19 .

Fig. 4b shows a detail of the device of Fig. 4a. The modulation filter 15 in the focal plane of the lens-pills system 10 includes location-specific filter elements 18th The modulated light beams are measured by the imaging detector array 17 of detector elements 19 . The embodiment is in principle prototypical for all modulations by the modulation device 12 . The calculation of the exact position of an edge element 21 around the beam path immediately in front of the modulation device 12 is carried out as follows:

The brightness H _i on a filter element 18 be zero on the edge element 21 and change on the other filter elements with coordinates x _i and y _{i of} the original image 20 of a detector element 19 as follows

H _i = y _i -mx _i -b (1)

The filter elements 18 are either permeable, F _i = 1, or impermeable to the x-polarization direction, F _i = 0. The F _i are chosen randomly. In the Detektoreinrich device 13 , both the brightness H _j = Σ _{i ε Ij} H _i and the brightness H _j ^p = Σ _{i ε Ij} F _i H _i behind a polarization filter (not shown), which is opaque for the y-direction sig is measured. By inserting Eq. (1) one gets

where x _i and y _i are known by design and H _{j is} measured, and

where F _i , x _i and y _i are known by design and H _j ^{p is} measured. Overall, equations (2) and (3) represent a linear system of equations with the two unknown quantities m and b, which is linearly independent due to the randomized filter elements F _i . Therefore, m and b and thus the position of the edge element ( 21 ) can be calculated directly in cash.

Fig. 5 shows the situation when using a curved 3D diffraction grating in the xy plane of a Modulationsein direction 12 in which the grating elements represented by local coordinate systems, one for the x-coordinate per-proportional azimuth angle 32 and a to the y-Koordi nate proportional polar angle 33 are rotated. For each local maximum brightness of the zeroth diffraction order, the associated diffracted light beams are determined using the color determined by the Bragg condition. The position of the diffracted light rays is used to determine the azimuth and polar angles. The x coordinate and the y coordinate of the place of origin in the modulation device 12 are determined from the azimuth angle and the polar angle with increased accuracy.

If one uses one in the x-y plane of the modulation device uncurved 3D diffraction grating, so be one agrees with every local brightness maximum of the zeroth Diffraction order the associated diffracted light rays based on the color determined by the Bragg condition. On hand the position of the undeflected and diffracted light rays one determines the position of the place of origin in the modu lationseinrichtung subpixel due to redundancy the data and with the help of an adequate interpolationfor mel.

It is advantageous to carry out the following steps detention:

An edge element 21 and an uncurved 3D diffraction grating are used as a striking pattern. The resultant diffraction patterns are determined for a detector element 19 in a detector arrangement 17 for all positions and orientations of an object 30 and the detector elements 19 (see FIG. 6) and the assignment of the position and orientation to each diffraction pattern is stored. Because of the translational invariance of the uncurved 3D diffraction grating, this assignment is used for each detector element 19 .

Advantageously, the patterns or distinctive 3D features are chosen with regard to a good differentiation of the resulting signal pattern of the detector elements 19 sufficiently large and different, so that the crosstalk known from the theory of neural networks becomes irrelevant due to the laws of large numbers. Furthermore, the patterns or distinctive features are chosen relatively differently for the same purpose, ie with relatively little overlap in the sense of the theory of neural networks. In general, such an application of neural networks enables the detection of relations between spatial and / or temporal object parameters of stationary and / or moving 3D features and the resulting signal patterns of the detector elements 19 , this applies particularly when using perceptron learning algorithms and neurons some multilinear couplings, ie with couplings between two, three, four etc. neurons. In particular, these relations can be learned through the network, even if no classic calculation rules and / or algorithms have been developed for these relations. In the case of complicated learning tasks for the assignments occurring here, a perceptron learning algorithm with couplings between a maximum number of neurons, which is adapted to the complexity of the learning task and for which there is a convergence proof valid for the learning task, is to be used. Such learning algorithms and convergence evidence can be found in the book "Neuronal Adaptation Theory" published by Carmesin in early 1996.

In view of the always limited storage capacity of neural networks are initially only defined as important  most distinctive features and patterns and stores the associated assignments in the network. Whilst saving further assignments of further distinctive features and / or samples you always check the quality of the recognition nung; if this decreases, the storage capacity is ver probably achieved. You then use a network with more Neurons.

The one in the previous description, in the drawing as well as features of the invention disclosed in the claims can be used individually or in any combination for the realization of the invention in its various NEN embodiments may be essential.

Reference list

10 lens-pupil system
11 light beam
12 modulation device
13 detector device
14 Place of origin of a light beam in the modulation device
15 modulation filters
16 second lens-pupil system
17 imaging detector arrangement
18 filter element or image element
19 detector element
20 archetype of a detector element
21 edge element
22 Device for site-specific diversion
23 deflection element
24 , 25 light rays that are not deflected
26 , 27 light rays that are deflected
28 Airy disk diameters
29 Distance between modulation device and detector device
30 object
32 azimuth angle
33 polar angles

Claims (61)

1. The method and apparatus for the precise determination of spatial and / or temporal distances in the focussed image of a lens-pupil system and / or of spatial and / or temporal object parameters, such as. B. speed or depth, in particular for the purpose of precise and intelligent control of machines, which comprises the following step:

• (a) focusing the incident electromagnetic rays by means of a lens-pupil system ( 10 ),

characterized by the following steps:

• (b) Location-specific modulation of the direction of propagation and / or intensity and / or wavelength and / or polarization and / or the temporal modulation frequency of the electromagnetic rays ( 11 ) by a modulation device ( 12 ) behind the lens-pupil system ( 10 ) in or close to the focal plane of the lens-pupil system ( 10 ).
• (c) Detection of the modulated electromagnetic rays by a detector device ( 13 ) which is located behind the modulation device ( 12 ).
• (d) Calculation of the spatial and / or temporal distances ( 14 ) of the electromagnetic rays in the modulation device ( 12 ) by determining the modulation and / or calculation of spatial and / or temporal object parameters.

2. The method of claim 1, comprising the following steps includes: (b-1) Location-specific modulation of the electromagnetic rays by a modulation device ( 12 ) with a modulation filter ( 15 ) in the focal plane of the lens-pupil system ( 10 ), which an incident electromagnetic beam by location-specific, with i-indexed filter elements ( 18 ) modulated, the filter elements ( 18 ) being selected to be relatively varied with regard to the solvability of equation systems for performing step (d) modulating or non-modulating. (c-1) focusing within the detector device ( 13 ) by a second lens-pupil system ( 16 ). (c-2) Measurement of the modulated electromagnetic beams in the focal plane of the detector device ( 13 ) by an imaging detector arrangement ( 17 ) of detector elements ( 19 ) indexed with j, into which the filter elements ( 18 ) with index i from an associated index set I _j be projected. (d-1) calculation of the exact position of an intensity distribution in the beam path determined by parameters immediately before the modulation device ( 12 ) in the original image ( 20 ) of a j-th detector element ( 19 ) from the measurements in the j-th detector element ( 19 ), the calculation being carried out in such a way that the number of measurement results exceeds the number of parameters to be determined on the basis of the modulation measurements and an independent system of equations occurs. 3. The method of claim 2, comprising the following step includes: (d-1-1) Carrying out the calculation on an edge element ( 21 ) parameterized by slope m and y intercept b in the xy plane of the modulation device ( 12 ) according to y = mx + b. 4. The method according to any one of claims 2 or 3, characterized by the use of j-th detector elements ( 19 ) together with adjacent detector elements ( 19 ). 5. The method according to any one of claims 2-4, characterized by the use of linear filter elements ( 18 ). 6. The method of claim 1, comprising the following steps includes: (b-1) Site-specific modulation by a modulation device ( 12 ) in the form of a device for location-specific deflection ( 22 ) with deflection elements ( 23 ) and possibly also for modulating the intensity, wavelength, polarization and modulation frequency of electromagnetic rays in or close to the focal plane of the lens-pupil system ( 10 ). (d-1) Calculation of the origin ( 14 ) of a considered electromagnetic beam in the modulation device ( 12 ) on the basis of the signals detected in detector elements ( 19 ), and the resulting spatial and / or temporal distances in the focused image of the lens Pupil system ( 10 ) and / or spatial and / or temporal object parameters. 7. The method of claim 1, comprising the following steps includes: (b-1) Location-specific modulation of the electromagnetic rays by a modulation device ( 12 ) in the form of a curved 3D diffraction grating in or near the focal plane of the lens-pupil system ( 10 ), the diffracted rays corresponding to the reflection on inclined network planes of the curved 3D diffraction grating oriented and / or modulated in intensity according to the grating aperture conditions and / or modulated in wavelength according to the Bragg condition and / or polarized in accordance with the anisotropies of the diffraction grating and / or temporally according to a possible impressed temporal vibration are modulated, the location-specific curvature and the other diffraction grating properties with regard to the location-specific modulation with regard to the solvability of equation systems for carrying out step (d) being sufficiently varied. (d-1) Calculation of the origin ( 14 ) of a considered electromagnetic beam in the modulation device ( 12 ) on the basis of the signals detected in detector elements ( 19 ) and the resulting spatial and / or temporal distances in the focused image of the lens pupils -System ( 10 ) and / or spatial and / or temporal object parameters. 8. The method according to any one of claims 1, 6 or 7, characterized by the following steps: (c ') Differentiation of the electromagnetic rays running into a detector element ( 19 ) into those that come from a 0 diffraction order ( 24 and 25 ) and those that come from a higher diffraction order ( 26 and 27 ), based on the specific wavelength the diffracted electromagnetic rays according to the Bragg condition and / or with the help of direction-specific detector elements ( 19 ). (d ′) calculation of the origin ( 14 ) of an electromagnetic beam under consideration in the modulation device ( 12 ) on the basis of the signals detected in detector elements ( 19 ) and the resulting spatial and / or temporal distances in the focused image of the lens pupil. Systemes ( 10 ) and / or spatial and / or temporal object parameters. 9. The method according to any one of claims 1, 6, 7 or 8, which includes the following steps: (C '') application of a temporal high-pass filter to the signals detected by a detector element ( 19 ), the frequencies passed are to be set so adaptively that the significantly increased speed of the higher diffraction orders occurring in image sequences is detected, and thus differentiation in a detector element ( 19 ) incoming electromagnetic radiation into those that are not deflected ( 24 and 25 ) or come from a 0th diffraction order, and those that are deflected ( 26 and 27 ) or come from a higher diffraction order. (d '') Calculation of the origin ( 14 ) of a considered electromagnetic beam in the modulation device ( 12 ) on the basis of the signals detected in detector elements ( 19 ), and the resulting spatial and / or temporal distances in the focused image of the lens Pupil system ( 10 ) and / or spatial and / or temporal object parameters. 10. The method of claim 1, comprising the following steps includes: (b-1) Site-specific modulation of the electromagnetic rays by a modulation device ( 12 ) in the form of an uncurved 3D diffraction grating in or near the focal plane of the lens-pupil system ( 10 ), the diffracted rays corresponding to the reflection on inclined network planes of the uncurved 3D diffraction grating oriented and / or modulated in intensity according to the grating aperture conditions and / or modulated in wavelength in accordance with the Bragg condition and / or polarized in accordance with the anisotropies of the diffraction grating and / or temporally according to a possible impressed temporal vibration are modulated, the diffraction grating properties with regard to the site-specific modulation with regard to the solvability of equation systems for carrying out step (d) being sufficiently varied. (d-1) Calculation of the origin ( 14 ) of a considered electromagnetic beam in the modulation device ( 12 ) on the basis of the signals detected in detector elements ( 19 ) and the resulting spatial and / or temporal distances in the focused image of the lens pu pills system ( 10 ) and / or spatial and / or temporal object parameters. 11. The method according to any one of claims 7-10, characterized net by performing steps (c) and (d) in Fraunhofer Fernfeld.   12. The method according to any one of claims 7-11, characterized net by performing steps (c) and (d) in Talbot or Fresnel planes in the near field of 3D diffraction lattice. 13. The method according to claim 12, characterized by the following steps:

• - Registration of the depth of focus in Talbot and / or Fresnel planes in the near field behind a 3D diffraction grating with a depth-sensitive detector element ( 19 ).
• - Calculation of the difference between two adjacent distances from objects ( 30 ).

14. The method according to any one of claims 1-13, the fol steps include: (a′-1) Preliminary definition of distinctive 3D features or Patterns in the form of picture elements, together with one drawn position picture element as the origin of a lo cal coordinate system for each pattern or for each 3D feature. (a′-2) carrying out a so-called learning phase, each 3D feature being brought into every 3D position and 3D orientation, or every pattern being brought into every position and orientation,
then the detector element ( 19 ) j is determined, onto which the position picture element is projected,
subsequently the associated signal pattern, including signal pattern of the modulations, of the detector elements ( 19 ) in the vicinity of the j-th detector element ( 19 ) is determined,
and then the assignment of the j-th detector element ( 19 ) together with the signal pattern of the surrounding detector elements ( 19 ) to the pattern or 3D feature, position of the position image element and orientation is stored in an associative neuronal network. (a'-3) application of the learned and stored assignment to the measured j-th detector element ( 19 ) with the signal pattern of the surrounding detector elements ( 19 ) to pattern or 3D feature, position of the position picture element and orientation by an associative neural network . 15. The method according to claim 14, characterized by the use of patterns in the form of picture elements based on intensity patterns in the focal plane of the lens-pupil len system ( 10 ) and / or movement patterns and / or frequency patterns and / or color or wavelength patterns and / or polarization patterns and / or intensity maxima and / or 3D shapes. 16. The method according to any one of claims 14 or 15, characterized by the use of neural network works that use the Hopfield rule or a Per zeptron learning rule or a back propagation learning rule learn and the couplings between two neurons and / or  Couplings between three neurons and / or couplings between four neurons and / or couplings between one Maximum number of neurons that depend on the complexity of the Signal pattern is adapted, and that with complicated learning tasked a perceptron learning algorithm together with Couplings with a maximum number of neurons, which on the complexity of the signal pattern is adjusted, so that a solution of the learning task according to a Kon evidence of vergence is ensured. 17. The method according to any one of claims 14-16, characterized by the following step: (a'-2-1) Storage of the assignment by an electronic or optical storage. 18. The method according to any one of claims 1-17, characterized by the execution of the calculation step te and / or neural networks before or after the detection gate level through a correlator-optical structure. 19. Device for performing the method according to one of claims 1-18 with a lens-pupil system ( 10 ) for imaging a scene and a detector beam ( 13 ) located behind it in the imaging beam path, characterized by a modulation device ( 12 ) for location-specific modulation the direction of propagation and / or intensity and / or wavelength and / or polarization and / or modulation frequency of the electromagnetic rays, which is located behind the lens-pupil system ( 10 ) in or near the focal plane of the lens-pupil system ( 10 ) . 20. The apparatus according to claim 19, characterized in that the modulation device ( 12 ) comprises a Modulationsfil ter ( 15 ) in the focal plane of the lens-pupil system, which contains deflection elements ( 23 ) in the form of mirror elements. 21. The apparatus according to claim 19, characterized in that the modulation device ( 12 ) comprises a curved 3D diffraction grating which converts incident electromagnetic rays into diffracted electromagnetic beams, the diffracted beams being oriented in accordance with the reflection on inclined network planes of the curved 3D diffraction grating and / or in accordance with the lattice aperture conditions modulated in their intensity and / or in accordance with the Bragg condition in the wavelength modulated and / or polarized in accordance with the anisotropies of the diffraction grating and / or modulated in time in accordance with a possible impressed temporal oscillation. 22. The apparatus according to claim 19, characterized in that the modulation device ( 12 ) comprises an uncurved 3D diffraction grating, wherein the diffracted rays are oriented according to the reflection on inclined network planes of the curved 3D diffraction grating and / or accordingly the grating aperture conditions in their intensity mo dulated and / or modulated in the wavelength according to the Bragg condition and / or polarized according to the anisotropy of the diffraction grating and / or modulated in time according to a possible impressed temporal oscillation. 23. Device according to one of claims 21 or 22, there characterized in that the 3D diffraction grating is one of the Represents 231 possible periodic 3D crystal lattice. 24. Device according to one of claims 21 or 22, there characterized in that the 3D diffraction grating is not pe is periodic. 25. The device according to any one of claims 21-24, characterized characterized in that the 3D diffraction grating several layers 2D grids. 26. Device according to one of claims 21-24, characterized characterized that the 3D diffraction grating acousto-optical Cells includes. 27. The device according to any one of claims 21-24, characterized characterized in that the 3D diffraction grating Wigner crystals includes. 28. The device according to any one of claims 21-25, characterized characterized in that the 3D diffraction grating polymer latices includes. 29. Device according to one of claims 21-25, characterized characterized in that the 3D diffraction grating is holographic produced layers comprises.   30. Device according to one of claims 19-29, characterized in that the detector device ( 13 ) has spectral sensitivity. 31. Device according to one of claims 19-30, characterized in that the detector device ( 13 ) Rich direction sensitivity. 32. Device according to one of claims 19-31, characterized in that the detector device ( 13 ) contains at least one polarization filter. 33. Device according to one of claims 21-32, characterized characterized in that in the imaging optical beam path In addition to the 3D diffraction grating, additional gratings are added are brought, which increase the degree of coherence. 34. Device according to one of claims 19-33, characterized in that the detector device ( 13 ) contains a CCD array. 35. Device according to one of claims 19-34, characterized in that the detector elements ( 19 ) contain waveguides. 36. Device according to one of claims 19 to 35, characterized in that the lens-pupil system ( 10 ) and the modulation device ( 12 ) are designed for electromagnetic waves, including X-rays, outside the visible light. 37. Device according to one of claims 19 to 36, characterized in that the modulation device ( 12 ) vibrates carrier frequency. 38. Device according to one of claims 19-37, characterized characterized by a neural network. 39. Device according to one of claims 19-38, characterized characterized by a correlator-optical structure or behind the detector level to perform the Offsets. 40. Device according to claim 39, characterized in that the correlator-optical structure comprises 4f optics. 41. Device according to one of claims 19-40, characterized in that in the modulation device ( 12 ) filter elements ( 18 ) and / or deflection elements ( 23 ) and / or grating elements are arranged in a relatively varied manner by means of a random number generator.