Method and system for optical measurement of an object having a reflective and / or partially reflective surface and corresponding measuring device

Abstract

The invention discloses a method for optical measurement of an object having a reflective and / or partially reflective surface. Here, a planar pattern (13) is generated by means of a pattern generator (1), which varies in at least one optical property, so that a large number of different points (p) or a large number of different point groups can be distinguished from one another at least in subregions (10). At least a part of the pattern (13) is reflected as a reflected pattern onto a detector (14) of a camera unit (4) by means of the reflective surface (2) of the object (3), wherein the reflected pattern is converted into a camera image (9) by means of the detector (14). The relationship between the points (q) of the camera image (9) and the corresponding points (p) of the pattern (13) can be described by means of a correspondence function, which is related to the geometric properties of the reflective surface (2) of the object (3). At least one of the geometric properties of the reflective surface (2) of the object (3) is determined by using the differential geometric properties of the transformation given by the correspondence function. Furthermore, a corresponding system and a corresponding measuring device are disclosed.

Description

Technical Field

[0001] The present invention relates to a method and system for optical measurement of objects having reflective and / or partially reflective surfaces, as well as a corresponding measuring device. Background Technology

[0002] In particular, this method, system, and measuring device are used to perform comprehensive optical measurements of the surface morphology of objects with reflective or partially reflective free surfaces. Applications include the automotive industry (e.g., for painting body parts), the optical industry (e.g., for measuring lenses and reflectors), and also in the consumer sector (e.g., for measuring mobile phones and flat panel displays), to name just a few use cases.

[0003] For this type of measurement, phase-deflection measurement is a well-established and frequently used method. The basic characteristics of this deflection measurement method are described in DE 199 44354A1. In this case, a sinusoidal fringe pattern is typically displayed on a calibrated screen and observed on the reflective surface of the object to be measured using one or more calibrated cameras. By performing appropriate phase evaluations in multiple directions, the coordinates of the observed pattern points in the screen plane can be inferred for each camera pixel. Under certain conditions, the surface normal can be inferred by beam tracing of the camera line of sight to the observed screen point, and the shape of the observed object can also be inferred after numerical integration.

[0004] A known problem with this approach is the so-called height ambiguity: since each camera pixel contains only information about the position of the observed screen point, and not about its emission direction, the surface normal can only be definitively determined if the object's orientation ("height") is known. Different height assumptions result in different surface normals. This ambiguity, often referred to as the "regularization problem or height problem" of deflection measurement, can be solved through various schemes, but these schemes all introduce specific drawbacks.

[0005] In one approach described by T. Bonfort et al., “General Specular Surface Triangulation, Proceedings of the 7th Asian Conference on Computer Vision - Volume Part I, 872-882, Berlin-Heidelberg: Springer-Verlag (2006),” the unknown emission direction is determined by a second measurement using a screen position shifted perpendicular to the screen plane. However, this requires complex mechanical devices for screen shifting and demanding calibration of the structure at both screen positions. This approach also increases measurement time.

[0006] In another approach, the orientation of an unknown object can be determined by an absolute measurement at a single point using an additional, independent measurement method. This measurement can be optical (e.g., using a triangulation sensor) or tactile. The unknown height of other points on the surface can be determined based on this measured surface point by iteratively integrating the surface normal calculated under each current height assumption until self-consistent. This action is described, for example, in E. Slogsnat et al., “Non-stereoscopic Method for Deflectometric Measurement of Reflecting Surfaces, Proc. DGaO 110A11 (2009)”. A drawback of this approach is that the additional point height measurements must be integrated into the structure and calibrated accordingly, which negatively impacts equipment cost and installation space.

[0007] In EP 1 837 623 A1, particularly for partially reflective objects, a method is known to resolve ambiguity by combining a deflection measurement of the directional reflection component with a "shape-from-shading" evaluation of the diffuse reflection component. However, this forcibly presupposes a sufficiently strong (partial) roughness of the surface to produce the desired diffuse reflection component, which is not given in the case of very smooth or completely glossy surfaces.

[0008] Another possibility for resolving the aforementioned ambiguity lies in assigning a definite orientation to each screen point, independent of the orientation of the respective object. For example, a method called orientation-encoded deflection measurement is described in DE 100 14 964A1. However, this method is mandatory because the screen must be mapped to infinity, and the principal beam used for observing the object must be parallel. The very large lens required for this screen mapping, along with the equally space-consuming and costly telecentric observation optics, renders this approach largely unusable in practical applications (especially in large image fields and / or large angular dynamic situations).

[0009] DE 10 2004 020 419 B3 indicates that ambiguity can also be resolved by utilizing observations from one or more additional cameras (the so-called "stereo or absolute deflection measurement method"). However, this approach is disadvantageous due to the higher cost of equipment and calibration, as well as the additional installation space required. Furthermore, for this scheme, the image fields of the cameras must be sufficiently overlapping. For good height resolution, a sufficiently large angle is also needed between the observation directions of the participating cameras, which necessitates a very large screen for each individual camera or the use of multiple individual screens. In many practical applications, this can only be implemented to a limited extent.

[0010] When using only a single camera (monocular view), it is theoretically possible to calculate the surface normal observed in one pixel for each (unknown) potential object height. In this way, a so-called "vector field of potential normals" is derived in the measurement volume. Under appropriate assumptions of continuity and differentiability of the object surface, it can be shown that, given a certain geometry of the measurement structure, this vector field can only be locally integrated on the "real" object surface. One criterion for local integrability is the "Frobenius condition." For example, this method is described in J. Kaminski's paper "Geometrische Rekonstruktionspiegelnder" (The Geometrische Rekonstruktionspiegelnder). aus deflektometrischen Messdaten”, Erlangen- (2008) or by SBWerling's "Deflektometrie zur automatischen" und Rekonstruktion spiegelnder ,Karlsruher Institut Technik (2011) published this paper. While it is theoretically possible to solve the height problem in this way using only a single camera view, this approach is practically unusable because the vector field of the underlying normal varies only very slightly and at a very low spatial frequency for typical structural geometries, and therefore the deviation from local integrability is practically unmeasurable in most cases. In “Local Shape from Mirror Reflections” by S. Savarese et al., International Journal of Computer Vision, 64(1), 31-67 (2005), a method is described for determining the differential geometric parameters of an object surface from a mapping of discrete intersecting curves on a screen at discrete points at zero-order (orientation / height), first-order (slope / normal), and second-order (II. basic shape / curvature). For this purpose, the straight edges of a binary checkerboard pattern with triangular or square fields are used as curves. The disadvantage of this method is that information can only be evaluated at individual discrete (spatially separated) points that are fixedly given in advance by the pattern. Furthermore, the positions of these discrete measurement points or support points (Stuetzpunkte) on the object surface (which will then apply to interpolation) depend on the shape of the object itself. Summary of the Invention

[0011] The objective of this invention is to configure and improve methods, systems, and measuring devices of the type mentioned at the beginning, so as to enable optical measurements of objects as comprehensively as possible, wherein the equipment costs and expenses should be kept as low as possible.

[0012] According to the present invention, the above-mentioned task is solved by the features of the following scheme. In the method discussed, a planar pattern is generated by means of a pattern generator, the pattern being modified in at least one optical property such that a large number of different points or a large number of different groups of points can be distinguished from each other, at least in sub-regions, wherein at least a portion of the pattern is reflected as a reflective pattern by the reflective surface of an object onto a detector of a camera unit, wherein the reflective pattern is converted into a camera image by the detector, wherein the relationship between points in the camera image and corresponding points in the pattern can be described by means of a correspondence function, the correspondence function being related to the geometric properties of the reflective surface of the object, and wherein at least one of the geometric properties of the reflective surface of the object is determined by using the differential geometric properties of the transformation given by the correspondence function.

[0013] Regarding the system, the above task is solved through the characteristics of the following scheme. Accordingly, the system under discussion includes:

[0014] A pattern generator for producing planar patterns, wherein the planar patterns are altered in at least one optical property such that a large number of distinct points or groups of distinct points can be distinguished from each other, at least in sub-regions.

[0015] A camera unit with a detector, wherein at least a portion of a pattern is reflected as a reflective pattern onto the detector of the camera unit via a reflective surface of an object, wherein the detector is configured to convert the reflective pattern into a camera image, and wherein the relationship between points in the camera image and corresponding points in the pattern can be described by means of a correspondence function relating to the geometric properties of the reflective surface of the object.

[0016] A computational unit configured to determine at least one of the geometric properties of the reflective surface of the object, wherein in this case, the differential geometric properties of the transformation given by the corresponding function are used.

[0017] Regarding the measuring device, the above task is solved by the features of the following scheme.

[0018] According to the invention, it is first recognized that by combining aspects of deflection measurement with phase measurement to provide comprehensive measurement data and a differential geometry scheme, a method can be achieved that allows for comprehensive measurement of objects with reflective and / or partially reflective surfaces with minimal equipment consumption, particularly without the introduction of additional cameras, additional measurement methods, and shifting mechanisms. This differential geometry scheme is described for the discrete intersections of intersecting curves in the aforementioned publication "Local Shape from Mirror Reflections" by S. Savarese et al. For this purpose, a planar pattern is generated by means of a pattern generator, the planar pattern being altered in its optical properties. Here, the optical properties of the pattern are altered such that a large number of distinct points or groups of points can be distinguished from each other, at least within sub-regions of the pattern. This means that a point or group of points differs from other points or other groups of points at least in terms of optical properties, such as brightness or color. In this way—similar to the case of deflection measurement with phase measurement—a pattern is formed such that, when observed with a camera, the position of the observed point or group of observed points within the pattern (or sub-regions of the pattern) can be definitively inferred from the reflective surface of the subject.

[0019] At least a portion of the pattern is reflected as a reflective pattern onto the detector of the camera unit via the reflective surface of the object. The detector of the camera unit converts the received reflective pattern into a camera image, which typically consists of a large number of pixels arranged in columns and rows. This camera image, together with the pattern generated by the pattern generator, forms the starting point for determining the geometric characteristics of the reflective surface of the object.

[0020] The relationship between the points in the camera image and the corresponding points in the pattern can be mathematically established using a correspondence function, especially using... The correspondence function is described here. It relates to the geometric characteristics of the reflective surface of the object; that is, depending on the configuration of the reflective surface, the pattern represented by the pattern generator is mapped in different ways and at different positions of the detector in the camera unit, which is reflected in the correspondence function. Thus, for example, in the case of a curved object, straight lines of the pattern are mapped onto curves whose curvature depends on the curvature of the surface, the emission direction of the reflective pattern, and other optical conditions. Therefore, if the parameters of the correspondence function are known, the geometric characteristics of the reflective surface can be inferred. Advantageously, the correspondence function is determined and one-to-one within the appropriately defined local environmental background (Umgebung) at each individual measurement point; global birefringence across the entire measurement field is not required.

[0021] To determine at least one of the geometric properties of the reflective surface of an object, differential geometric properties of transformations locally given by corresponding functions are used. This means that the corresponding functions are locally differentiated at least once, and information about the geometric properties of the reflective surface is obtained from consideration of the corresponding functions and their derivatives.

[0022] According to the invention, due to the distinguishability of planar patterns and points or groups of points, points of a camera image can be assigned to points of a pattern, and thus planar regions of the camera image can be assigned to (correspondingly distorted) regions of the image on the screen of the pattern generator, from which parameters of the corresponding function and / or differential geometric properties of the transformation given by the corresponding function can be calculated.

[0023] For example, the corresponding function can be locally represented in the form of a Taylor expansion (or other suitable polynomial approximation). In the case of a local approximation down to second order, this representation can be, for example, made with 10 parameters to be determined—4 parameters a. 11 a 12 a 21 a 22 Used for Jacobi matrices (first order) and 6 parameters a 13 a 14 a 15 a 23 a 24 a 25 For Hessian matrices (second order):

[0024] in

[0025] The length from the line of sight to the surface can be determined by solving a system of nonlinear equations with four unknowns (parameter s for zero order, i.e., the orientation of the object point in space along the respective line of sight of the camera, and three parameters a, b, and c for second order, i.e., the second basic shape of the object surface). In this configuration, a key feature is determining the coefficients of the Taylor expansion of the corresponding function (also referred to herein as the “transformation”), which exist in the form of comprehensive measurement data at each point to be evaluated within a properly defined local context using deflection measurements of phase measurements or other suitable pattern coding methods. This allows for the determination of (e.g., 10) parameters of the aforementioned transformation within this local context using an overdetermined system of equations in the form of a least-squares fit.

[0026] In principle, the method according to the invention can be used for a wide variety of objects having reflective and / or partially reflective surfaces. A “reflective surface” means that a beam of light incident at an angle of incidence is reflected at an exit angle (relative to the surface normal) corresponding to the angle of incidence. This reflection is also referred to as “direct reflection.” A “partially reflective surface” means that, in addition to the directly reflected component, a component of the incident beam is also diffusely reflected. This means that the exit angle of the diffuse component deviates from the angle of incidence relative to the surface normal. The extent of the directly reflected and diffusely reflected beam components is secondary in the case of a partially reflective surface. The only important factor is that the directly reflected beam component can be mapped onto the detector of the camera unit in a sufficiently distinguishable manner. Since the diffuse component—unlike EP 1 837 623 A1—does not need to be evaluated in the method according to the invention, no additional conditions need to be imposed on it. In this respect, it can also be as weak as desired.

[0027] Pattern generators can be implemented in a variety of ways. Importantly, the pattern generated by the pattern generator must be reproducible, meaning the representation of the pattern remains unchanged at different points in time. This can be achieved without using a pattern generator, as is known in practice. Reference should be made, by way of example only, to the use of TFT (Thin Film Transistor), OLED (Organic Light Emitting Diode), LCD (Liquid Crystal Display), or plasma screens, or to projecting patterns onto a screen or frosted glass, in order to list only a few available, but not limited, pattern generators.

[0028] The camera unit can also be configured in a variety of ways. Importantly, the camera unit can convert the pattern reflected onto the detector into a camera image. Here, it is appropriate that the mapping optics, detector, and other components of the camera unit are of such high quality that the camera image obtained through the detector represents the reflected pattern with the lowest possible distortion. In the case of components of lower quality, these components can be calibrated accordingly using standard methods of near-field photogrammetry. Furthermore, especially when used in production systems, it is appropriate that the camera unit is correspondingly robust and industrially suitable.

[0029] In the case of a pattern generator and a camera unit, it is suitable that the pattern generator and camera unit are each constructed using calibrated units. This means that distortions or changes in color representation that occur when generating a pattern or mapping a reflective pattern in a camera image are detected by measurement techniques and can be used for error correction. The corresponding calibration methods are well-known in practice.

[0030] In principle, various geometric properties of the reflective surface can be determined using the method according to the invention. Importantly, these geometric properties are reflected, at least indirectly, in the camera image and corresponding functions. Preferably, the geometric properties of the object's reflective surface are composed of height information and / or slope information and / or curvature information. Height information is the (absolute) position of a point on the reflective surface at which a beam of light originating from the pattern is reflected to the detector. This height information can be given, in particular, as the distance *s* between the point on the reflective surface and the illuminated point on the detector of the camera unit (or, in the case of calibration using an extended pinhole camera model). Slope information can be calculated from the height information using the first derivative, and curvature information can be calculated from the height information using the second derivative.

[0031] In one improved embodiment, within the calibration system, the slope information can be determined with high precision by evaluating it using a "classical" deflection measurement method along a line of sight known through calibration, with the elevation information s determined according to the invention, thereby improving the overall accuracy of the measurement. Therefore, in this case, the elevation information can also be improved until it becomes self-consistent through iterative integration.

[0032] In one configuration, when calculating the parameters of the corresponding function and / or when calculating the differential geometry of the transformation given by the corresponding function, the ambient background of a point in the camera image is used for that point. Here, in an improved embodiment, the "size" of the ambient background region can be made dependent on the curvature of the object. The more pronounced the curvature of the object's reflective surface, the smaller the ambient background can be selected. For example, the ambient background can be constructed using a 3×3 matrix, where the points defining the ambient background of the camera image are arranged in the middle element (2 2), and the immediate neighbors of that point are arranged on the remaining points (thus there will be 2×9 = 18 measurements used to determine 10 parameters describing the local transformation up to the second order corresponding function, approximated by Taylor expansion). A 5×5 matrix or larger ambient background is also conceivable.

[0033] In one improved approach, a curve defined by multiple points in the environment and background of a point in the camera image can be mathematically approximated. For example, such a curve can be represented by columns or rows of the matrix described above. Various approximation methods known in practice can be used for mathematical approximation. Taylor expansion is preferably used.

[0034] In one configuration, the curve defined by multiple points of the ambient background from which a point in the camera image is located is composed of straight lines. If we assume that the correspondence function maps points in the camera image to points in the pattern, then in the case of an object with a curved reflective surface, the straight lines in the camera image will be mapped to the curved curves in the pattern. In this way, any number of curve pairs can be evaluated for each measurement point in the camera image, as used in the aforementioned publication by Savarese. With the configuration of the pattern according to the invention, in an improved embodiment, it is possible to determine the corresponding curve of the pattern for the curve in the camera image when calculating the parameters of the correspondence function and / or when calculating the differential geometry of the transformation given by the correspondence function. Here, the corresponding curve of the pattern is composed of points of the pattern that, after reflection at the reflective surface of the object, produce the considered curve in the camera image. Since points or groups of points can be distinguished from each other by the optical properties of the pattern, this assignment is largely possible without problems. In this way, the position of each pattern point can be fully calculated. The prototype of the correspondence function, i.e., the curve in the camera image, in this case can be, for example, coordinate lines arbitrarily arranged in the chip plane of the detector. The graph of the corresponding function is then the (usually curved) curve derived from the points of the pattern observed separately. In this way, the locally unfavorable curve geometry that may occur when using a fixed, pre-given curve on the pattern, as in S. Savarese et al., “Local Shape from Mirror Reflections”, International Journal of Computer Vision, 64(1), 31-67 (2005), can be avoided.

[0035] In principle, the varying optical properties within a pattern can be constructed in a variety of ways. The important thing is that each optical property is reflected in the camera image and the point from which the beam of light originates in the pattern can be deduced as clearly as possible. How these optical properties are constructed is not decisive. However, the optical properties are preferably composed of the brightness and / or color of the points in the pattern. If the pattern is composed, for example, of grayscale values, then these optical properties will be, for example, different brightness levels of the points in the pattern. Depending on the color space used, such as 256 different grayscale levels (including white and black), a large number of distinguishable points or groups of points can be represented in this way. Different brightness levels of color are also conceivable, such as different brightness levels of green or blue hues. Optical properties can also be composed of color. In this case, a color distribution map can be used on a series of adjacent pattern points, for example, a color distribution map from yellow through green to blue. In this way, an even larger number of different points or groups of points can be distinguished. If the above two schemes, namely different brightness and different color, are additionally combined, the number of distinguishable points or groups of points can be further increased.

[0036] In one improved embodiment, the pattern of dots with the same optical properties can be formed by adjacent dots. These adjacent dots can, for example, form curves. In principle, these curves can be arbitrarily arc-shaped and of any width. However, these curves are preferably formed by straight lines, which are narrow, particularly preferably less than 1 cm, and very particularly preferably less than 3 mm.

[0037] In one improved scheme, the optical properties are changed quasi-continuously within the pattern. This means that the changes in optical properties are so small that the observer perceives the changes as a continuous transition. Here, a sinusoidal distribution pattern is preferred. The quasi-continuous changes can also be confined to sub-regions, such as stripes or squares.

[0038] In principle, patterns can be constructed in a variety of ways. Here, a pattern can be constructed such that every individual point in the entire pattern is distinguishable relative to another point in the pattern. In this way, a one-to-one assignment between the camera image and the pattern is possible. However, this usually means that the differences between the individual points are very close. In cases with a large number of distinguishable points, this can lead to these differences becoming practically indistinguishable in measurement techniques. Therefore, in an improved approach, sub-regions are constructed within the pattern, and the optical properties of the pattern are changed within these sub-regions. In this way, the bandwidth of the change in one or more optical properties can be limited to small sub-regions of the pattern. This is applicable if the individual sub-regions are constructed identically to each other or comprise fragments of sub-regions. The ambiguity formed between different sub-regions in this way can usually be easily grasped, for example, by means of a so-called "phase unwrapping" method.

[0039] The various sub-regions of the pattern can be arranged relative to each other in different ways. However, to simplify pattern generation, the sub-regions are preferably arranged in a grid. This grid can have grid elements of different designs. Since the overall pattern should be constructed planarly, the grid elements should ensure that a planar overall pattern can be formed, i.e., the grid elements should be able to be arranged together without gaps. Preferred configurations of grid elements here are squares, rectangles, triangles, and hexagons. However, more complex grid element designs can also be used.

[0040] The overall pattern or sub-regions of the pattern can be constructed in different ways. Here, in one configuration, the pattern is a reversed pattern adapted to the object to be measured and / or a (band-limited) random pattern and / or a color-coded pattern and / or a striped pattern with a brightness and / or color distribution map. The corresponding patterns are known from practice.

[0041] In one improved approach, the pattern is generated such that the screen coordinates are encoded within the pattern within the range of phase encoding known from the deflection measurement method. Especially when using a phase-shifted sinusoidal pattern or when two sinuses are superimposed and the reconstructed phase is subsequently demodulated (phase jump cleared), the phase of the sinusoidal pattern directly encodes the screen coordinates (as known from the phase-shifted deflection measurement method).

[0042] In one configuration, the pattern is configured such that the geometric properties of the reflective surface of an object can be determined from a single pattern and a single camera image. In another configuration, multiple different patterns are generated sequentially in time by a pattern generator. Here, the multiple patterns may be phased, rotated by a defined angle (e.g., 90°), flipped, or otherwise correlated with each other. The calculation of the parameters of the correspondence function and / or the calculation of the differential geometric properties of the transformation given by the correspondence function are repeated for each of the multiple different patterns. By appropriately selecting multiple patterns, it is possible to calculate the required parameters and / or properties in an iterative process, even when patterns are not sufficient to determine the parameters or differential geometric properties of the correspondence function. With each iteration, the calculation of the parameters and / or properties is thus supplemented or refined.

[0043] Depending on the number of points in the camera image and the complexity of the reflective surface, evaluating each point on the reflective surface can become very complex and computationally intensive. Therefore, in an improved approach, points in the camera image are intentionally undersampled, so that not every camera point is used, or only an average of multiple camera points is used. Support points are then formed by undersampling, on which interpolation or regression of the model surface can be performed. Using this simplified model surface, geometric properties at least similar to the real reflective surface can then be determined.

[0044] In particular, a system according to the invention for optically measuring an object, specifically configured to perform the method according to the invention, comprises a pattern generator, a camera unit, and a computing unit. The pattern generator is configured to generate a planar pattern, the planar pattern being altered in at least one optical property such that multiple distinct points or multiple distinct groups of points can be distinguished from each other, at least in sub-regions. The camera unit has a detector, wherein at least a portion of the pattern is reflected as a reflective pattern by the reflective surface of the object onto the detector of the camera unit, wherein the detector is configured to convert the reflective pattern into a camera image, and wherein the relationship between points in the camera image and corresponding points in the pattern can be described by means of a correspondence function relating to the geometric properties of the reflective surface of the object. This correlation means that the parameters of the correspondence function, which can be determined by measurement, depend on the geometric properties of the reflective surface. The computing unit is configured to determine at least one of the geometric properties of the reflective surface of the object, wherein in this case—that is, in determining the geometric properties—differential geometric properties of the transformation given by the correspondence function are used.

[0045] In principle, the arrangement of the camera unit and the pattern generator relative to each other is irrelevant as long as the pattern generator produces a pattern that can be reflected onto the detector of the camera unit through the reflective surface of the object. This premise can usually be satisfied relatively easily. However, it has been shown here that certain arrangements can lead to less desirable situations. Therefore, in an improvement of the system according to the invention, the camera unit and the pattern generator (i.e., the surface representing the pattern) are arranged relative to each other such that the viewing plane used and the pattern generator are not perpendicular to each other. Here, the viewing plane is spanned by the line of sight of the camera unit and the light beam formed by the pattern reflected at the reflective surface of the object. When the viewing plane is arranged perpendicular to the pattern generator, the objective function to be minimized in computation has a wide minimum value that is only badly performed. This is numerically disadvantageous and limits measurement accuracy in the case of noisy measurement data. Especially in the case of predominantly flat objects, this arrangement can be avoided by deliberately viewing the pattern generator from above the object at an angle.

[0046] In principle, the method and system according to the invention are constructed such that only a single camera unit is used. Even in the case of such a monocular system, one or more geometrical characteristics of the reflective surface of an object can be determined by a suitably constructed pattern. Most ambiguities formed in the periodic pattern can be eliminated without much problem. For this purpose, corresponding “phase unwrapping” methods known in the literature should be consulted, for example. Alternatively, in a particular configuration, such ambiguity caused by the periodic pattern can also be resolved by using the “Frobenius criterion,” which is robustly applicable in this case compared to a weakly varying potential normal field (as described in the papers mentioned in the introduction of the specification by J. Kaminiski or SB Werling), since the “candidate normals” calculated from the periodically offset screen points (due to ambiguity) typically vary very strongly outside the “real” object surface and thus very explicitly violate the local integrability criterion.

[0047] In one improved embodiment, the system includes multiple camera units arranged spaced apart from each other. Each of these cameras converts the pattern reflected at the reflective surface of the object into a camera image. For each of these individual camera images, the geometric properties of the object's reflective surface can be determined independently. If ambiguity is to be created, the results of the evaluations of the different camera images can be combined, thereby reducing or eliminating ambiguity. Attached Figure Description

[0048] Various possibilities now exist for configuring and improving the teachings of the present invention in an advantageous manner. For this purpose, reference can be made, on the one hand, to the foregoing solutions, and on the other hand, to the following description of preferred embodiments of the invention with reference to the accompanying drawings. General preferred configurations and improvements of the teachings are also described in conjunction with the description of preferred embodiments of the invention with reference to the accompanying drawings.

[0049] In the attached diagram

[0050] Figure 1 A schematic diagram of a deflection measurement system is shown to illustrate the high degree of ambiguity in the deflection measurement method.

[0051] Figure 2 The image shown is a first camera phase image obtained from a reflection pattern with a sinusoidal fringe distribution in a first orientation.

[0052] Figure 3 The image shows a second camera phase image obtained from a reflection pattern with a sinusoidal fringe distribution at a second orientation perpendicular to the first orientation.

[0053] Figure 4 Showing according to Figure 2 and 3 The additive combination of sinusoidal stripe patterns, the pattern reflected at the reflective surface of the object, and the phase image from a third camera.

[0054] Figure 5 shows a sinusoidal cross pattern ( Figure 5A ), camera images ( Figure 5B ) and the 3×3 environment background of the camera image points ( Figure 5C A diagram illustrating the correspondence between curves in camera images and patterns, and

[0055] Figure 6 A schematic diagram of an embodiment of a system according to the present invention, having a double sine pattern, is shown. Detailed Implementation

[0056] Figure 1 A schematic diagram illustrating a typical deflection measurement scenario is shown. Pattern generator 1 generates a pattern that is reflected as a reflective pattern at the reflective surface 2 of object 3 into camera unit 4. A detector (not shown) in camera unit 4 generates a camera image from the reflected pattern. The camera's line of sight (formed by the optical axis of the camera optics) is drawn using reference numeral 5. The point q visible in the camera image, derived from point p of the pattern by reflection at reflective surface 2, can be generated by different light beams, wherein... Figure 1 Three different possibilities are illustrated in the figure. Figure 1 Each of the beams 6, 6', and 6" shown can cause a reflected beam 7 and thus an illuminated point q on the detector of camera unit 4. Different heights are derived depending on which of the beams 6, 6', and 6" is correct. Since the angles of the incident beams 6, 6', and 6" relative to the surface normal and the angle of the reflected beam 7 relative to the surface normal are the same in each case, different surface normals 8, 8', and 8" are derived based on the correct beams 6, 6', and 6" respectively. It can be seen that this ambiguity must be resolved in order to correctly determine the geometric characteristics of the reflecting surface 2: "height" and "slope".

[0057] exist Figures 2 to 4 Various camera images 9, 9', and 9'' are shown in the figure. These images, exemplified by a phase-measured deflection method, are formed by reflecting a pattern generated by pattern generator 1 onto the slightly curved reflective surface 2 of object 3. Figure 2 In the image, camera image 9 consists of horizontally oriented curved stripes with a brightness distribution map that can be obtained from the original sinusoidal brightness distribution map after appropriate phase evaluation. Figure 3In this process, the pattern is rotated 90° clockwise to form vertical curved stripes, which have a brightness distribution pattern as obtained from the original sinusoidal brightness distribution pattern after appropriate phase evaluation. Figure 4 In the middle, the combination comes from Figure 2 and Figure 3 The two patterns in the camera image 9” constitute a combination of horizontally curved stripes and vertically curved stripes. Furthermore, the positions of the same phases 11 and 12 are marked in color, with the position of the same phase 11 indicated in green in the horizontal direction and the position of the same phase 12 indicated in red in the vertical direction. In each of the shown camera images 9, 9', and 9”, a sub-region 11 is formed in which at least one optical property of the pattern—i.e., the phase of the brightness distribution map of the pattern with a respective sinusoidal variation—changes, such that a set of points—i.e., lines with the same phase—can be inferred on the camera image. All three camera images 9, 9', and 9” can be used to determine the geometry of the reflecting surface using the differential geometry of the transformation given by the corresponding function, and thus determine the point cloud describing the surface structure.

[0058] Figure 5 shows that the correspondence function can be locally approximated as a Taylor expansion in the context of the environment. In this case, the prototype of the correspondence function is a coordinate line arranged arbitrarily in the detector plane, and the image is a (typically curved) curve derived from the separately observed phase values ​​φ on the screen. Figure 5A The image shows pattern 13, which has multiple sub-regions 10” in a square grid pattern. Within each sub-region 10”, a double sinusoidal distribution pattern of optical characteristics is constructed, which in this case is composed of the brightness of the pattern points. Specifically, the double sinusoidal distribution pattern consists of two sinusoidal patterns offset from each other by 90°, which are derived after appropriate phase evaluation. Figure 4 Camera images. Figure 5B A segment of detector 14 of camera unit 4 is schematically shown, where each pixel 15 is positioned by a midpoint 16. Figure 5B In addition, point p and its immediate neighbors (3×3 ambient background) are drawn, with a total of 9 points connected to each other by horizontal or vertical lines, forming curve 17 in the camera image. Figure 5C In the image, the 3×3 environment background is shown again in a magnified view, with variables used for matrix representation plotted at each point. If we search for the corresponding point of pattern 13 for each point in a 3×3 background environment, then for example, we can obtain... Figure 5A Curve 18 is shown in the figure.

[0059] Figure 6 The system shown in the invention has a mapping geometry and largely corresponds to the mapping geometry. Figure 1 The representation in the text. However, by using specially configured patterns and by calculating differential geometric properties, the high degree of ambiguity can now be eliminated.

[0060] For other advantageous configurations, please refer to the general section of the instruction manual to avoid repetition.

[0061] Finally, it should be clearly stated that the above exemplary embodiments are only used to discuss the claimed teachings and are not intended to limit the teachings to the embodiments.

[0062] List of reference numerals

[0063] 1. Pattern Generator

[0064] 2. Reflection indicates

[0065] 3 objects

[0066] 4 camera units

[0067] 5. Visual axis

[0068] 6 beams

[0069] 7 reflected beam

[0070] 8 Surface Normals

[0071] 9. Camera images

[0072] 10 sub-regions

[0073] 11. Positions of the same phase (horizontal)

[0074] 12 positions with the same phase (perpendicular)

[0075] 13 Patterns

[0076] 14 detectors

[0077] 15 pixels

[0078] 16 (pixels) midpoint

[0079] 17 (on the detector / in the camera image) curve

[0080] 18 (in the pattern) curve.

Claims

1. Method for optical measurement of an object having a reflective and / or partially reflective surface, wherein - generating a planar pattern (13) by means of a pattern generator (1), which pattern is varied in at least one optical property, such that a large number of different points (p) or a large number of different point groups can be distinguished from one another at least in subregions (10), - wherein at least a part of the pattern (13) is reflected by the reflective surface (2) of the object (3) as a reflected pattern onto a detector (14) of a camera unit (4), which reflected pattern is converted into a camera image (9) by the detector (14), a relationship between points (q) of the camera image (9) and corresponding points (p) of the pattern (13) can be described by means of a correspondence function, which correspondence function is related to geometric properties of the reflective surface (2) of the object (3), and - wherein at least one of the geometric properties of the reflective surface (2) of the object (3) is determined by using differential geometric properties of a transformation given by the correspondence function, - wherein, when calculating parameters of the correspondence function and / or when calculating differential geometric properties of a transformation given by the correspondence function, for a curve (17) in the camera image (9) a corresponding curve (18) of the pattern (13) is determined, wherein the corresponding curve (18) of the pattern (13) consists of points of the pattern (13) which, after reflection at the reflective surface (2) of the object (3), produce a curve (17) in the camera image (9).

2. The method of claim 1, wherein, The geometric properties of the reflective surface (2) of the object (3) comprise height information and / or slope information and / or curvature information.

3. The method according to claim 1 or 2, characterized in that, When calculating parameters of the correspondence function and / or when calculating differential geometric properties of a transformation given by the correspondence function, for a point (q) of the camera image (9) the ambient context of this point in the camera image (9) is used.

4. The method of claim 3, wherein, Curves (17) defined by a plurality of points of the ambient context of a point (p) of the camera image (9) are mathematically approximated, wherein preferably a Taylor expansion is used.

5. The method according to claim 1 or 2, characterized in that, The at least one optical property varied within the pattern (13) consists of the brightness and / or the color and / or the phase of a sinusoidal brightness profile of the points of the pattern (13).

6. The method according to claim 1 or 2, characterized in that, Point groups of the pattern (13) having the same optical property consist of adjacent points, which adjacent points are preferably arranged on a curve.

7. The method according to claim 1 or 2, characterized in that, The optical property within the pattern (13) is varied quasi-continuously, preferably sinusoidally, at least in subregions (10), such that points of the pattern are encoded, preferably by the local phase of a sinusoidal brightness profile.

8. The method of claim 1 or 2, wherein, The pattern (13) has subregions (10) within which the optical property of the pattern (13) is varied, wherein the individual subregions (10) are preferably constructed identically to one another.

9. The method of claim 8, wherein, The subregions (10) are arranged in a grid, wherein the grid is particularly preferably composed of square, rectangular, triangular or hexagonal grid elements.

10. The method of claim 1 or 2, wherein, For the pattern (13) an inverse pattern and / or a band-limited random pattern and / or a color-coded pattern and / or a stripe pattern with a brightness and / or color profile is used.

11. The method of claim 1 or 2, wherein, The pattern (13) is generated such that screen coordinates in the pattern (13) are encoded.

12. The method of claim 1 or 2, wherein, A plurality of different patterns (13) is successively generated by the pattern generator (1), wherein the plurality of patterns (13) is preferably phase-shifted with respect to each other, such that the calculation of the parameters of the correspondence function and / or the calculation of the differential geometric properties of the transformation given by the correspondence function is repeated for the changed pattern (13), wherein the plurality of patterns (13) is preferably chosen such that the calculated parameters and / or properties are complemented or refined with the calculation for one of the plurality of patterns (13).

13. The method of claim 1 or 2, wherein, The points in the camera image (9) are undersampled and an interpolation or regression of the model surface is performed for the support points thus obtained.

14. The method of claim 1 or 2, wherein, In case of ambiguities in the encoding of the pattern, the ambiguities are resolved by applying a Frobenius criterion or other suitable local integrability or smoothness criterion to the vector field of potential ambiguity normals due to the periodicity of the pattern.

15. A system for optical measurement of an object having a reflective and / or partially reflective surface, in particular for carrying out the method according to any one of claims 1 to 14, the system comprising: a pattern generator (1) for generating a planar pattern (13) which is varied in at least one optical property such that a large number of different points or a large number of different point groups can be distinguished from each other at least in sub-areas (10), a camera unit (4) having a detector (14), wherein at least a portion of the pattern (13) is reflected as a reflected pattern onto the detector (14) of the camera unit (4) by the reflective surface (2) of the object (3), wherein the detector (14) is configured for converting the reflected pattern into a camera image (9), and a relationship between points of the camera image (9) and corresponding points of the pattern (13) can be described by means of a correspondence function, which is related to geometric properties of the reflective surface (2) of the object (3), a calculation unit configured for determining at least one of the geometric properties of the reflective surface (2) of the object (3), in this case using differential geometric properties of the transformation given by the correspondence function.

16. The system of claim 15, wherein, The camera unit (4) and the pattern generator (1) are arranged relative to each other such that the viewing plane used for evaluation and the pattern generator (1) are not perpendicular to each other, wherein the viewing plane is spanned by the viewing axis (5) of the camera unit (4) and the light beam formed by the pattern (13) reflected at the reflective surface (2) of the object (3).

17. The system of claim 15 or 16, wherein A plurality of camera units (4) arranged at a distance from each other, which respectively convert the pattern (13) reflected at the reflective surface (2) of the object (3) into a camera image (9), and from the plurality of camera images (9) determine the geometric properties of the reflective surface (2) of the object (3). In case of ambiguities in the encoding of the pattern, the ambiguities are resolved by applying a Frobenius criterion or other suitable local integrability or smoothness criterion to the vector field of potential ambiguity normals due to the periodicity of the pattern.

18. A measuring device comprising the system according to any one of claims 15 to 17 and an object to be measured.

Images