Method and system for optically measuring object with specular reflecting surface and / or partially specular reflecting surface as well as measuring device corresponding thereto

Abstract

To provide a method for optically measuring an object with a specular reflecting surface and / or a partially specular reflecting surface.SOLUTION: A two-dimensional pattern 13 in which optical characteristics change in the pattern is generated by a pattern generator 1 so that a plurality of different points or a plurality of different point groups can be distinguished from each other in a sub-region. At least part of the pattern 13 is reflected as a reflection pattern on a detector of a camera unit 4 by a reflecting surface 2 of an object 3, and the reflection pattern is converted into a camera image by the detector. The relationship between a point in the camera image and a point in the pattern 13 corresponding to the point can be described by a corresponding function that depends on the geometric characteristics of the reflecting surface 2 of the object 3. The geometric characteristics of the reflecting surface 2 of the object 3 are determined by using differential geometric characteristics in the conversion used for the corresponding function. The present invention further relates to a system and a measuring device corresponding thereto.SELECTED DRAWING: Figure 6

Description

[Technical field]

[0001] The present invention relates to a method and system for optically measuring objects having specular and / or partially specular surfaces and a corresponding measurement device. [Background technology]

[0002] These types of methods, systems and measuring devices are used for optically measuring the surface profile of objects having specular or partially specular freeform surfaces over the entire surface. Applications can be found in the automotive industry (e.g. for painted body parts), the optical industry (e.g. for the measurement of lenses and mirrors) and even the consumer industry (e.g. for the measurement of mobile phone and tablet displays), to name a few.

[0003] Phase measuring deflectometry is a well-established and frequently used method for such measurements. The fundamentals of deflectometry are described in US Pat. No. 5,399,633. Typically, a sinusoidal stripe pattern is displayed on a calibrated screen and viewed with one or more calibrated cameras through the mirrored surface of the object to be measured. By performing appropriate phase estimation in multiple directions, it is possible to estimate the coordinates of the pattern points of the observed object on the screen for each camera pixel. Under certain conditions, by tracing the camera's line of sight to a point on the screen of the object being viewed, it is possible to infer the surface normal, and by numerical integration, the shape of the viewed object can also be inferred.

[0004] A well-known problem with this method is the so-called height ambiguity problem. Since each camera pixel only contains information about the position of the observed screen point, and not about the beam direction, the surface normal can only be unambiguously determined if the object's position ("height") is known. As a result, if the height assumptions are incorrect, the surface normals will be incorrect as well. This ambiguity, also called the "deflectometry regularization problem or height problem", can be solved by different approaches, all of which have certain drawbacks.

[0005] In the approach described in [1], the unknown beam direction is determined by a second measurement. In this second measurement, the screen position is shifted vertically with respect to the screen position in the first measurement. However, this approach requires a complex mechanism for shifting the screen, in addition to highly calibrating the device at both screen positions. This approach also results in longer measurement times.

[0006] In another approach, the unknown object position can be determined by making an additional absolute measurement for a point using an independent measurement method. The measurement can be optical (eg, using a triangulation sensor) or tactile. The unknown heights of other points on the surface can be determined in a self-consistent manner by estimating their respective heights based on the measured points on the surface, calculating their surface normals, and iteratively integrating these surface normals. Such a procedure is described, for example, in "Chemical Chemistry and Biomolecular Biology", vol. A drawback of this method is that the measurement device must incorporate an additional measurement of the height of the point and must be calibrated accordingly. This has a negative impact on the cost and installation space of the measuring device.

[0007] US Pat. No. 5,399,433 describes a method for partially specularly reflecting objects. This method solves the height ambiguity problem by combining deflectometry measurements of the directional reflectance parts with "shape from light" evaluation of the diffuse reflectance parts. However, this method requires that the surface roughness is (partially) sufficiently rough so that the necessary diffuse reflection portions are formed. This is not the case for very smooth or perfectly glossy surfaces.

[0008] Assigning a unique orientation to each screen point, regardless of object position, could potentially solve the ambiguity problem mentioned above. Such a method is known as directionally ordered deflectometry and is described, for example, in US Pat. No. 5,399,633. However, this method requires that the mapping onto the screen is very large and that the chief rays for observing the object are parallel. The need for very large lenses to image the screen and the need for space-consuming and costly telecentric viewing optics mean that this approach is usually not feasible for practical applications, especially when large image fields or large dynamic angles are required.

[0009] US Pat. No. 5,399,633 indicates that this ambiguity problem can also be solved by observation with one or more additional cameras (so-called "stereodeflectometry or absolute deflectometry"). However, the drawback of this method is that it requires additional installation space, as well as more complex equipment and calibration. This approach also requires that the image fields of the cameras overlap sufficiently. To obtain good resolution, the angle between the viewing directions of the associated cameras must also be large enough. This would require either a very large screen or multiple separate screens for each individual camera. In many practical applications, this is limited in feasibility.

[0010] When working with only one camera (monocular view), the surface normal observed to within one pixel can be calculated for every (unknown) potential height of the object. This gives rise to a so-called "potential normal vector field" in the measurement volume. Assuming that the object surface is suitably continuous and differentiable, it can be seen that, for a particular geometry of the measuring device, this vector field is only locally integrable on the "real" object surface. One criterion for local integrability is the "Frobenius condition". This method is described in academic papers, for example, in Non-Patent Document 3 and Non-Patent Document 4. Although the height problem could theoretically be solved this way with a single camera view, this approach is impractical. The reason is that for typical structures, the potential normal vector field varies only slightly at very low spatial frequencies, so that the departures from local integrability are usually not measurable in practice. Non-patent document 5 describes a method for determining the zeroth (position / height), first (slope / normal) and second (II basic shape / curvature) differential geometric parameters of an object surface at discrete points from the mapping of discrete intersection curves on a screen. Straight edges of a two-color checkerboard pattern with triangular or square sections are used as curves. A drawback of this method is that the individual discrete (spatially separated) points are fixedly predetermined by the pattern and the information can only be evaluated at those discrete points. The locations on the object surface at which these discrete measurements are taken, and the locations of the interpolation points (which are then suitably interpolated), further depend on the shape of the object itself. [Prior art documents] [Patent documents]

[0011] [Patent Document 1] DE 19944354 A1 [Patent Document 2] European Patent Application Publication No. 1837623(A1) [Patent Document 3] German Patent Application Publication No. 10014964(A1) [Patent Document 4] German Patent Application Publication No. 102004020419(B3) [Non-Patent Document]

[0012] [Non-Patent Document 1] 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) [Non-Patent Document 2] E. Slogsnat et al. “Non-stereoscopic Method for Deflectometric Measurement of Reflecting Surfaces”, Proc. DGaO 110 A11 (2009) [Non-Patent Document 3] J. Kaminski “Geometrische Rekonstruktion spiegelder Oberflaechen aus deflektometrischen Messdaten [Geometric reconstruction of specular surfaces from deflectometric measurement data]”, University of Erlangen-Nuremberg (2008) [Non-Patent Document 4] SB Werling “Deflectometry for automatic visual inspection and reconstruction of specular surfaces]”, Karlsruhe Institute of Technology (2011) [Non-Patent Document 5] S. Savarese et al. “Local Shape from Mirror Reflections”, International Journal of Computer Vision, 64(1), 31-67 (2005) Summary of the Invention [Problem to be solved by the invention]

[0013] The object of the present invention is to construct and further develop a method, a system and a measuring device of the type mentioned at the beginning so as to enable optical measurement of an object over as full a surface as possible while keeping the complexity and costs of the equipment as low as possible. [Means for solving the problem]

[0014] The above mentioned object is achieved according to the invention by means of the features of claim 1. The method of the present invention involves using a pattern generator to generate a two-dimensional pattern in which optical properties vary within the pattern such that different points or groups of points within a sub-region are distinguishable from one another, at least a part of the pattern is reflected by a reflective surface of an object onto a detector of a camera unit as a reflection pattern, the reflection pattern is transformed by the detector into a camera image, the relationship between a point in the camera image and a corresponding point in the pattern can be described by a correspondence function that depends on geometric properties of the reflective surface of the object, and the geometric properties of the reflective surface of the object are determined using differential geometric properties within a transformation used for the correspondence function.

[0015] Regarding the system, the aforementioned object is achieved by the features of claim 16. According to claim 16, the system comprises: a pattern generator for generating a two-dimensional pattern in which at least one optical property is varied within the pattern such that different points or different groups of points within the sub-regions are distinguishable from one another; a camera unit including a detector, at least a portion of the pattern being reflected by a reflective surface of the object onto a detector of the camera unit as a reflection pattern, the detector being configured to convert the reflection pattern into a camera image, a relationship between a point in the camera image and a corresponding point in the pattern being describable by a correspondence function that depends on geometric properties of the reflective surface of the object; and a calculation unit configured to determine geometric properties of the reflective surface of the object using differential geometric properties within a transformation used for the correspondence function.

[0016] The object of the measuring device described above is achieved by the features of claim 18.

[0017] It was initially recognized that the method of the present invention combines several aspects of phase measuring deflectometry to provide full surface measurement data and the differential geometric approach described in the publication of [5] for each discrete intersection point of intersecting curves, thereby enabling full surface measurement of objects with specular and / or partially specular surfaces, while at the same time enabling an approach with low complexity in terms of the measurement equipment, in particular without the need for additional cameras, additional measurement methods or shifting machines. For this purpose, a two-dimensional pattern, in which the optical properties vary within the pattern, is generated by a pattern generator. The optical properties within the pattern vary such that different points or different groups of points are distinguishable from one another in sub-regions within the pattern. This means that a point or a group of points differs from other points or groups of points with respect to at least one optical property, for example the brightness or color of the point. Similar to phase measurement deflectometry, this method allows one to unambiguously infer the location of an observation point or points within a pattern (or within a subregion of the pattern) when the pattern is observed with a camera through a reflective surface of an object.

[0018] At least a portion of this pattern is reflected by a reflective surface of the object as a reflective pattern to a detector in the camera unit. A detector in the camera unit converts the received reflection pattern into a camera image. A camera image is usually composed of a number of pixels arranged in a matrix. This camera image together with the pattern generated by the pattern generator is the starting point of the process of determining the geometric properties of the object's reflective surface.

[0019] The relationship between points in the camera image and corresponding points in the pattern is JPEG2025029169000002.jpg6168 It can be mathematically described by: This correspondence function depends on the geometric properties of the object's reflecting surface. That is, the pattern provided by the pattern generator is mapped differently to different locations within the detector of the camera unit depending on the configuration of the reflective surface. This is reflected in the correspondence function. For example, for a curved object, the straight lines in the pattern are mapped to curves, the curvature of which depends on the curvature of the reflective surface, the beam direction of the reflective pattern, and other optical conditions. Therefore, knowing the parameters of the correspondence function, the geometric properties of the reflecting surface can be inferred. It is advantageous if this correspondence function is well-defined and unique in the local surrounding area of ​​each individual measurement point, and does not have to be globally bidirectional over the entire field of measurement.

[0020] A differential geometric property within a transform locally applied to the correspondence function is used to determine at least one geometric property of the reflecting surface of the object. This means that the correspondence function is derived locally at least once and information about the geometric properties of the reflecting surface is obtained by considering the correspondence function and its derivatives.

[0021] Points in the camera image can be assigned to points in the pattern based on the two-dimensional pattern and configuration according to the present invention that makes the points or point clouds identifiable. Thus, two-dimensional regions in the camera image can be assigned to (correspondingly distorted) regions in the pattern on the screen of the pattern generator, and the parameters of the correspondence function and / or the geometric differential properties of the transformation used for the correspondence function can be calculated therefrom.

[0022] The correspondence function can be locally expressed, for example, in the form of a Taylor expansion (or other suitable polynomial approximation). For local approximations up to second order, this representation can be expressed, for example, by determining and using ten parameters, namely, four parameters a11, a12, a21, a22 of the Jacobian matrix (first order) and six parameters a13, a14, a15, a23, a24, a25 of the Hessian matrix (second order). JPEG2025029169000003.jpg28169 The line-of-sight length to the surface can be determined by solving a nonlinear simultaneous equation with four unknowns: one parameter of order 0, s, i.e., the spatial position of a point on the object along each of the camera's lines of sight, and three parameters of order 2, a, b, c, i.e., the second fundamental shape of the object surface. The feature in this construction is to determine the coefficients of the Taylor expansion (also referred to herein as the "transformation") of the correspondence function. Here, each point, which is evaluated in the form of full surface measurement data within a suitably determined local surrounding area, is analyzed using phase measuring deflectometry or other suitable pattern analysis methods. In this way, the (eg, 10) parameters in the above transformation can be determined by using the least squares method on an overdetermined system of simultaneous equations in the local surrounding area.

[0023] The method according to the invention can be used with a wide variety of objects having specular and / or partially specular surfaces. By "specular reflective surface" it is meant that an incident ray of light incident at an angle of incidence is reflected at an angle of reflection that matches the angle of incidence. For each reflection, the surface normal is the reference. This reflection is also called "direct reflection." By "partially specular surface" it is meant that in addition to directly reflecting a portion of the incident light, the surface also diffusely reflects a portion of the incident light. This means that the angle of reflection of the diffusely reflected portion is at an angle relative to the surface normal that is different from the angle of incidence. The magnitude of the proportion of directly reflected light and diffusely reflected light of a partially specular surface is not critical. All that matters is that the directly reflected portion of the light beam can be mapped onto the camera unit's detector in a sufficiently distinct manner. In the method according to the present invention, unlike Patent Document 2, the diffuse reflection portion is not necessary for the evaluation, so no further conditions need to be imposed. Thus, the diffusely reflected portion of the light beam can be weakened if desired.

[0024] The pattern generator can also be implemented in a variety of ways. It is important that the patterns generated by the pattern generator can be produced in a reproducible manner. That is, even if the time is different, the pattern display does not change. This can be achieved with a variety of pattern generators known in the art. Some examples of pattern generators include using TFT (thin film transistor), OLED (organic light emitting diode), LCD (liquid crystal display), plasma screens, or projecting the pattern onto a screen or frosted glass.

[0025] The camera unit can likewise be configured in a variety of ways. It is important that the camera unit is able to convert the pattern reflected on the detector into a camera image. In this regard, it is recommended that the mapping optics, the detector and other components of the camera unit are of high quality so that the camera image acquired by the detector displays the reflection pattern with as little distortion as possible. The lower quality components can be adequately calibrated using standard near-field photogrammetry methods. Also, for use in manufacturing facilities, it is recommended that the camera unit be suitably rugged and suitable for industrial use.

[0026] It is recommended that both the pattern generator and the camera unit be constructed as calibrated units. This means that distortions and changes in color reproduction that occur during pattern generation and mapping of the reflected pattern onto the camera image can be measured and used to correct for errors. Suitable calibration methods are well known in practice.

[0027] The method according to the invention can be used to determine various geometrical properties of a reflecting surface. It is important to note that the geometric properties are indirectly reflected in the camera images and the correspondence functions. The geometrical properties of the object's reflective surface consist of height information and / or slope information and / or curvature information. The height information is the (absolute) position of the point on the reflecting surface where the light ray emanating from the pattern is reflected towards the detector. This height information is available via the distance s between a point on the reflecting surface and the illuminated point on the camera unit's detector (or the center of the camera in case of calibration using the extended pinhole camera model). The slope information can be calculated from the height information using a first derivative, and the curvature information can be calculated from the height information using a second derivative.

[0028] A further aspect is that in a calibrated system the tilt information can be determined with high precision by a “classical” deflectometry evaluation, tracing rays along a line of sight known by calibration, using the height information s determined according to the invention. Thus improving the overall accuracy of the measurement. The height information can be refined by iterative integration until it becomes self-consistent.

[0029] The calculation of the parameters of the correspondence function and / or the differential geometric properties within the transformation to be used for the correspondence function for a point in the camera image uses the area surrounding this point in the camera image. A further aspect is to make the "size" of the surrounding region dependent on the curvature of the object. The stronger the curvature of the reflective surface, the smaller the size of the selected surrounding area can be. The surrounding area can be configured, for example, as a 3x3 matrix. This places the point in the camera image that defines the surrounding area in the central element (2,2) and assigns its direct neighbors to the remaining points (thus, 2x9=18 measurements are available to determine the 10 parameters of the local transformation that describes the correspondence function up to second order, approximated by a Taylor expansion). A 5x5 matrix or a larger surrounding area is also contemplated.

[0030] Furthermore, the curve defined by a number of points in the area surrounding a point in the camera image can be mathematically approximated. This curve can be formed, for example, by the columns or rows of the matrix representation mentioned above. For the mathematical approximation, various approximation means known in practice can be used. The use of Taylor expansion is recommended.

[0031] The curve defined by a number of points in the area surrounding a point in the camera image is formed by straight lines. Assuming that the correspondence function maps points in the camera image to points in the pattern, for objects with curved reflective surfaces, straight lines in the camera image will be mapped to curved lines in the pattern. In this way, any number of pairs of curves can be evaluated for each measurement point in the camera image, as used in [5]. Furthermore, by configuring the pattern in the present invention, for a curve in the camera image, a corresponding curve in the pattern can be determined when calculating the parameters of the correspondence function and / or when calculating the differential geometric properties in the transformation used for the correspondence function. A corresponding curve in the pattern is defined by a number of points in the pattern. These points are the points which, after being reflected by the reflective surfaces of the object, form the curve under consideration in the camera image. This assignment can generally be made without any problems since the points or points can be distinguished from one another by their optical properties within the pattern. Therefore, the position of each point in the pattern can be calculated over the entire surface. The original image of the correspondence function, i.e. the curve in the camera image, can be, for example, a coordinate line arbitrarily placed on the chip surface of the detector. In this way, the image of the correspondence function is a (usually curved) curve consisting of each observation point in the pattern. It is therefore possible to avoid locally disadvantageous curve shapes when using a predetermined fixed curve on the pattern, as described in Non-Patent Document 5.

[0032] The varying optical properties within the pattern can be created in a variety of ways. It is important that each optical characteristic is reflected in the camera image, allowing the light ray generation points within the pattern to be estimated as clearly as possible. It is not important what this optical property consists of. However, preferably the optical properties consist of the brightness and / or colour of the points in the pattern. For example, if the pattern is made up of grey values, these optical properties are the differences in luminance of the points in the pattern. Thus, depending on the color space used, for example 256 different grey levels (including black and white), a large number of distinct points or point clouds can be displayed. The colors may also have different intensities, for example varying degrees of green or blue intensity. The optical property can also consist of color. A color curve is available for successive adjacent points in the pattern. For example, a color curve that goes from yellow to green to blue. Thus, in addition, many different points or points can be identified. Further combining the two approaches mentioned above, i.e. combining different intensities with different colors, can further increase the number of distinguishable points or point clouds.

[0033] Furthermore, a pattern of points having the same optical properties can be formed by multiple adjacent points. Such a plurality of adjacent points may, for example, form a curve. These curves may have any desired curvature and may have any desired width. However, these curves are preferably formed by thin straight lines, the width of which is particularly preferably less than 1 cm, very particularly preferably less than 3 mm.

[0034] Furthermore, the optical properties vary quasi-continuously within the pattern. This means that the change in optical properties is so slight that the observer perceives the change as a continuous transition. The use of a sine curve is recommended. The quasi-continuous variation can also be limited to sub-regions, for example stripe or square shapes.

[0035] The pattern can be formed in a variety of ways. The pattern can be constructed such that every individual point throughout the pattern is distinguishable from other points within the pattern. Therefore, a unique association between the camera image and the pattern is possible. However, this means that the differences between individual points are usually very small relative to one another. When there are many distinguishable points, making the differences between the points very slight may mean that the differences cannot be resolved in practical measurements. Thus, sub-regions are formed within the pattern, and within the sub-regions the optical properties of the pattern vary. This allows the area with altered optical properties to be restricted to a small sub-region within the pattern. It is useful if the individual sub-regions have identical structure to each other or are made up of portions of sub-regions. Thus, ambiguities arising between different sub-regions can usually be easily resolved, for example by performing a phase disambiguation procedure.

[0036] The individual sub-regions within the pattern can be positioned relative to one another in a variety of ways. However, to simplify pattern generation, the sub-regions are preferably arranged in a grid. The grid may be made up of grid elements of different structures. However, since the entire pattern needs to be two-dimensional, the grid elements need to be capable of creating the entire pattern in two dimensions. In other words, the grid elements need to be able to be arranged without any gaps. Preferred grid element configurations are square, rectangular, triangular and hexagonal. However, more complex grid elements may also be used.

[0037] The entire pattern or sub-regions within the pattern can be structured in various ways. The pattern may be an inverse pattern adapted to the object being measured and / or a (band-limited) stochastic pattern and / or a colour-coded pattern and / or a stripe pattern with a brightness curve and / or a stripe pattern with a colour curve. Suitable patterns are known in practice.

[0038] Furthermore, the pattern is generated in such a way that the analysis from coordinates in the screen to the pattern can be performed using the phase analysis known from deflectometry. In particular, when using a phase-shifted sinusoidal pattern or a superposition of two sinusoids and then reconstructing the original phase from the shifted phase (resolving the phase shift), the phase of the sinusoidal pattern (as known from phase-shifting deflectometry) is analyzed directly from its coordinates in the screen.

[0039] The pattern is constructed such that the geometric properties of the object's reflective surface can be determined from a single pattern and a single camera image. In another arrangement, different patterns are generated in time sequence by a pattern generator. In the process, the multiple patterns may be phase shifted relative to one another, rotated by a defined angle (eg, 90°), inverted, or otherwise correlated with one another. The calculation of the parameters of the correspondence function and / or the differential geometric properties within the transformation used for the correspondence function may then be repeated for each of a number of different patterns. Even for patterns for which the parameters of the corresponding functions or the differential geometric properties cannot be fully determined, it may be possible to calculate the necessary parameters and / or properties by iterative processing by appropriately selecting multiple patterns. With each iteration, the calculation of parameters and characteristics may be supplemented and / or made more accurate.

[0040] Depending on the number of points in a camera image and the complexity of the reflective surface, evaluating individual points on the reflective surface can become very complex and computationally intensive. Therefore, in a more advanced embodiment, multiple points in the camera image are acquired with deliberate decimation. This avoids using every point in the camera, only an average of multiple points in the camera. The thinned out points are then used as support points on which an interpolation or regression analysis of the model surface can be performed. Using such a simplified model surface, it is possible to determine geometric properties that are at least similar to the actual reflecting surface.

[0041] The system of the present invention for optically measuring an object comprises a pattern generator, a camera unit, and a computing unit. The pattern generator is configured to generate a two-dimensional pattern in which at least one optical property varies within the pattern such that different points or groups of points are distinguishable from one another, at least within sub-regions of the pattern. The camera unit is a camera unit including a detector, where at least a part of the pattern is reflected by a reflective surface of the object onto the detector of the camera unit as a reflection pattern, and the detector is configured to convert the reflection pattern into a camera image, where a relationship between a point in the camera image and a corresponding point in the pattern can be described by a correspondence function that depends on the geometric properties of the reflective surface of the object. Dependence here means that the parameters of the correspondence function, which can be determined by measurements, depend on the geometrical properties of the reflecting surface. The calculation unit is configured to determine at least one geometric property of the reflective surface of the object, and a differential geometric property in the transformation used for the correspondence function in determining the geometric property is used.

[0042] It is not important how the camera unit and the pattern generator are positioned relative to each other, so long as the pattern generator generates a pattern that can be reflected by the reflective surface of an object to a detector in the camera unit. This requirement is usually relatively easy to meet. However, it is known that certain configurations can result in disadvantages. Therefore, in the system according to the invention, the camera unit and the pattern generator (i.e. the surface on which the pattern is displayed) are arranged such that the viewing plane used and the pattern generator are not perpendicular to each other. The viewing plane consists of the camera unit's line of sight and the light rays generated by the pattern and reflected by the object's reflective surfaces. When the viewing plane is oriented perpendicular to the pattern generator, the target function that needs to be minimized in the computations has a broad, ill-defined minimum. This is numerically unfavorable and limits the measurement accuracy in the case of noisy measurement data. Particularly for mostly flat objects, this arrangement can be systematically avoided by observing the pattern generator with a camera diagonally above the object.

[0043] The method and system according to the present invention are adapted so that only a single camera unit is used. Even with this type of monocular system, a suitably constructed pattern can determine one or more geometric properties of the object's reflective surface. When the ambiguity occurs in a periodic pattern, it can usually be easily removed. By way of example, reference is made here to the corresponding "phase unwrapping" methods known from the literature. Such ambiguity caused by periodic patterns can also be resolved using the "Frobenius criterion." This is distinct from the application to vector fields of slightly varying potential normals (as explained in detail in the papers [3] and [4] mentioned at the beginning of this specification), which is certainly applicable in this case. This is because the "candidate normals" computed from each of the periodically offset (due to ambiguities) screen points typically vary very significantly outside the "real" object surface, and therefore very clearly violate the criterion of local integrability.

[0044] The system of the present invention further includes a plurality of camera units spaced apart from one another. Each of these cameras converts the pattern reflected by the object's reflective surface into a camera image. For these individual camera images, the geometric properties of the object's reflective surfaces can be determined independently of each other. If ambiguities arise, the evaluation results of the various camera images can be combined with each other, so that the ambiguity can be reduced or eliminated.

[0045] There are various further developments of the present invention. To this end, reference is made to the dependent claims and to the following description of the drawings. Further developments of the configuration are also discussed in relation to the drawings. [Brief description of the drawings]

[0046] [Figure 1] FIG. 1 is a schematic diagram of a deflectometry system to illustrate the problem of height ambiguity in deflectometry. [Diagram 2] 1 is a first camera phase image acquired of a reflection pattern having a sinusoidal fringe profile in a first direction. [Diagram 3] 13 is a second camera phase image acquired of a reflection pattern having a sinusoidal fringe profile in a second direction perpendicular to the first direction. [Figure 4]4 is a third camera phase image of the pattern reflected by the object's reflective surface, which is the sum of the sinusoidal stripe patterns of Figures 2 and 3. [Diagram 5] 5A is a schematic diagram of a sine wave cross pattern (FIG. 5A), a camera image (FIG. 5B), and a 3x3 surrounding area of ​​a point in the camera image (FIG. 5C) to illustrate the correspondence between curves in the camera image and curves in the pattern. [Figure 6] FIG. 1 is a schematic diagram of a design example of a system according to the present invention having a double sine wave pattern. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0047] FIG. 1 shows a schematic diagram of a typical situation in which deflectometry is used. A pattern generator 1 generates a pattern, which is reflected by a reflective surface 2 of an object 3 to a camera unit 4 as a reflection pattern. A detector (not shown) of the camera unit 4 generates a camera image from the reflected pattern. The viewing axis of the camera (formed by the optical axis of the camera optics) is indicated with reference numeral 5 . The point q displayed in the camera image arises from point p in the pattern by reflection on the reflective surface 2 and can be generated by various light rays, three different possibilities are shown as examples in Figure 1. The reflected light ray 7 and the point q illuminated on the detector of the camera unit 4 can be generated by each of the light rays 6, 6', 6'' shown in FIG. Depending on whether the correct ray is ray 6, 6' or 6'', the height of the object 3 will result to be different. Since the angle of the incident ray 6, 6', 6" with respect to the surface normal and the angle of the reflected ray 7 with respect to the surface normal are always the same, different surface normals 8, 8', 8" result depending on which of the incident rays 6, 6', 6" is correct. This ambiguity needs to be resolved in order to be able to correctly determine the geometrical properties "height" and "slope" of the reflecting surface 2.

[0048] Figures 2 to 4 show various camera images 9, 9', 9" created by a slightly curved reflective surface 2 of an object 3 reflecting a pattern generated by a pattern generator 1 using an example of phase measuring deflectometry. In FIG. 2, the camera image 9 is formed by horizontally aligned curved stripes, which have an intensity curve that, with appropriate phase estimation, can be obtained from the original sinusoidal intensity curve. In Figure 3, the pattern has been rotated 90° clockwise to create a vertically aligned curved stripe pattern that has a luminance curve that can be obtained from the original sinusoidal luminance curve with appropriate phase estimation. In FIG. 4, the two patterns in FIGS. 2 and 3 are combined, forming in the camera image 9″ a combination of horizontally curved stripes and vertically curved stripes. Additionally, locations with the same phases 11 and 12 are color coded, with locations with the same phase 11 being shown horizontally in green and locations with the same phase 12 being shown vertically in red. In each of the illustrated camera images 9, 9', 9", sub-regions 10 are created in which at least one optical property of the pattern (i.e. the phase of the respective luminance curve which varies sinusoidally within the pattern) is varied, so that a cloud of points, i.e. lines with the same phase, can be inferred from the camera image. All three camera images 9, 9', 9" are used to determine the geometric properties of the reflecting surface using differential geometric properties in the transformation used for the correspondence function. Thus, a cloud of points describing the structure of the surface is determined.

[0049] FIG. 5 shows that the correspondence function can be locally approximated as a Taylor expansion in the surrounding region. The original image of the correspondence function is a coordinate line arbitrarily placed in the plane of the detector, and its image is a (usually curved) curve obtained for each observed phase value φ on the screen. FIG. 5A shows a pattern 13 that includes multiple sub-regions 10'' within a square grid. Within each sub-region 10" there is a double sinusoid having an optical characteristic, which in this case is again the brightness of a point in the pattern. Specifically, a double sine curve is two sine patterns offset by 90° from each other, which, when properly phased, produces the camera image in FIG. FIG. 5B shows diagrammatically a portion of the detector 14 of the camera unit 4 , with individual pixels 15 having centres 16 . Also shown in FIG. 5B are point p and its direct neighbors (a 3x3 surrounding area), all nine points connected to each other by horizontal and vertical lines respectively, forming a curve 17 in the camera image. FIG. 5C shows the 3×3 surrounding area again in an enlarged scale, with the variables φx(i,j) and φy(i,j) in the matrix representation written at each point. If the corresponding point in the pattern 13 is found for each point in the 3x3 surrounding area, then for example the curve 18 shown in Figure 5A is obtained.

[0050] FIG. 6 shows the mapping configuration in the system of the present invention, most of which coincides with the configuration in FIG. However, by using specially constructed patterns and calculating differential geometric properties, it is possible to eliminate the height ambiguity.

[0051] For further advantageous configurations, in order to avoid repetition, reference is made to the description section and to the appended claims.

[0052] Finally, the above design examples are illustrative of the claimed teachings and are not limited to these design examples. [Explanation of symbols]

[0053] 1. Pattern Generator 2...Reflective surface 3...object 4. Camera unit 5 ... line of sight 6...ray of light 7. Reflected light 8. Surface normal 9. Camera image 10 Sub-areas 11: Places with the same phase (horizontal direction) 12 - Places with the same phase (vertical direction) 13...Pattern 14 Detector 15 pixels 16... (pixel) center 17 Curve (on the detector / in the camera image) 18... Curve (within the pattern)

Claims

1. A method for optically measuring an object (3) having a specular and / or partially specular surface, comprising the steps of: a two-dimensional pattern (13) is generated by a pattern generator (1) in which optical properties vary within the pattern such that different points (p) or different groups of points are distinguishable from one another within the sub-regions (10); the pattern (13) is reflected by a reflective surface (2) of the object (3) onto a detector (14) of the camera unit (4) as a reflection pattern; The reflected pattern is converted by the detector (14) into a camera image (9); the relationship between a point (q) in the camera image (9) and a corresponding point (p) in the pattern (13) can be described by a correspondence function that depends on the geometric properties of the reflective surface (2) of the object (3); A geometric characteristic of the reflecting surface (2) of the object (3) is determined using differential geometric characteristics in a transformation used for the correspondence function; locally deriving said correspondence function at least once and determining at least one geometric property from said correspondence function and its differential or derivative; A method for optically measuring objects with specular and / or partially specular surfaces, characterized in that within a sub-area (10), a point or a group of points that differs from other points or groups of points in terms of at least one optical property, such as the brightness or color of the point, is made distinguishable from other points or groups of points.

2. 2. A method for optically measuring objects having specular and / or partially specular reflecting surfaces as claimed in claim 1, characterized in that the geometrical properties of the reflecting surface (2) of the object (3) include height information and / or tilt information and / or curvature information.

3. 3. A method for optically measuring objects having specular and / or partially specular reflecting surfaces as claimed in claim 1 or 2, characterized in that the area surrounding the point (q) of the camera image (9) is used for the calculation of the parameters of the correspondence function and / or the calculation of the differential geometric properties within the transformation used for the correspondence function for the point (q) of the camera image (9).

4. 4. A method for optically measuring objects having specular and / or partially specular reflecting surfaces as claimed in claim 3, characterized in that a curve (17) defined by a plurality of points in a surrounding area of ​​a point (q) in the camera image (9) is mathematically approximated using a Taylor expansion.

5. Calculating the parameters of the correspondence function and / or the differential geometric properties of a transformation used for the correspondence function for a curve (17) in the camera image (9) comprises determining a corresponding curve (18) in the pattern (13), a corresponding curve (18) in said pattern (13) is formed by a plurality of points in said pattern (13); 5. A method for optically measuring an object having a specular and / or partially specular surface as claimed in claim 1, characterized in that the plurality of points, after reflection on a reflective surface (2) of the object (3), generate the curve (17) in the camera image (9).

6. 6. A method for optically measuring an object having a specular and / or partially specular reflecting surface as claimed in any one of claims 1 to 5, characterized in that the optical properties varying within the pattern (13) consist of the brightness of a number of points within the pattern (13) and / or the color of a number of points within the pattern (13) and / or the phase of a sinusoidal brightness curve.

7. 7. A method for optically measuring an object having a specular and / or partially specular reflecting surface as claimed in claim 1, wherein a group of points having identical optical properties in the pattern (13) is formed by a plurality of points arranged adjacently on a curve.

8. the optical properties in the pattern (13) vary quasi-continuously, at least in sub-regions (10), with small discernible variations as continuous transitions or sinusoidally, 8. A method for optically measuring objects with specular and / or partially specular reflecting surfaces as claimed in claim 1, characterized in that a number of points in the pattern are preferably analysed using the local phase of a sinusoidal luminance curve.

9. The pattern (13) comprises a plurality of sub-regions (10) having internally varying optical properties within the pattern (13), 9. A method for optically measuring an object having a specular and / or partially specular surface as claimed in claim 1, characterized in that each of the sub-areas (10) is structured identically to one another.

10. The plurality of sub-regions (10) are arranged in a grid, 10. The method for optically measuring objects with specular and / or partially specular reflecting surfaces according to claim 9, characterized in that the grid is formed by square, rectangular, triangular or hexagonal grid elements.

11. 11. A method for optically measuring objects having specular and / or partially specular reflecting surfaces as claimed in claim 1, characterized in that an inverse pattern and / or a band-limited stochastic pattern and / or a colour-coded pattern and / or a stripe pattern with a brightness curve and / or a stripe pattern with a colour curve adapted to the object to be measured is used for the pattern (13).

12. A method for optically measuring an object having a specular and / or partially specular surface as claimed in any one of claims 1 to 11, characterized in that the pattern (13) is generated such that screen coordinates are analysed within the pattern (13).

13. A plurality of different patterns (13) which are mutually phase shifted are generated successively by said pattern generator (1), 13. A method for optically measuring objects having specular and / or partially specular reflecting surfaces as claimed in any one of claims 1 to 12, characterized in that the calculation of the parameters of the correspondence function and / or the differential geometric properties in the transformation used for said correspondence function are calculated iteratively for each modified pattern (13) and the different patterns (13) are selected in such a way that the calculated parameters and / or differential geometric properties are complemented and / or made more accurate with each calculation for each of the different patterns (13).

14. 14. A method for optically measuring objects with specular and / or partially specular surfaces as claimed in claim 1, characterized in that a number of points in the camera image (9) are acquired in a culled form and an interpolation or regression analysis of the model surface is carried out on the basis of the support points acquired in this way.

15. A method for optically measuring objects with specular and / or partially specular surfaces as claimed in any one of claims 1 to 14, characterized in that in the case of ambiguities in the analysis of the pattern, these ambiguities are resolved by applying a Frobenius criterion or another suitable local integrability or local smoothness criterion to the vector field, whose potential normals are ambiguous due to the periodicity of the pattern.

16. A system for optically measuring objects with specular and / or partially specular surfaces, in particular a system implementing a method for optically measuring objects with specular and / or partially specular surfaces according to any one of claims 1 to 15, comprising: a pattern generator (1) for generating a two-dimensional pattern (13) in which optical properties vary within the pattern such that different points or different groups of points are distinguishable from one another within a sub-region (10); a camera unit (4) including a detector (14), where at least a part of the pattern (13) is reflected by a reflective surface (2) of the object (3) onto the detector (14) of the camera unit (4) as a reflection pattern, the detector (14) being configured to convert the reflection pattern into a camera image (9), where a relationship between a point in the camera image (9) and a corresponding point in the pattern (13) can be described by a correspondence function that depends on geometric properties of the reflective surface (2) of the object (3); and a calculation unit configured to determine geometric properties of the reflecting surface (2) of the object (3) using differential geometric properties in a transformation used for the correspondence function.

17. the camera unit (4) and the pattern generator (1) are arranged such that a viewing plane used for evaluation and the pattern generator (1) are not perpendicular to each other; 17. The system according to claim 16, characterized in that the viewing plane is constituted by the visual axis (5) of the camera unit (4) and a light ray generated by the pattern (13) and reflected by the reflective surface (2) of the object (3).

18. A plurality of camera units (4) arranged at a distance from each other, a plurality of camera units (4), each of which converts a pattern (13) reflected by a reflective surface (2) of the object (3) into a camera image (9); 18. A system according to claim 16 or 17, characterized in that the geometrical properties of the reflective surface (2) of the object (3) are determined from the multiple camera images (9).

19. A measuring device comprising a system according to any one of claims 16 to 18 and a measuring object.

Images