Method and apparatus for three-dimensional measurement of moving objects

Abstract

The invention relates to a method and a device for three-dimensional measurement of a moving object, the method having the following method steps: calculating a trajectory of potential object points on the basis of imaging parameters, the imaging parameters being related to a first image sequence consisting of N patterns projected onto the moving object and to a second image sequence consisting of the N patterns and to movement data of the moving object, determining a first image point trajectory in the first image sequence and a second image point trajectory in the second image sequence on the basis of the trajectory of potential object points and the imaging parameters related to the first and second image sequences, comparing image points along the first image point trajectory with image points along the second image point trajectory in order to determine a correspondence between the first and second image point trajectories, and performing a three-dimensional measurement of the moving object on the basis of the corresponding first and second image point trajectories.

Description

[0001] This application is a divisional application of the application patent application with the application date of 20 June 2017, the application number 201780045560.0 (international application number PCT / EP2017 / 065118), the invention name "Method for three-dimensional measurement of a moving object in the case of known motion". TECHNICAL FIELD

[0002] The invention relates to a method for three-dimensional measurement of a moving object. One example of this is the measurement of components moving on a flow line, in which the sensor is fixedly arranged relative to the flow line, or the measurement of large objects, in which a 3D sensor is moved continuously over the measurement object by a device, in particular a robot or a coordinate measuring machine. BACKGROUND

[0003] For the applications mentioned, 3D sensors with laser-line-triangulation are usually used. These 3D sensors are not sensitive with respect to the relative motion between object and sensor. However, only a line on the measurement object is always measured in the case of such sensors. However, many single measurements can be made by the motion of the sensor or the object, which can be combined into an area-like measurement.

[0004] It is also possible to use area-like 3D sensors with two cameras and a projector, with which a sequence of patterns is projected. Here, a method from photogrammetry is involved, with structured illumination that changes over time. Here, with one measurement a large area can be captured, so that an area-like 3D measurement can be achieved in the shortest time, in particular compared to methods based on laser-line-triangulation. However, such a method is very sensitive with respect to motion.

[0005] From this, an important prerequisite for the method is that each image point maps the same object point during the entire recording time. Neither the measurement object nor the sensor are allowed to move relative to each other during the measurement.

[0006] There are already ways for compensating for motion during the recording of a sequence of pattern images. One of these ways is described for example in Harendt, B.; Grosse, M.; Schaffer, M. & Kowarschik, R., "3D shape measurement of static and moving objects with adaptive spatiotemporal correlation", Applied Optics, 2014, 53, 7507-7515 or in Breitbarth, A.; Kuehmstedt, P.; Notni, G. & Denzler, J., "Motion compensation for three-dimensional measurements of macroscopic objects using fringe projection", DGaO Proceedings, 2012, 113.

[0007] These known ways work iteratively. Here a rough measurement is first performed which is insensitive to motion. The result of this measurement is then used to compensate for motion and to perform point assignment in the case of features over time.

[0008] The disadvantage of these ways is the iterative approach which significantly limits the accuracy of the 3D measurement. SUMMARY

[0009] From this there is the task to on the one hand constitute a 3D acquisition which is insensitive to motion and on the other hand to overcome the disadvantages of the iterative acquisition.

[0010] This task is solved by a method for three-dimensional measurement of a moving object, the method having the following method steps:

[0011] calculating a trajectory of potential object points on the basis of imaging parameters which are related to a first sequence of images consisting of N patterns projected onto the moving object and to a second sequence of images consisting of the N patterns and to motion data of the moving object,

[0012] deriving a first image point trajectory in the first sequence of images and a second image point trajectory in the second sequence of images on the basis of the trajectory of potential object points and the imaging parameters related to the first and second sequence of images,

[0013] comparing image points along the first image point trajectory with image points along the second image point trajectory to determine that a correspondence exists between the first image point trajectory and the second image point trajectory,

[0014] performing a three-dimensional measurement of the moving object based on the corresponding first image point trajectory and second image point trajectory.

[0015] According to an implementation form, comparing image points along the first image point trajectory with image points along the second image point trajectory comprises:

[0016] deriving a first series of grey values from the first image sequence and a second series of grey values from the second image sequence, and

[0017] determining a similarity of the first series of grey values and the second series of grey values.

[0018] In an implementation form, a first sensor takes the first image sequence and a second sensor takes the second image sequence, and the first sensor and the second sensor are synchronized.

[0019] In an implementation form, the second image sequence consisting of N patterns is projected onto the moving object.

[0020] Furthermore, the invention also relates to a device for performing a three-dimensional measurement of a moving object, the device comprising a processor in communication with a memory, the processor being arranged to execute instructions stored in the memory, the instructions causing the processor to perform a method comprising the steps of:

[0021] calculating a trajectory of potential object points based on imaging parameters related to a first image sequence consisting of N patterns projected onto the moving object and a second image sequence consisting of the N patterns and to motion data of the moving object,

[0022] deriving a first image point trajectory in the first image sequence and a second image point trajectory in the second image sequence based on the trajectory of potential object points and the imaging parameters related to the first image sequence and the second image sequence,

[0023] comparing image points along the first image point trajectory with image points along the second image point trajectory to determine that a correspondence exists between the first image point trajectory and the second image point trajectory,

[0024] performing a three-dimensional measurement of the moving object based on the corresponding first image point trajectory and second image point trajectory.

[0025] Furthermore, the application also relates to a non-transitory computer-readable storage medium having stored thereon instructions for three-dimensional measurement of a moving object, which, when executed by a processor, cause the processor to perform a method comprising the following steps:

[0026] calculating a trajectory of potential object points based on imaging parameters, which are related to a first image sequence consisting of N patterns projected onto the moving object and a second image sequence consisting of the N patterns and to motion data of the moving object,

[0027] deriving a first image point trajectory in the first image sequence and a second image point trajectory in the second image sequence based on the trajectory of potential object points and the imaging parameters related to the first and second image sequences,

[0028] comparing image points along the first image point trajectory with image points along the second image point trajectory to determine a correspondence between the first and second image point trajectories,

[0029] three-dimensionally measuring the moving object based on the corresponding first and second image point trajectories.

[0030] Furthermore, the task can also be solved with a method for three-dimensionally measuring a moving object during a relative movement of the moving object and at least two sensors, which has the following method steps:

[0031] calculating a trajectory of potential object points based on imaging parameters, which are related to a first image sequence consisting of N patterns projected onto the moving object and a second image sequence consisting of the N patterns projected onto the moving object and to motion data of the moving object, wherein a first sensor of the at least two sensors takes the first image sequence and a second sensor of the at least two sensors takes the second image sequence,

[0032] deriving a first image point trajectory in the first image sequence and a second image point trajectory in the second image sequence based on the trajectory of potential object points and the imaging parameters related to the first and second image sequences,

[0033] comparing image points along the first image point trajectory with image points along the second image point trajectory to determine a correspondence between the first and second image point trajectories,

[0034] three-dimensionally measuring the moving object based on the corresponding first and second image point trajectories.

[0035] The present invention also relates to a device for three-dimensional measurement of a moving object during relative movement of the moving object and at least two sensors, the device comprising a processor in communication with a memory, the processor being arranged to execute instructions stored in the memory, the instructions causing the processor to execute a method comprising the steps of:

[0036] calculating a trajectory of potential object points based on imaging parameters related to a first sequence of images consisting of N patterns projected onto the moving object and a second sequence of images consisting of the N patterns projected onto the moving object and related to movement data of the moving object, wherein a first sensor of the at least two sensors takes the first sequence of images and a second sensor of the at least two sensors takes the second sequence of images,

[0037] deriving a first image point trajectory in the first sequence of images and a second image point trajectory in the second sequence of images based on the trajectory of potential object points and the imaging parameters related to the first and second sequence of images,

[0038] comparing image points along the first image point trajectory with image points along the second image point trajectory to determine that a correspondence exists between the first and second image point trajectories,

[0039] performing a three-dimensional measurement of the moving object based on the corresponding first and second image point trajectories.

[0040] The present invention also relates to a non-transitory computer readable storage medium storing instructions for three-dimensional measurement of a moving object during relative movement of the moving object and at least two sensors, the instructions causing a processor to execute a method comprising the steps of:

[0041] calculating a trajectory of potential object points based on imaging parameters related to a first sequence of images consisting of N patterns projected onto the moving object and a second sequence of images consisting of the N patterns projected onto the moving object and related to movement data of the moving object, wherein a first sensor of the at least two sensors takes the first sequence of images and a second sensor of the at least two sensors takes the second sequence of images,

[0042] deriving a first image point trajectory in the first sequence of images and a second image point trajectory in the second sequence of images based on the trajectory of potential object points and the imaging parameters related to the first and second sequence of images,

[0043] image points along a first image point trajectory are compared with image points along a second image point trajectory in order to determine that a correspondence exists between the first image point trajectory and the second image point trajectory,

[0044] based on the corresponding first image point trajectory and the second image point trajectory, a three-dimensional measurement of the moving object is carried out.

[0045] Furthermore, this task can also be solved with a method for carrying out a three-dimensional measurement of a moving object in the case that the relative movement between the object and the sensor acting as a measuring sensor is known. The method has the following method steps:

[0046] a pattern sequence consisting of N patterns is projected onto the moving object,

[0047] a first image sequence consisting of N images is captured by a first camera and a second image sequence consisting of N images is captured by a second camera, which is synchronized to the first image sequence,

[0048] image points corresponding to one another in the first image sequence and in the second image sequence are determined, wherein

[0049] for each pair of image points of the two cameras, which should be checked for correspondence, a trajectory of a potential object point is calculated from the imaging parameters of the camera system and from the known movement data, which potential object point is mapped by the two image points, which actually correspond to one another, and from which the object point position determined at each of the N recording time points of the cameras is respectively simulated in the image plane of the first camera and in the image plane of the second camera, wherein the position of the respective image point is determined as a first image point trajectory in the first camera and as a second image point trajectory in the second camera, and the image points are compared with one another along the previously determined image point trajectories in time and space in the recorded image sequences and are checked for correspondence,

[0050] in a final step, a three-dimensional measurement of the moving object is carried out from the corresponding image points by means of triangulation.

[0051] A prerequisite for this solution is that information about the relative movement of the measuring object with respect to the sensor is known (for example in the case of a measuring object which is moved with a sensor on a measuring line or in the case of a static measuring object which is moved over it by a robot).

[0052] The method for carrying out a three-dimensional measurement of a moving object in the case of known movement data is carried out in the following method steps:

[0053] carrying out a projection of a pattern sequence consisting of N patterns onto a moving object.

[0054] Subsequently, a first image sequence consisting of N images is captured by means of a first camera and a second image sequence consisting of N images is captured by means of a second camera, which is synchronized to the first image sequence.

[0055] Then, mutually corresponding image points are determined in the first image sequence and in the second image sequence, wherein a trajectory of the potential object points is calculated from the known movement data and the determined object positions are projected onto the image planes of the first camera and the second camera, respectively, wherein the positions of the corresponding image points are determined in advance as a first image point trajectory in the first camera and as a second image point trajectory in the second camera.

[0056] The image points are compared with each other along the trajectories determined in advance and a check for correspondence is carried out. In a final step, a three-dimensional measurement of the moving object is carried out from the corresponding image points by means of triangulation.

[0057] In one embodiment, when comparing the image points along the trajectories determined in advance, a first series of grey values is determined in the first camera and a second series of grey values is determined in the second camera, and a similarity of the first series of grey values and the second series of grey values is determined.

[0058] Depending on the embodiment, a normalized cross correlation, a sum of absolute differences and / or a phase evaluation is carried out when determining the similarity of the series of grey values.

[0059] In one embodiment, a sequence of static patterns is used as the pattern sequence for the projection.

[0060] For example, a sequence of phase-shifted, sinusoidal bar patterns can be used as the pattern sequence for the projection.

[0061] It is also possible to use as the pattern sequence for the projection a static pattern whose projection onto the measurement object changes in an arbitrary manner in terms of position and / or shape. BRIEF DESCRIPTION OF DRAWINGS

[0062] The method will be explained in more detail below by means of an example. In the drawings:

[0063] Figure 1 a diagram showing the geometry of the method,

[0064] Figure 2 and 3 exemplary image point trajectories depending on time. DETAILED DESCRIPTION

[0065] The method according to the application is based on the basic idea that

[0066] Under the premise that it is known at which position (position vector) an object point is at a point in time and how this object point moves (movement vector) from there on, it is possible to predict how the corresponding image point of that object point moves in the cameras of the calibrated camera system. This information about the movement can be utilized in such a way that, when assigning the corresponding image points, not the time series of the gray values of the fixed pixels are compared with one another, but the time series of the gray values are compared along the trajectory of the image point of the object point in the image planes of the two cameras.

[0067] If there is a rigid translational movement, the movement of all object points, i.e. their movement vectors, is identical. In the case of a pipeline this movement can be determined by calibrating the movement direction of the pipeline. Nevertheless, the movement of the image points in the cameras cannot be predicted when the distance of the respective object point is unknown, because the further away an object point is from the image acquisition device, the smaller the movement of the image point in the image plane is. That is, in addition to the movement information, information about the position of the object point is also required in order to compensate for the movement of the object point.

[0068] It is in principle possible to solve this problem iteratively and to first roughly determine the position vectors of all object points using a method that is insensitive to movement and to estimate the trajectories of the object points in the images of the two sensor cameras of the sensor camera from this. However, this is very inaccurate.

[0069] The method according to the application is sufficient without such an iterative method. Rather, here the case is applied when the gray value series of two image points are compared with one another, at which time it is also implicitly checked whether the two image points map the same object point. From this it is implicitly checked whether the possible object point is at a completely determined location in space, i.e. at a location at which the viewing beams of the two image points intersect. In the case of skewed viewing beams this corresponds to the location which has the smallest distance with respect to the two viewing beams. That is, with each comparison of image points the assumption is checked whether the object point is at the respective location in space.

[0070] When now comparing the two image points it is assumed that the searched object point actually lies in a position in which the viewing beams of the image points at least approximately intersect, then it can be predicted, given the knowledge of the motion of the measured object, how the respective image points change in the two cameras when the object point moves. For this purpose, from the position vector of the object point and the motion data a trajectory of the object point is first constructed. This trajectory is now projected back into the image planes of the two cameras by means of the calibration data of the camera system. That is, at each time point of the camera it is calculated where the potential object point is at that time point and this more or less simulated 3D position is embedded, that is, projected into the image planes of the two cameras.

[0071] In order to check the similarity of the two image points, it is thus according to the application not as in the case of a static object according to Figure 2 The time-dependent grey value is compared on these image points which are fixed in the image plane. Conversely, according to Figure 1 and 3 The grey values are compared along the time-dependent image point-trajectory.

[0072] The method is of course not limited to translational motion only. If the respective motion data are known, arbitrary motion of the measured object, that is, also rotational motion, can be compensated. In theory, if the respective motion data are available, also deformations of the object can be compensated.

[0073] The method is implemented as follows.

[0074] First, information about the motion of the measured object is predetermined from the outside. In the case of a pipeline, the motion direction can be calibrated for this purpose in advance. Furthermore, the current motion speed and / or position of the pipeline can be determined by means of an encoder or another position measuring device of the pipeline.

[0075] Next, a pattern image sequence consisting of N images is projected and filmed with two synchronized cameras. That is, each camera acquires an image sequence of the respective N images of the measured object with the projected pattern. During the filming, the measured object is moved with a known motion.

[0076] The pattern image sequence consisting of N images can consist, for example, of a sinusoidal strip pattern which is shifted and / or of a random pattern, for example a static pattern with limited bandwidth. By projecting a single pattern onto the measured object and changing it there in position and / or shape in an arbitrary manner, it is also possible to generate a pattern sequence (that is, for example, a static pattern which continuously moves on a circular track above the measured object). The latter example shows that the concept "pattern image sequence" should not be understood so narrowly that a pattern image sequence always has to consist of a discrete sequence of N different pattern images which are projected onto the measured object.

[0077] In the sequence of captured images the corresponding image points, that is to say the image points mapping the same object point, are searched for in such a way that for each pixel of one camera the pixel in the other camera is searched for which has the highest similarity to the previous pixel.

[0078] The similarity of two pixels from two cameras is determined, for example, as follows, where reference is made in this regard to the diagram of Figure 1

[0079] First of all it is determined by triangulation where the respective object point 5 is located when the two pixels 9 and 16 actually map the same object point.

[0080] The trajectory 19 of the potential object point 5 is reconstructed in three dimensions from the known motion information of the measured object, in particular 17 and 18, so that at each point in time of the sequence of captured pattern images 5, 6, 7,..., 8, t = 1, 2, 3,..., N, the potential position of the moving object point is known: P(t = 1), P(t = 2), P(t = 3),..., P(t = N). This is indicated here by the reference signs 5, 6, 7,..., 8.

[0081] The now known position of the object point at each point in time of the sequence of pattern images is back-projected into the image planes 1 and 2 of the cameras with the projection centers 3 and 4, respectively. In the example shown here the pinhole camera model is applied optically, but in principle other camera models are also possible.

[0082] By this projection the trajectories 21 and 20 of the image points of the potential object point in the camera images are obtained. From this the position of the image points mapping the potential object point is known for both cameras and each point in time t. In detail this is the image points B1(t = 1), B1(t = 2), B1(t = 3),..., B1(t = N) for camera 1, indicated here by the reference signs 9, 10, 11, 12, and the image points B2(t = 1), B2(t = 2), B2(t = 3),..., B2(t = N) for camera 2, indicated here by the reference signs 16, 15, 14 and 13.

[0083] Next, the gray value series are extracted along the two image point trajectories 21, 20, more precisely the gray value series G1(t = 1), G1(t = 2), G1(t = 3),..., G1(t = N) for camera 1 and the gray value series G2(t = 1), G2(t = 2), G2(t = 3),..., G2(t = N) for camera 2. G2(t = 3), for example, denotes the gray value on the image point B2(t = 3) in the third image of the sequence of captured images in camera 2. ​

[0084] exist Figure 3 In the diagram, image point trajectories 21 and 20 are shown in a temporal and spatial manner. Thus, as the shooting time progresses, different image points are excited, and the position of this excitement moves within the corresponding camera's shooting area. For comparison, Figure 2 This illustrates the trajectory in time and space, where image points remain in constant position as the shooting time progresses, as would be the case with a static object. (Source: [Original Source Name]) Figure 2 The two trajectories are like those from Figure 3 That would provide a corresponding grayscale value series, and the two trajectories could be treated completely equivalently in terms of methodology.

[0085] Since image points are usually located at sub-pixel locations, the corresponding intensity value can also be interpolated from the intensity values ​​of adjacent pixels.

[0086] Next, the similarity of the grayscale value series is determined. Normalized cross-correlation can be used as a similarity measure. However, other similarity measures, such as sum of absolute differences or phase assessment, are also feasible. The choice of similarity measure depends on the type of projected pattern. Normalized cross-correlation is particularly useful in the case of static patterns, while phase assessment is particularly useful in the case of phase-shifted sinusoidal bar patterns.

[0087] When an image point in camera 2 has the highest similarity among all image points in camera 2 with respect to an image point in camera 1, that image point in camera 2 is assigned as the corresponding image point in camera 1. If necessary, the search area in camera 2 can be limited to a locally meaningful area, such as a so-called epipolar line.

[0088] When needed, the correspondence of points with the highest similarity can also be accurately determined at the sub-pixel level. Clearly, the method described for evaluating the similarity between two image points can also be applied to sub-pixel locations.

[0089] Then, 3D points are reconstructed from the corresponding image points using triangulation in the conventional manner.

[0090] Depending on the time point t = 2, 3, ... or N at which the measured object should be reconstructed, in the case of triangulation, the corresponding image point pairs (B1(t=1), B2(t=1)), (B1(t=2), B2(t=2)), (B1(t=3), B2(t=3)), ... or (B1(t=N), B2(t=N)) must be used, which are indicated here by the figure reference numerals (9, 16), (10, 15), (11, 14), ... or (12, 13).

[0091] The described method is not limited to translational motion only. Rotational or a combination of translation and rotation can also be compensated.

[0092] The complexity of the image point trajectories 21 and 20 depends on the complexity of the motion. In the case of arbitrary translation and rotation, the position and shape of the trajectories depend on the respective image point pairs 9, 16. In the case of translational motion of a straight line, for example in the case of a measurement object on a conveyor belt, the image point trajectories simplify to straight lines. This is the case at least in the case of an undistorted camera, which is sufficient for a pinhole camera model.

[0093] In principle, even arbitrary motion, i.e. also deformations, can be compensated if the respective motion information can be determined and is present. This can be done without problems in the case of translation and rotation. If, for example, a sensor is fixed on a robot hand and the position and orientation of the sensor relative to the flange of the robot hand is calibrated, in particular in the context of so-called hand-eye calibration, the robot hand can signal at each point in time the current position and orientation of the sensor in the coordinate system of the robot hand. From this information, the trajectory of an object point that does not move relative to the coordinate system of the robot hand in the coordinate system of the moving sensor can be determined.

[0094] The motion of a movable 3D sensor relative to a static measurement object can also be determined by means of an additional sensor, not necessarily an optical sensor, in synchronization with the acquisition of the measurement object by the 3D sensor. For example, a marker can be fixed on the primary 3D sensor, the motion of which is tracked by an additional camera, and from the motion of the marker the position change of the 3D sensor relative to the static measurement object is determined. This motion information can then be compensated in the context of the described method when reconstructing the measurement object in three dimensions.

[0095] The relative motion of the 3D sensor and the measured object can also be determined by the 3D sensor itself. For example within the scope of the described method, a static pattern can be used for the pattern projection and each synchronously taken stereo image pair (t = 1, t = 2,..., t = N) can additionally be evaluated separately with a coarse spatial correlation, although not motion-sensitive. The result is a relatively coarse, that is to say dense but imprecise, point cloud of the measured object at each taking point in time t = 1, t = 2,..., t = N. If the point clouds from the respectively successive taking points in time t = i and t = i + 1 (i = 1, 2,..., N - 1 ) are brought into agreement by means of the ICP method (iterative closest point), information about the relative motion, that is to say movement and rotation, of the measured object between t = i and t = i + 1 is obtained and, as a result, when this step is performed for i = 1 to i = N - 1, information about the entire trajectory of the measured object from t = 1 to t = N is obtained. This motion information can then be compensated in the three-dimensional reconstruction of the measured object within the scope of the described method.

[0096] Similarly to the previously described way of obtaining motion information, the motion of the measured object relative to the 3D sensor can also be determined by an additional sensor rigidly connected to the 3D sensor and calibrated relative to the 3D sensor. This additional sensor acquires the measured object synchronously with the actual 3D sensor and provides a coarse point cloud of the measured object at each taking point in time t = 1, t = 2,..., t = N. From these coarse point clouds, motion information can be obtained in the previously described way by means of the ICP method. The advantage of the additional sensor is that it can be specifically aimed at the task of providing precise motion information. Depending on which method provides the best motion information, the additional sensor can be based on a method which is not motion-sensitive, which provides either a small number of precise or a large number of imprecise measurement points of the measured object under investigation.

[0097] The method according to the invention is explained by means of embodiments. Further embodiments are possible within the scope of the technical means of the person skilled in the art.

Claims

1. A method for performing three-dimensional measurement of a moving object, the method comprising the following steps: Determine the motion data of a moving object from the outside; The trajectory of potential object points of the moving object is calculated based on the motion data of the moving object. Based on the trajectories of the potential object points, a first image sequence consisting of N patterns projected onto the moving object captured by a first sensor, and a second image sequence consisting of the N patterns captured by a second sensor, the first image point trajectory of the potential object points in the first image sequence and the second image point trajectory of the potential object points in the second image sequence are obtained. Image points along the first image point trajectory are compared with image points along the second image point trajectory to determine whether there is a correspondence between the first and second image point trajectories. Based on the corresponding first image point trajectory and second image point trajectory, the moving object is measured in three dimensions.

2. The method according to claim 1, characterized in that, Comparing image points along the trajectory of the first image point with image points along the trajectory of the second image point includes: A first grayscale value series is obtained from the first image sequence, and a second grayscale value series is obtained from the second image sequence. Determine the similarity between the first grayscale value series and the second grayscale value series.

3. The method according to claim 2, characterized in that, Determining the similarity between the first grayscale value series and the second grayscale value series includes applying normalized cross-correlation, sum of absolute differences, phase assessment, and / or similarity measures to determine the correlation between the grayscale value series.

4. The method according to claim 1, characterized in that, The N patterns projected onto the moving object include a sequence of static patterns.

5. The method according to claim 1, characterized in that, The N patterns projected onto the moving object include a sequence of phase-shifted, sinusoidal bar patterns.

6. The method according to claim 1, characterized in that, The N patterns projected onto the moving object include static patterns, and the method further includes: changing the projection of the static patterns onto the moving object in terms of position and / or shape.

7. The method according to claim 1, characterized in that, The first sensor is synchronized with the second sensor.

8. The method according to claim 1, characterized in that, The second image sequence, consisting of N patterns, is projected onto the moving object.

9. An apparatus for performing three-dimensional measurement of a moving object, the apparatus comprising a processor in communication with a memory, the processor configured to execute instructions stored in the memory, the instructions causing the processor to perform a method comprising the following steps: Determine the motion data of a moving object from the outside; The trajectory of potential object points of the moving object is calculated based on the motion data of the moving object. Based on the trajectories of the potential object points, a first image sequence consisting of N patterns projected onto the moving object captured by a first sensor, and a second image sequence consisting of the N patterns captured by a second sensor, the first image point trajectory of the potential object points in the first image sequence and the second image point trajectory of the potential object points in the second image sequence are obtained. Image points along the first image point trajectory are compared with image points along the second image point trajectory to determine whether there is a correspondence between the first and second image point trajectories. Based on the corresponding first image point trajectory and second image point trajectory, the moving object is measured in three dimensions.

10. The apparatus according to claim 9, characterized in that, Comparing image points along the trajectory of the first image point with image points along the trajectory of the second image point includes: A first grayscale value series is obtained from the first image sequence, and a second grayscale value series is obtained from the second image sequence. Determine the similarity between the first grayscale value series and the second grayscale value series.

11. The apparatus according to claim 10, characterized in that, Determining the similarity between the first grayscale value series and the second grayscale value series includes applying normalized cross-correlation, sum of absolute differences, phase assessment, and / or similarity measures to determine the correlation between the grayscale value series.

12. The apparatus according to claim 9, characterized in that, The N patterns projected onto the moving object include a sequence of static patterns and / or a sequence of phase-shifted, sinusoidal bar patterns.

13. The apparatus according to claim 9, characterized in that, The N patterns projected onto the moving object include static patterns, and the instructions also cause the processor to change the projection of the static patterns onto the moving object in terms of position and / or shape.

14. The apparatus according to claim 9, characterized in that, The first sensor is synchronized with the second sensor.

15. The apparatus according to claim 9, characterized in that, The second image sequence, consisting of N patterns, is projected onto the moving object.

16. A non-transitory computer-readable storage medium storing instructions for performing three-dimensional measurement of a moving object, wherein when the instructions are executed by a processor, the processor performs a method comprising the following steps: Determine the motion data of a moving object from the outside; The trajectory of potential object points of the moving object is calculated based on the motion data of the moving object. Based on the trajectories of the potential object points, a first image sequence consisting of N patterns projected onto the moving object captured by a first sensor, and a second image sequence consisting of the N patterns captured by a second sensor, the first image point trajectory of the potential object points in the first image sequence and the second image point trajectory of the potential object points in the second image sequence are obtained. Image points along the first image point trajectory are compared with image points along the second image point trajectory to determine whether there is a correspondence between the first and second image point trajectories. Based on the corresponding first image point trajectory and second image point trajectory, the moving object is measured in three dimensions.

17. A method for performing three-dimensional measurement of a moving object during relative motion with at least two sensors, the method comprising the following steps: Determine the motion data of a moving object from the outside; The trajectory of potential object points of the moving object is calculated based on the motion data of the moving object, wherein, The first sensor of the at least two sensors captures a first image sequence consisting of N patterns projected onto the moving object, and the second sensor of the at least two sensors captures a second image sequence consisting of the N patterns. Based on the trajectories of the potential object points and the first and second image sequences, the first image point trajectory of the potential object points in the first image sequence and the second image point trajectory of the potential object points in the second image sequence are obtained. Image points along the first image point trajectory are compared with image points along the second image point trajectory to determine whether there is a correspondence between the first and second image point trajectories. Based on the corresponding first image point trajectory and second image point trajectory, the moving object is measured in three dimensions.

18. The method according to claim 17, characterized in that, Comparing image points along the trajectory of the first image point with image points along the trajectory of the second image point includes: A first grayscale value series is obtained from the first image sequence, and a second grayscale value series is obtained from the second image sequence. Determine the similarity between the first grayscale value series and the second grayscale value series.

19. The method according to claim 18, characterized in that, Determining the similarity between the first grayscale value series and the second grayscale value series includes applying normalized cross-correlation, sum of absolute differences, phase assessment, and / or similarity measures to determine the correlation between the grayscale value series.

20. The method according to claim 17, characterized in that, The N patterns projected onto the moving object include a sequence of static patterns.

21. The method according to claim 17, characterized in that, The N patterns projected onto the moving object include a sequence of phase-shifted, sinusoidal bar patterns.

22. The method according to claim 17, characterized in that, The N patterns projected onto the moving object include static patterns, and the method further includes: changing the projection of the static patterns onto the moving object in terms of position and / or shape.

23. The method according to claim 17, characterized in that, The first sensor is synchronized with the second sensor.

24. An apparatus for performing three-dimensional measurements of a moving object during relative motion with at least two sensors, the apparatus comprising a processor in communication with a memory, the processor configured to execute instructions stored in the memory, the instructions causing the processor to perform a method comprising the following steps: Determine the motion data of a moving object from the outside; The trajectory of potential object points of the moving object is calculated based on the motion data of the moving object, wherein, The first sensor of the at least two sensors captures a first image sequence consisting of N patterns projected onto the moving object, and the second sensor of the at least two sensors captures a second image sequence consisting of the N patterns. Based on the trajectories of the potential object points and the first and second image sequences, the first image point trajectory of the potential object points in the first image sequence and the second image point trajectory of the potential object points in the second image sequence are obtained. Image points along the first image point trajectory are compared with image points along the second image point trajectory to determine whether there is a correspondence between the first and second image point trajectories. Based on the corresponding first image point trajectory and second image point trajectory, the moving object is measured in three dimensions.

25. The apparatus according to claim 24, characterized in that, Comparing image points along the trajectory of the first image point with image points along the trajectory of the second image point includes: A first grayscale value series is obtained from the first image sequence, and a second grayscale value series is obtained from the second image sequence. Determine the similarity between the first grayscale value series and the second grayscale value series.

26. The apparatus according to claim 25, characterized in that, Determining the similarity between the first grayscale value series and the second grayscale value series includes applying normalized cross-correlation, sum of absolute differences, phase assessment, and / or similarity measures to determine the correlation between the grayscale value series.

27. The apparatus according to claim 24, characterized in that, The N patterns projected onto the moving object include a sequence of static patterns.

28. The apparatus according to claim 24, characterized in that, The N patterns projected onto the moving object include a sequence of phase-shifted, sinusoidal bar patterns.

29. The apparatus according to claim 24, characterized in that, The N patterns projected onto the moving object include static patterns, and the instructions also cause the processor to change the projection of the static patterns onto the moving object in terms of position and / or shape.

30. The apparatus according to claim 24, characterized in that, The first sensor is synchronized with the second sensor.

31. A non-transitory computer-readable storage medium storing instructions for performing three-dimensional measurements of a moving object during relative motion with at least two sensors, wherein when the instructions are executed by a processor, the processor performs a method comprising the following steps: Determine the motion data of a moving object from the outside; The trajectory of potential object points of the moving object is calculated based on the motion data of the moving object, wherein, The first sensor of the at least two sensors captures a first image sequence consisting of N patterns projected onto the moving object, and the second sensor of the at least two sensors captures a second image sequence consisting of the N patterns. Based on the trajectories of the potential object points and the first and second image sequences, the first image point trajectory of the potential object points in the first image sequence and the second image point trajectory of the potential object points in the second image sequence are obtained. Image points along the first image point trajectory are compared with image points along the second image point trajectory to determine whether there is a correspondence between the first and second image point trajectories. Based on the corresponding first image point trajectory and second image point trajectory, the moving object is measured in three dimensions.

32. The non-transitory computer-readable storage medium according to claim 31, characterized in that, Comparing image points along the trajectory of the first image point with image points along the trajectory of the second image point includes: A first grayscale value series is obtained from the first image sequence, and a second grayscale value series is obtained from the second image sequence. Determine the similarity between the first grayscale value series and the second grayscale value series.

33. The non-transitory computer-readable storage medium according to claim 32, characterized in that, Determining the similarity between the first grayscale value series and the second grayscale value series includes applying normalized cross-correlation, sum of absolute differences, phase assessment, and / or similarity measures to determine the correlation between the grayscale value series.

34. The non-transitory computer-readable storage medium according to claim 31, characterized in that, The N patterns projected onto the moving object include a sequence of static patterns.

35. The non-transitory computer-readable storage medium according to claim 31, characterized in that, The N patterns projected onto the moving object include a sequence of phase-shifted, sinusoidal bar patterns.

36. The non-transitory computer-readable storage medium according to claim 31, characterized in that, The N patterns projected onto the moving object include static patterns, and the instructions also cause the processor to change the projection of the static patterns onto the moving object in terms of position and / or shape.

Images