3D measurement with object movement

Abstract

Method for 3D measurement of objects (5) in which at least one constant and / or variable and / or several different optical patterns with a high imaging rate are projected onto the object surface to be measured three-dimensionally, and in which at least one projected pattern is captured as a change by at least one camera (2) as location-different corresponding image patterns of the object surface to be measured and evaluated to obtain 3D information of the object (5), characterized in that as many object points as possible are measured at calibrated measuring points in a calibrated spatial area.

Description

[0001] The invention relates to a method for highly accurate, non-contact 3D measurement of surfaces, in which the object is moved relative to the measurement volume, into which structured illumination patterns are projected, during the measurement process. The three-dimensional reconstruction of the object surface can then be carried out by evaluating the image sequences of the object surface recorded with at least one camera.

[0002] Improving measurement accuracy and increasing measurement point density remain key challenges for optical 3D metrology. A long-established approach involves acquiring multiple object views (multi-view methods) using one or more cameras. This method has been used in classical photogrammetry since the mid-19th century and remains relevant today, particularly for surveying larger objects (structures, technical components) [1]. However, as a passive optical measurement method, this type of photogrammetry requires a suitable surface texture, which is often insufficient for the measurement process. Therefore, markers must be applied to the object surface, which is time-consuming and often impractical.

[0003] In contrast, active optical methods for 3D object measurement project patterns of various kinds onto the object surface to be measured and record them during the measurement process with one or more cameras [2]. Using special calibration and evaluation methods (e.g., temporal and / or spatial correlation), corresponding object points can be found in the camera images, and from these, the points of the object surface in space can usually be determined by triangulation. However, these methods rely either on the use of multiple cameras or on the use of a calibrated projector and a calibrated camera. The reconstructed 3D points are then generally based on interpolations, i.e., "best-fitting solutions" of the triangulation. The absolute accuracy is limited by this interpolation and is often insufficient.

[0004] Even with active 3D measurement methods, measurement accuracy and point density can be increased by increasing the number of projected patterns. However, this requires that the projected patterns are sufficiently distinct, which quickly reaches its physical and technical limits. Therefore, similar approaches to those used in classical photogrammetry have been proposed and implemented in active methods over the last few decades.

[0005] One possibility is to use multiple cameras and / or projectors that are appropriately positioned around the object being measured [3, 9].

[0006] Another possibility is a defined rotational movement of the object in a stereo arrangement [4, 6, 8]. By using adaptive methods, even uncooperative parts of object surfaces, which are highly reflective, can be measured [7].

[0007] By rotating the object in the measurement volume of the projector-camera unit, specific areas of the object surface can be selected for more detailed investigations (regions of interest) [8].

[0008] Multiple views of the object can also be obtained if the 3D measurement system is moved around the object during the measurement [10, 11, 12]. Such measuring devices can also be used, for example, as intraoral dental digitizers with a freely movable measuring head. However, none of these methods provides a true reference for evaluating the measurement accuracy of the reconstructed 3D points in terms of ground-truth data.

[0009] Even with laser point and laser line triangulation methods, measurement accuracy and measurement point density can be increased by appropriately varying the projected light structure (laser spot or laser line) spatially during the measurement or by moving the object being measured. Ultimately, it is always only the relative movement of the object to the measurement system that matters

[13] .

[0010] In

[25] , a method is described that uses line scan cameras with structured illumination for the 3D measurement of surfaces. By using at least two line scan cameras instead of matrix cameras, it is possible to scan continuously moving surfaces in industrial continuous processes (e.g., with strip material). However, only a small spatial area can be represented per 3D reconstruction, and a true reference is also lacking, analogous to the other methods mentioned.

[0011] However, according to current technology, 3D point clouds of the same object, which represent different but partially overlapping sub-areas of the same object, can be numerically combined to form a common 3D point cloud

[26] .

[0012] In addition to improving measurement accuracy and increasing measurement point density, high-speed applications (such as crash tests) are becoming increasingly important in active optical 3D metrology

[14] . This requires not only high-speed cameras but also new methods for rapid pattern projection, for which conventional projectors like beamers are generally no longer sufficient. These include, for example, the projection of laser speckles, GOBO projectors with aperiodic fringe patterns, and multi-aperture array projectors

[15] . With these novel projectors, high-speed 3D measurement systems have been realized in recent years, which are generally operated with stereo cameras. However, these sensor systems typically have a large footprint and are therefore also very heavy. Furthermore, they are expensive, particularly due to the high-speed cameras.

[0013] Furthermore, there are prior art 3D sensors that operate with only one camera. The usual approach is to transfer the pinhole camera model to these novel projection systems. However, this is not readily possible, especially with multi-aperture array projectors. A new geometric model combined with a corresponding calibration is required. A method and a device for this purpose were disclosed in

[16] . The viability of this concept has already been demonstrated, for example, in

[17] for multi-aperture array projectors. In this way, 3D measurement point calculation analogous to the standard methods of photogrammetry and phasogrammetry is also possible for these novel projection systems with only one camera.

[0014] The calibration model, developed specifically for the Multi-Aperture Array Projector (MAAP), utilizes the measurement of projected illumination patterns (e.g., aperiodic fringes) in various planes of the measurement volume perpendicular to the projection direction. For this purpose, a calibration plane is moved in incremental steps within the measurement volume using a linear actuator. After all intrinsic and extrinsic parameters of the 3D sensor have been calibrated, 3D measurement of objects can then take place in the calibrated measurement volume using the same repeatable projection patterns used during calibration [17, 18]. The corresponding measurement points can be determined by triangulation at the positions where the calibration planes were located during calibration. Additional points between the calibrated planes in the measurement volume are determined by subpixel interpolation, as is standard practice.

[0015] In triangulation-based photogrammetric 3D measurement technology, the camera's focal length is typically fixed after sensor calibration. For a given objective lens focal length, the lateral and axial resolution of the 3D measurement then depend on the imaging detector, with the pixel pitch determining the sampling rate [19, 20]. Besides the inaccuracy resulting from interpolation during triangulation, the fixed focus range inherently limits the measurement volume. Any change in the measurement volume required to accommodate different objects, which involves a change in the camera's focal length, necessitates a complete and time-consuming recalibration of the sensor.

[0016] An alternative method for 3D measurement using triangulation is the acquisition of image stacks at different focal lengths with an electronically tunable lens (ETL) to determine the object topography. The associated algorithms / techniques are referred to as "autofocus," "focus sweep," or "shape-from-focus" [21-23]. In contrast to triangulation methods, this technique does not use a fixed focal length. The variable focal length technique is similar in some respects to tomographic techniques, in which discrete image slices can be acquired along the axial direction of the sample to reconstruct the surface topography. In particular, in X-ray-based computed tomography

[24] , which operates without an objective lens for imaging, the sample is guided through the detector system in a defined manner, and images are acquired in discrete planes / slices.

[0017] However, no true reference is used here either, and interpolation must again be used for 3D point reconstruction. Literature: [1] Luhmann, Thomas. Close-range photogrammetry. Wichmann Verlag, 2000. [2] Andres G. Marrugo et al. “State-of-the-art active optical techniques for three-dimensional surface metrology: a review” [Invited, Vol. 37, No. 9 / September 2020 / Journal of the Optical Society of America A. [3] P. Kuehmstedt, G. Notni, W. Schreiber, and J. Gerber, “Full-hemisphere automatic optical 3D measurement system,” SPIE 3100, 261-265 (1997). [4] W. Schreiber and G. Notni, “Theory and arrangements of self-calibrating whole-body 3-D measurement systems using fringe projection techniques,” Opt. Eng. 39(1), 159-169 (2000) [5] RM Kowarschik, J. Gerber, G. Notni, and W. Schreiber, “Adaptive optical three-dimensional measurement with structured light,” Opt. Eng. 39(1), 150-158 (2000). [6] US 10,704,899 B2, Takahashi S., July 7, 2020 (stereo with turntable). [7] RM Kowarschik, J. Gerber, G. Notni, and W. Schreiber, “Adaptive optical three-dimensional measurement with structured light,” Opt. Eng. 39(1), 150-158 (2000). [8] US 10,482,592 B2, Yamada T., 19.11.2019 (Object on a rotary table for more detailed examination of parts of the surface) [9] DE 10 2016 119 819 B3, Lilienblum T., 18.10.2016 (several spatially distributed light projections instead of a central projection).

[10] DE 10 2016 002 398 B4, Häusler G., 26.02.2016 (Free flight sensor).

[11] DE 10 2007 022 361.9, Munkelt C., 04.05.2007 (IOF hand scanner).

[12] DE 10 2006 049 695.7, Palme M., 16.10.2006 (IOF hand scanner for e.g. intraoral applications).

[13] US 10,837,757 B1, Ikebuchi M., 17.11.2020 (Objekt auf Drehtisch für hochgenaue Koordinatenmessung)

[14] Zhang S. „High-speed 3D shape measurement with structured light methods: A review“, Optics and Lasers in Eng. 106(2018)119-131.

[15] DE 10 2013 013 791 B4, Heist S., 14.08.2013.

[16] DE 10 2017 220 720 A1, Bräuer-Burchardt C., 20.11.2017.

[17] E. Wong, S. Heist, C. Bräuer-Burchardt, A. Stark, H. Babovsky and R. Kowarschik „Optimizationbased extrinsic calibration of a three-dimensional sensor composed of an array projector and a single camera“, Opt. Eng. 58(10), 104109 (2019).

[18] E. Wong, S. Heist, C. Bräuer-Burchardt, H. Babovsky and R. Kowarschik, „Calibration of an array projector used for high-speed three-dimensional shape measurements using a single camera“, Appl. Opt. 57, 7570-7578 (2018).

[19] Rainer G. Dorsch, Gerd Häusler, and Jürgen M. Herrmann, „Laser triangulation: fundamental uncertainty in distance measurement,“ Appl. Opt. 33, 1306-1314 (1994)

[20] Florian Willomitzer and Gerd Häusler, „Single-shot 3D motion picture camera with a dense point cloud,“ Opt. Express 25, 23451-23464 (2017)

[21] Xiaowei Hu, Guijin Wang, Jae-Sang Hyun, Yujin Zhang, Huazhong Yang, and Song Zhang, „Autofocusing method for high-resolution three-dimensional profilometry,“ Opt. Lett. 45, 375-378 (2020)

[22] Xiaowei Hu, Song Zhang, Yujin Zhang, Yongpan Liu, and Guijin Wang, „Large depth-of-field three-dimensional shape measurement with the focal sweep technique,“ Opt. Express 28, 31197-31208 (2020)

[23] Xiangjun Kong, Erwan Dupont, Al Hajjar Hani, and Frédéric Lamarque “Three-dimensional measurement using a shape from focus method applied on a context of structured light profilometry,” Proc. SPIE 11782, Optical Measurement Systems for Industrial Inspection XII, 117821B (20 June 2021)

[24] AM Cormack, 1973. Reconstruction of densities from their projections, with applications in radiological physics. Physics in Medicine & Biology, 18(2), p.195.

[25] EP 2 753 896 B1, Lilienblum T., 27.06.2013

[26] Koide, Kenji, et al. “Voxelized gicp for fast and accurate 3d point cloud registration.” 2021 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2021.

[0018] In summary, it can be stated that in previous methods for 3D measurement of objects, any object movement serves to measure as many objects as possible in a short time.

[0019] The invention is based on the objective of capturing moving objects quickly and accurately in three dimensions (with only one camera). Preferably, the number of calibrated measurement points and the measurement accuracy are to be optimized.

[0020] According to the invention, the problem is solved by a method for 3D measurement of objects using pattern projection with at least one camera, in which, to increase the number of calibrated measuring points and improve the measurement accuracy, a movement of the object in a calibrated measuring volume is used in a defined manner, so that the maximum measurement accuracy resulting from the calibration of the measuring volume can be achieved for any number of 3D points.

[0021] The method combines tomographic techniques with variable focal length technology for use in photogrammetric optical 3D measurement to achieve high depth resolution and, in triangulation-based methods using fixed focal length lenses, to capture stacks of point clouds at various sample depths. This eliminates the need for additional components such as ETLs (External Telescopes). Advantageously, a single calibration of the measurement volume at specific surfaces is sufficient. The surface topography of the object is then determined within the measurement volume at these calibrated surfaces. A camera with precisely known ("intrinsically calibrated") imaging characteristics, as is standard practice in photogrammetry, can typically be used for 3D object measurement.The surface used to calibrate the measurement volume can be analytical, such as a plane or sphere, or freeform. In many cases, calibration can also be performed with just one plane whose height deviation is small and which is positioned with high precision within the measurement volume, allowing the desired spatial area to be calibrated step by step. This is made possible, for example, with a precise (piezo) translation stage. At the individual, defined positions, which are, for example, 10 µm apart, images of the plane are taken with structured illumination using, for example, 10 band-limited patterns. These images are stored and serve as reference data sets. For the subsequent 3D measurement, the object being measured is placed in the measurement volume and measured analogously to other measurement methods, by taking one image for each illumination pattern used.By comparing the measurement images with the reference patterns, 3D points are obtained from the surface areas of the object that were located at the respective plane positions in the measurement volume during the calibration process (= calibrated measurement points). The accuracy of this point reconstruction is significantly higher than that of points determined between the planes by interpolation. However, the number of 3D points generated in this way is usually also lower and, depending on the required accuracy, can be less than 2% of all possible measurement points. After this initial 3D measurement of the object, the object is, for example, translated or rotated within the measurement volume, and 3D measurement points are determined again. For translational movement of the object, it is advantageous to use the same translation table with which the calibration of the measurement volume was initially performed.The object is moved in a smaller increment, for example 5 µm, than the plane spacing chosen during calibration. This generates additional 3D point clouds of the object, which can then be combined and / or merged into a single point cloud reconstruction of the object's surface. The displacement of the object between the point clouds in the stack then determines the depth resolution of the 3D profile. Unlike methods that process multiple camera views or interpolate using only one camera and projector, this method reconstructs only those points that correspond to the reference points on the calibrated surfaces within the measurement volume, thus achieving higher accuracy. Consequently, the achievable depth resolution is determined by the minimum detectable intensity change of a detector pixel resulting from the depth-wise translation and / or rotation.The point density can be significantly increased by repeating the position change multiple times, while the height resolution depends solely on the quality of the calibration. Alternative design options:

[0022] The process can also be implemented by rotation. For this, either the object and / or the measuring system is rotated by a known or unknown angle around a known or unknown axis of rotation. The point clouds generated after the movement are added to the other point clouds after the rotation has been compensated for.

[0023] The procedure can include translational, rotational, or translational and / or rotational movement, which can also alternate.

[0024] The method can be used to numerically simplify the summarization of point clouds with defined position changes during the measurement process, but can also be implemented by freely choosing the movement.

[0025] The process can also be implemented with more than one camera.

[0026] A multispectral implementation is also conceivable, for example by generating pattern sequences in specific spectra or by using several cameras operating in different spectral ranges in parallel.

[0027] The procedure can also be implemented on a microscopic or nanoscopic scale.

[0028] The invention will now be explained in more detail, for example with reference to the accompanying drawings, which also reveal essential features of the invention. They show: Fig. Figure 1: Schematic view of an exemplary measurement setup consisting of a projector 1 and a camera 2 in perspective front view (a) and top view (b). Also shown is the calibrated measurement volume with the calibration planes 3 and the stripe patterns 4 projected onto them. Fig. 2: Measurement setup with the measurement object 5 (a simple cuboid serves as an illustrative example) placed in the calibrated measurement volume, shown in the perspective front view (a) and in the top view (b). Fig. Figure 2c shows the 3D points (point cloud 6) determined from this first measurement in a top view. For clarity, only 3D points on the edges of the cuboid are shown. Fig. 3: In a first embodiment, the object being measured 5 (cuboid) is now linearly moved within the measuring volume (here the translational movement is perpendicular to the calibration planes towards the rear) and measured again ( Fig. 3a and b)). The 3D points determined from this measurement (point cloud 7) are again only shown at the edges of the cuboid in the top view (c). Fig. 4: The two measurements ( Fig. 2 and Fig. 3) Certain 3D points (point clouds 6 and 7) are now combined into a common 3D point reconstruction (point cloud 8). Fig. 5: In a second embodiment, the object 5 (cuboid) is not moved linearly (translatively) within the calibrated measurement volume as in the first embodiment, but rather rotated by a specific angle about an axis perpendicular to the horizontal plane of the measurement setup. The perspective front view and the top view are shown in Fig. 5a and Fig. 5b shown schematically. The 3D points determined from this measurement (point cloud 9) are again only shown at the edges of the cuboid in the top view (c). Fig. 6: The two measurements ( Fig. 2 and Fig. 5) Certain 3D points (point clouds 6 and 9) are now combined again into a common 3D point reconstruction (point cloud 10).

[0029] In Fig. Figure 1 schematically depicts an exemplary setup for the 3D measurement of objects using pattern projection, consisting of a projector 1 and a camera 2. The projector and camera are first calibrated in a known manner with respect to their internal and external parameters. Subsequently, the measurement volume is calibrated. In the Fig. In the measurement setup shown in Figure 1, this is achieved, for example, by moving a flat calibration surface to different positions within the measurement volume to be calibrated using a linear translator. Fig. For clarity, only five calibration planes are shown in the measurement volume. At each of these positions, repeatable stripe patterns (e.g., aperiodic stripes) are projected onto the calibration planes by the projector and detected by the camera synchronized with the projector. Subsequently, 3D points of the surfaces of the calibration planes at the various positions in the measurement volume can be calculated from the sequences of images using standard correlation and triangulation algorithms. This completes the calibration of the measurement volume.

[0030] Subsequently, a measurement object 5 (here, for example, a simple cuboid) is placed into the calibrated volume ( Fig. 2) and illuminated with the sequence of repeatable stripe patterns from projector 1, which was already used in the calibration of the measurement volume. Then, using known algorithms for pixel correlation and triangulation, 3D points on the surface of object 5 can be reconstructed from the images taken by camera 2 at the locations where the virtual calibration planes 3 intersect the object surface in the measurement volume. At these locations, the same (usually very good) measurement accuracy can be achieved as was present during the calibration of the planes in the measurement volume. Between the calibration planes, further 3D points on the object surface can be determined by interpolation methods, albeit with lower accuracy.To accurately reconstruct as many 3D points of the object's surface as possible using the calibrated measurement volume method, a large number of calibration planes within the measurement volume are employed, which is very time-consuming. In many cases, interpolation is still necessary to obtain as many measurement points as possible and / or to determine measurement points at locations on the object's surface that do not lie on the original calibration planes.

[0031] According to the invention, the number of reconstructed object points is increased with high measurement accuracy by performing at least two measurements of the object in the calibrated measurement volume, whereby the object is moved to a different position between the measurements. Fig. Figure 3 shows a first embodiment. After the first measurement of object 5 (cuboid), as in Fig. As described in section 2, object 5 is linearly shifted backwards within the calibrated measurement volume, e.g., by means of an actuator perpendicular to the calibration planes. The magnitude of the shift does not need to be known. However, it is advisable that object 5 remain within the calibrated measurement volume and that the magnitude of the shift also takes the object's surface structure into account. The latter is particularly important when measuring structural areas of object 5 that were not located on the calibration surfaces within the measurement volume during the first measurement. After the shift, the calibration planes 3 intersect the surface of object 5 at different points than during the first measurement. In this way, a new point cloud 7 of the object's surface is obtained, which can be merged with the point cloud 6 obtained in the first measurement into a single point cloud 8 using known algorithms, as described in [reference to be added]. Fig. Figure 4 is shown schematically. For clarity, only reconstructed 3D points on the edges of object 5 are shown. In the illustrated embodiment, the number of reconstructed 3D points on the surface of object 5 more than doubled without requiring interpolation. Furthermore, the front edge of object 5 could also be measured in this case.

[0032] In a second embodiment ( Fig. 5) will be after the first measurement ( Fig. 2) A rotation of the object 5 about an axis perpendicular to the horizontal plane of the measurement setup is performed. The calibration planes 3 now intersect the surface of the object 5 at different locations than in the first measurement. In this way, a new point cloud 9 of the object surface is obtained, which can be combined with the point cloud 6 obtained in the first measurement using known algorithms to form a common point cloud 10, as described in Fig.Figure 6 is shown schematically. For clarity, only reconstructed 3D points on the edges of object 5 are shown. In this second embodiment as well, the number of reconstructed 3D points on the surface of object 5 has more than doubled without requiring interpolation. The magnitude of the rotation of object 5 also does not need to be known. However, it is advisable that the object remains within the calibrated measurement volume and that the rotation is adapted to the surface structure of the object. Reference symbol list 1 projector 2 cameras 3 calibration levels 4 Projected stripe patterns 5. Measuring object 6 Point cloud from first measurement 7 Point cloud from the measurement after linear translation of the object 8. Combined point cloud from the first measurement and the measurement after translation of the object 9 Point cloud from the measurement after rotation of the object 10. Combined point cloud from the first measurement and the measurement after rotation of the object QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] US 10,704,899 B2

[0017] US 10,482,592 B2

[0017] DE 10 2016 119 819 B3

[0017] DE 10 2016 002 398 B4

[0017] DE 10 2007 022 361.9

[0017] DE 10 2006 049 695.7

[0017] US 10,837,757 B1

[0017] DE 10 2013 013 791 B4

[0017] DE 10 2017 220 720 A1

[0017] EP 2 753 896 B1

[0017] Zitierte Nicht-Patentliteratur

[0000] Andres G. Marrugo et al. „State-of-the-art active optical techniques for three-dimensional surface metrology: a review“ [Invited, Vol. 37, No. 9 / September 2020 / Journal of the Optical Society of America A

[0017] P. Kuehmstedt, G. Notni, W. Schreiber, and J. Gerber, „Full-hemisphere automatic optical 3D measurement system,“ SPIE 3100, 261-265 (1997

[0017] W. Schreiber and G. Notni, „Theory and arrangements of self-calibrating whole-body 3-D-measurement systems using fringe projection technique,“ Opt. Eng. 39(1), 159-169 (2000

[0017] R. M. Kowarschik, J. Gerber, G. Notni, and W. Schreiber, „Adaptive optical three-dimensional measurement with structured light,“ Opt. Eng. 39(1), 150-158 (2000

[0017] Zhang S. „High-speed 3D shape measurement with structured light methods: A review“, Optics and Lasers in Eng. 106(2018)119-131

[0017] E. Wong, S. Heist, C. Bräuer-Burchardt, A. Stark, H. Babovsky and R. Kowarschik „Optimizationbased extrinsic calibration of a three-dimensional sensor composed of an array projector and a single camera“, Opt. Eng. 58(10), 104109 (2019

[0017] E. Wong, S. Heist, C. Bräuer-Burchardt, H. Babovsky and R. Kowarschik, „Calibration of an array projector used for high-speed three-dimensional shape measurements using a single camera“, Appl. Opt. 57, 7570-7578 (2018

[0017] Rainer G. Dorsch, Gerd Häusler, and Jürgen M. Herrmann, „Laser triangulation: fundamental uncertainty in distance measurement,“ Appl. Opt. 33, 1306-1314 (1994

[0017] Florian Willomitzer and Gerd Häusler, „Single-shot 3D motion picture camera with a dense point cloud,“ Opt. Express 25, 23451-23464 (2017

[0017] Xiaowei Hu, Guijin Wang, Jae-Sang Hyun, Yujin Zhang, Huazhong Yang, and Song Zhang, „Autofocusing method for high-resolution three-dimensional profilometry,“ Opt. Lett. 45, 375-378

[0017] Xiaowei Hu, Song Zhang, Yujin Zhang, Yongpan Liu, and Guijin Wang, „Large depth-of-field three-dimensional shape measurement with the focal sweep technique,“ Opt. Express 28, 31197-31208 (2020

[0017] Xiangjun Kong, Erwan Dupont, Al Hajjar Hani, and Frédéric Lamarque „Three-dimensional measurement using a shape from focus method applied on a context of structured light profilometry“, Proc. SPIE 11782, Optical Measurement Systems for Industrial Inspection XII, 117821B (20 June 2021

[0017] A.M. Cormack, 1973. Reconstruction of densities from their projections, with applications in radiological physics. Physics in Medicine & Biology, 18(2), p.195

[0017] Koide, Kenji, et al. „Voxelized gicp for fast and accurate 3d point cloud registration.“ 2021 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2021

[0017]

Claims

Method for 3D measurement of objects (5), in which at least one constant and / or variable and / or several different optical patterns with a high imaging rate are projected onto the object surface to be measured three-dimensionally, and in which at least one projected pattern is captured in its change by at least one camera (2) as location-different corresponding image patterns of the object surface to be measured and evaluated to obtain 3D information of the object (5), characterized in that as many object points as possible are measured at calibrated measuring points in a calibrated spatial area. Method according to claim 1, characterized in that the calibrated spatial area is spanned by calibrated planes or surfaces. Method according to claim 1 or 2, characterized in that by a relative movement of the object (5) to the calibrated spatial area, as many object points as possible are measured at calibrated measuring points, i.e. at the points where the planes or surfaces of the calibrated spatial area are intersected by the surface of the object (5). Method according to one of claims 1 to 3, characterized in that the object (5) is moved by translation and / or rotation defined by the calibrated spatial area. Method according to one of claims 1 to 4, characterized in that the calibrated spatial area, preferably the surfaces spanning it, is adapted to the object (5) to be measured in such a way that the number of calibrated measuring points is optimized. Method according to one of claims 1 to 5, characterized in that more than one camera (2) and / or more than one projector (1) is used. Method according to any one of claims 1 to 6, characterized in that statistical, deterministic, coherent, incoherent, or partially coherent patterns are projected. Method according to one of claims 1 to 7, characterized in that different patterns are combined with each other.

Images