Computer-implemented method for determining geometric parameters required for at least one evaluation measurement data

Abstract

The invention relates to a computer-implemented method for determining at least one geometric parameter required for evaluating measurement data, wherein the measurement data are determined by means of a transmission radiographic measurement of a component having a component geometry, wherein a digital component representation is generated from the measurement data, wherein the method (100) comprises the following steps: a step (102) of determining measurement data by means of a transmission radiographic measurement of the component; a step (104) of identifying a region in the digital component representation and / or in the component geometry as a reference region; a step (106) of determining at least one geometric parameter required for evaluating the determined measurement data by means of the reference region. With the method (100), less computing power is required than in the prior art. Furthermore, the method (100) can be used without any effort.

Description

Technical Field

[0001] The present invention relates to a computer-based method for determining at least one geometric parameter required for evaluating measurement data. Background Technology

[0002] The interior of an object or organism can be visualized using transmitted radiographs, such as computed tomography (CT) images. To perform CT measurements, geometric calibration is necessary; for each individual transmitted radiograph, this calibration arises from the relative geometrical positioning and orientation of the radiation source, detector, and object under test. In a single transmitted radiograph, the relative geometrical positioning and orientation of the radiation source, detector, and components, particularly the orientation of the X-ray source and detector with respect to the component or turntable, can be referred to as the imaging geometry. Reconstruction requires geometric calibration, during which volumetric data of the object are determined from a large number of transmitted radiographs with different imaging geometries. Errors in geometric calibration can lead to image errors in the reconstructed volumetric data, such as blurred edges, double edges, or distortion, thus generating measurement errors when performing dimensional measurements or other analyses, such as defect analysis.

[0003] For geometric calibration, the geometric parameters of the apparatus used to perform transmission radiograph measurements on the object are required. Typically, each individual transmission radiograph requires nine parameters, or degrees of freedom (DoF). When a rigid object is chosen as the reference, these are three translational parameters and three rotational parameters about the detector, and three translational parameters about the X-ray source. However, other definitions are possible. In this way, a correspondence can be established between the pixels of the transmission radiograph corresponding to a two-dimensional projection and the spatial coordinates or voxels of the three-dimensional volume data. This general case, for example, requires determining the parameters for each transmission radiograph in the case of the complex motion trajectory of a computed tomography robot. Here, local parameters are determined for each transmission radiograph.

[0004] In certain motion trajectories, such as the circular motion trajectory of axial computed tomography or in spiral computed tomography, there may be fewer parameters. In such cases, global parameters are determined, from which global parameters for each individual transmission radiograph can be determined based on the motion trajectory settings. This does not preclude the possibility of recording deviations from the ideal imaging geometry for each individual transmission radiograph using global parameters, deviations already taken into account in the local parameters.

[0005] However, especially robotic computed tomography (CT) scanners and C-arm CT scanners generally cannot precisely position the detector and radiation source. This involves not only systematic bias but also random bias, so the deviation from the ideal imaging geometry may differ in each measurement. In addition to systematic bias, normal axial CT scanners also have random bias, which can be caused by factors such as thermal effects, rotational axis wobble, slow spot drift in the radiation source or axis, tube pitting, axis vibration, deviation, or error. Furthermore, the movement of the object or organism being measured during transmitted X-ray measurement can also be a contributing factor.

[0006] US 7186023B provides information on pre- or post-transmission standards for all required imaging geometries. This can be referred to as offline calibration. The imaging geometry can be recalculated individually based on the position of the standard in each individual radiograph. However, calibration accuracy decreases over time because the apparatus used to perform the radiographic measurements, especially computed tomography robots, is not perfectly stable. Therefore, the calibration or measurement process must be repeated periodically. This is time-consuming and can also cause drift effects during measurement and immediately after calibration. Furthermore, in cases of flexible or adaptive motion trajectories, pre-determination is often not possible because the motion trajectory to be used is unknown.

[0007] It is also known to derive calibration from the measurement data itself. This can be called online calibration. Online calibration eliminates the uncertainties associated with transferring offline calibration to the measurement point in time. However, these calibrations are often not very accurate because dedicated standards with known geometries cannot be used. Furthermore, these practices are computationally expensive and complex. Summary of the Invention

[0008] Therefore, the objective of this invention is to provide a computer-implemented method that requires less computing power than existing technologies and can be used with less effort.

[0009] In a first aspect, the present invention relates to a computer-implemented method for determining at least one geometric parameter required for evaluating measurement data, wherein the measurement data is determined by means of transmission ray measurement of a component having a component geometry, wherein a digital component representation is generated from the measurement data, wherein the method includes the steps of: determining the measurement data by means of transmission ray measurement of the component; identifying a region in the digital component representation and / or in the component geometry as a reference region; and determining at least one geometric parameter required for evaluating the determined measurement data by means of the reference region.

[0010] Using this invention, online calibration is performed, in which highly accurate geometric parameters are determined. For this purpose, at least one reference region is identified in a digital representation of the component generated from the reconstruction of measurement data or transmission radiographs. The reference region includes only a portion of the component, not the entire component. The geometric parameters required to evaluate the determined measurement data, i.e., dimensional measurements, are determined solely based on said at least one reference region. To this end, a region or geometry within or above the component geometry, i.e., in the transmission radiograph or reconstructed volume data or its segments, is determined that is as suitable as possible for geometric calibration. Ideally, this calibration is calculated as in measurements on a dedicated standard, without scanning it. This reduces calibration work. Furthermore, less computational power is required compared to known online calibration methods. Suitable reference regions are searched specifically for performing calibration or determining geometric parameters. The purposeful use of reference regions reduces the required computation time (especially at each iteration) or, where appropriate, enables higher accuracy compared to known online calibration methods.

[0011] If the geometric calibration is assumed to be known, this method can, for example, be further used to determine parameters of other geometric effects and correct for those effects in this way. These effects might be deviations of the detector from its nominal shape and non-rigid behavior, i.e., deformation of the object or component being measured during the measurement process.

[0012] In one example, the parameters of two of the three geometric effects are known, and the parameters of the third geometric effect can be determined. It is also conceivable to determine the parameters of two or even all three geometric effects simultaneously.

[0013] The selected geometric shape or digital component representing, and especially the selected portion of the volume data, of the measured object having the stated geometry is called the reference region. In one example, the reference region can also be identified using prior knowledge of the component's geometry. The reference region can also be determined before measurement.

[0014] Furthermore, a reference region can be identified in the transmitted radiograph or measurement data, such as an area with high contrast and / or sharp edges. The reference region, i.e., a selected portion of the volume data representing the component geometry or digital component, is imaged using the region in the transmitted radiograph. This can be conceived, for example, to be done individually for each projection.

[0015] The evaluation of the determined measurement data may in particular include reconstruction.

[0016] The geometric parameters required for evaluation may be, for example, parameters describing the imaging geometry during measurement, the deviation of the detector from the nominal shape, or the non-rigid behavior of the component.

[0017] According to one example, the at least one geometric parameter can describe the photographic geometry used for component transmission ray measurement.

[0018] This can be referred to as the geometric calibration of the apparatus used to perform transmission radiography measurements, particularly computed tomography (CT) systems. One possibility for parameterization is to describe each transmission radiograph of the measurement data using nine parameters, or degrees of freedom.

[0019] Furthermore, the method may include, for example, the step of determining measurement data and the step of identifying regions in the digital component representation and / or component geometry as reference regions between: determining a preliminary reconstruction of the component in the form of a digital component representation from at least a portion of the determined measurement data by means of at least one predetermined geometric initial parameter, wherein, in the step of identifying regions in the digital component representation and / or component geometry as reference regions, regions in the digital component representation and / or in segments based on the digital component representation are identified.

[0020] The reference area can be identified by means of preliminary reconstruction or digital component representation or segmentation based on the preliminary reconstruction, wherein the preliminary reconstruction is based on at least a portion of the transmission radiograph or measurement data. This may, for example, mean that the determination of the measurement data has not yet been completed, but existing measurement data or a portion of existing measurement data has been used for preliminary reconstruction.

[0021] Here, the method may, after the step of determining at least one geometric parameter, further include, for example, the following steps: repeating the following steps as long as a predetermined termination criterion is not met: determining, with the aid of at least one geometric parameter, a preliminary reconstruction of another component in the form of a digital component representation from at least a portion of the determined measurement data; identifying a region in the digital component representation as a reference region; and using the reference region to determine at least one other geometric parameter that replaces the at least one geometric parameter and is required for the evaluation of the measurement data.

[0022] Because the geometric parameters are now more accurate than before, the new initial digital component representation or reconstruction has a higher quality. Therefore, at least one reference geometry can be better selected. Furthermore, a better understanding of the reference region, such as the relative positioning of the reference region or the geometry it contains, is used in new iterations. The repetition process can terminate when, for example, a predetermined number of iterations is reached, the geometric parameters no longer change significantly between iteration steps (i.e., they converge), or the measurement data or the volume data reconstructed thereby meets a predetermined quality scale.

[0023] According to another example, the step of determining at least one geometric parameter may include at least one of the following sub-steps: determining the relative positions of reference regions in a digital component representation, determining the projected positions of the reference regions in at least a portion of the determined measurement data, and using the reference region positions and the projected positions of the reference regions to determine at least one geometric parameter; and / or simulating transmission ray measurements at segments in the digital component representation and / or based on segments of the digital component representation, wherein the simulation generates simulated measurement data, and at least one geometric parameter is determined by comparing the reference regions in the determined measurement data and the simulated measurement data; and / or changing at least one geometric parameter of the digital component representation in at least one projection of the digital component representation, determining at least one mass parameter in at least one reference region reconstructed using at least one projection, repeating the changing of the at least one geometric parameter and the determination of at least one mass parameter in at least one reference region until at least one optimized geometric parameter has a mass parameter that is the maximum value of all determined mass parameters, and using at least one optimized geometric parameter as the determined at least one geometric parameter.

[0024] When determining reference regions, it can be determined, for example, which of the three sub-steps they can be used for determining geometric parameters. For example, small spherical inclusions are advantageous in principle for all three sub-steps mentioned above. Irregularly shaped inclusions are advantageous for the last two sub-steps mentioned above. Material transition regions with high contrast are advantageous for the last sub-step: changing at least one geometric parameter of the digital component representation in at least one projection of the digital component representation, determining at least one quality parameter reconstructed using at least one projection, repeating the changing of the at least one geometric parameter and the determination of the at least one quality parameter until the quality parameter possessed by at least one optimized geometric parameter is the maximum value of all determined quality parameters, and using the at least one optimized geometric parameter as the determined at least one geometric parameter.

[0025] In one example of the sub-steps of determining the relative positions of reference regions in a digital component representation, determining the projected positions of the reference regions in at least a portion of the determined measurement data, and using the reference region positions and their projected positions to determine at least one geometric parameter, the relative positions of the reference regions are determined, and thus their spacing, disregarding the global translations and rotations that collectively constitute six degrees of freedom. Therefore, the positions of the reference regions can be described in their own coordinate system. Using this information, geometric parameters can be determined such that the projected positions of the reference regions and their positions in the initial digital component representation are preferably as consistent as possible. Therefore, in the case of geometric calibration, the respective attitudes, i.e., the position and orientation of the detector and the position of the X-ray source with respect to the reference regions and, consequently, the defined coordinate system, are determined based on these projected positions. This geometric information can then be used for the accurate reconstruction of the digital component representation, which can replace the initial digital component representation.

[0026] For this example, it is advantageous to use the following component geometries as reference regions, whose projected positions can be clearly seen or located in the transmission ray images. One example is a spherical geometry. Due to the possible degrees of freedom in each individual transmission ray image, a certain minimum number of geometries is required here. It is also advantageous to use three, four, or even more reference regions to obtain accurate geometric calibration. For example, it is also advantageous to use a relatively large number of reference regions, such as at least 8, 20, 50, or 100. Thus, it is not a problem that the projected positions of some reference regions in the transmission ray images may be less accurate, sometimes overlap with each other, or they may not be visible in all transmission ray images, because this can be compensated for by improved statistics.

[0027] Furthermore, in another example, reference regions in a single projection are not considered if they overlap or cover each other in the transmitted radiograph, or are at least very close to each other, or are located in projection regions where the measurements (grayscale values ​​in computed tomography) or the measurement background (such as a grayscale background) are relatively non-uniform. This makes it difficult to determine the precise location of the reference geometry in the projection.

[0028] Typically, nonlinear equations must be solved to determine the geometric parameters. Iterative solutions to the Levenberg-Marquardt Algorithm are conceivable, for example. However, the determined solutions may be limited to local minima.

[0029] For example, for axially oriented geometry, if a free-motion trajectory is not used, the elliptical fit of the path of the reference geometry in the transmitted ray photograph may be more stable.

[0030] To perform a complete geometric calibration, the sub-step of determining at least one geometric parameter may include at least one of the following sub-steps: determining the relative position of a reference region in a digital component representation, determining the projected position of the reference region in at least a portion of the determined measurement data, and using the reference region position and the projected position of the reference region to determine at least one geometric parameter, which may be performed for all transmitted radiographs.

[0031] In the sub-step of simulating ray measurements at segments based on digital component representations and / or digital component representations (wherein the simulation generates simulated measurement data) and determining at least one geometric parameter by comparing a reference region in the determined measurement data and the simulated measurement data, for example, the set of geometric parameters can be searched for a ray image of interest for which the simulated ray image of the simulated measurement data and the measured ray image of the determined measurement data have an optimal match for the reference region. For this purpose, a similarity metric, such as the correlation of pixel grayscale values ​​in the case of computed tomography, can be used. For example, comparing a gradient representation separately is conceivable. The simulated ray images can be generated, for example, using a ray-projection-based method or forward projection. To perform complete geometric calibration, this sub-step can be performed for all ray images.

[0032] The corresponding reference region that allows for projection-based correction is also identified. Instead of the sub-step of determining at least one geometric parameter, it includes at least one of the following sub-steps: determining the relative position of the reference region in the digital component representation, determining the projected position of the reference region in at least a portion of the determined measurement data, and using the reference region position and the projected position of the reference region to determine at least one geometric parameter, but comparing the position of the reference geometry in 3D with its projected position in 2D in an "analytical" manner; the sub-step of simulating radiometric measurements is applicable to the digital component representation and / or segmentation based on the digital component representation; simulating transmission ray measurements at the segments in the digital component representation and / or based on the digital component representation (wherein the simulation generates simulated measurement data); and the sub-step of using the comparison of the reference region in the determined measurement data with that in the simulated measurement data to determine at least one geometric parameter applies to the image data. The digital transmission is compared with measured transmission ray images, and a geometric parameter leading to the maximum matching is searched. This is performed only for identified, for example, well-characterized reference regions. This significantly reduces computation time.

[0033] The process involves: altering at least one geometric parameter of the digital component representation in at least one projection of the digital component; determining at least one quality parameter in at least one reference region reconstructed using the component from at least one projection; repeating the alteration of the at least one geometric parameter and the determination of at least one quality parameter in the at least one reference region until at least one optimized geometric parameter has a quality parameter that is the maximum value of all determined quality parameters; and using the at least one optimized geometric parameter as an example of a sub-step of the determined at least one geometric parameter. A set of geometric parameters with optimal characteristics for the volume data of the reference region of the at least one transmission radiograph of interest is searched. These characteristics may be, for example, contrast, entropy, resolution, or the sharpness of edge / material transitions, as erroneous geometric parameters generally adversely affect these characteristics. This can be evaluated based on the volume data itself and / or on a grayscale histogram, respectively. Furthermore, optimization can also be performed by displaying the most distinct peaks possible for individual materials in the grayscale histogram.

[0034] In one example used for geometric calibration, the individual projections or transmission ray photographs have a negligible effect on the volume data. Therefore, the geometric parameters of all projections are advantageously optimized together, i.e., global parameters, since otherwise there would be many degrees of freedom.

[0035] To further optimize volumetric data quality, the volumetric regions being evaluated are intelligently selected. These regions serve as reference regions. In this way, the optimization can be performed more quickly, more specifically, and therefore more accurately.

[0036] All three sub-steps described above are, in principle, suitable for determining the geometric parameters of the three geometric effects described above. Geometric calibration is also the most important function in practice here. For example, multiple steps in the above sub-steps can be combined or used simultaneously. For instance, some reference regions may be used only in one sub-step, while other reference regions may be used in two or three sub-steps, depending on their applicability.

[0037] When the reference region is a simple structure that is clearly identifiable in the projection (such as a sphere or small inclusion whose position can be precisely determined), it is often more appropriate to determine the relative position of the reference region in the digital component representation, determine the projected position of the reference region in at least a portion of the determined measurement data, and use the reference region position and the projected position of the reference region to determine at least one geometric parameter.

[0038] When using a relatively complex structure as the reference region, it is often more appropriate to simulate the transmission ray measurement at the digital component representation and / or the segment based on the digital component representation (wherein the simulation generates simulated measurement data), and use the comparison of the reference region in the determined measurement data and the simulated measurement data to determine at least one geometric parameter.

[0039] In the example of geometric calibration, the first two sub-steps are often best suited for computed tomography robots or free-motion trajectories, and also best suited for determining the deviations of these projections from their respective ideal motion trajectories. In principle, all three sub-steps are applicable to determining the global parameters of conventional axial computed tomography or helical computed tomography.

[0040] Here, for example, in the sub-steps of determining the relative position of a reference region in a digital component representation, determining the projection position of the reference region in at least a portion of the determined measurement data, and using the reference region position and the projection position of the reference region to determine at least one geometric parameter, the geometry of the reference region may be taken into account when determining the projection position.

[0041] In the sub-step: determine the positions of the reference regions relative to each other in the digital component representation, determine the projected positions of the reference regions in at least a portion of the determined measurement data, and use the positions of the reference regions and the projected positions of the reference regions to determine at least one geometric parameter; taking into account the geometry of the reference regions when determining the projected positions.

[0042] Consideration of the geometry of the reference region specifically involves deviations from spherical geometry. Therefore, for example, knowledge of the three-dimensional geometry of the reference region from defect analysis can be used to predict corresponding parameters, such as the centroid of the extreme values ​​of pixel values ​​in a 2D projection of a ray-transmitted photograph or the frame edges of the reference region's image. These can then be identified better or more accurately. In this way, the accuracy of the mutual correspondence of geometrical positions in the ray-transmitted photograph or projection, or the correspondence of positions relative to the digital component representation, is improved. For example, this can be achieved by determining the principal axes of the geometry and thus determining or simulating their effects on the projection.

[0043] Furthermore, for example, in the sub-step of simulating transmitted ray measurements at segments in the digital component representation and / or based on the digital component representation (wherein the simulation generates simulated measurement data), and using a comparison of reference regions in the determined measurement data and the simulated measurement data to determine at least one geometric parameter, the simulated measurement data is generated only for the regions in the digital component representation that have reference regions.

[0044] In other words, simulated transmitted rays are calculated only in the region of the reference area and, if necessary, in a small region surrounding the reference area, and compared with a determined transmitted ray photograph from the measurement data. Therefore, the required computational power is reduced. The projection, for example, can be viewed or compared using the derivative or gradient representation after applying the Sobel operator. Furthermore, transmission measurements can only be simulated for the conical region of the component, which includes the reference area. Thus, simulated measurement data is generated only for the component region where geometric parameters need to be evaluated. This further reduces the required computational power.

[0045] In another example, in the sub-steps of changing at least one geometric parameter of the digital component representation in at least one projection of the digital component representation, determining at least one quality parameter in at least one reference region reconstructed using the component with at least one projection, repeating the change of the at least one geometric parameter and the determination of at least one quality parameter in at least one reference region until at least one optimized geometric parameter has a quality parameter that is the maximum value of all determined quality parameters, and using at least one optimized geometric parameter as the determined at least one geometric parameter, the region with the reference region in the digital component representation is reconstructed and analyzed by means of the changed geometric parameters in the digital component representation.

[0046] The correct geometric parameters can be identified by iteratively changing the geometric parameters and reconstructing and analyzing them during optimization. This is faster or allows for more iterations because only a small region needs to be reconstructed.

[0047] The step of identifying regions in the digital component representation and / or component geometry as reference regions may, for example, include the following sub-step: segmenting the boundary surfaces of the digital component representation.

[0048] In this example, the segmentation may include not only the outer interface but also the inner interface. The inner interface may be an inner surface of the component or an inner surface within the component, such as an inner surface with possibly small inclusions, pores, or defects. In one example, the segmentation may advantageously be performed at sub-voxel accuracy. The segmentation may also be based on volume data represented by a digital component, but may also be based on measurement data or transmission radiographs. For example, it is also conceivable to consider the segmentation and volume data together to identify the reference region. Here, for example, a small spherical geometry with high absorption and / or good contrast may be used in the volume data represented by the digital component.

[0049] Furthermore, for example, in the step of identifying regions in the digital component representation and / or component geometry as reference regions, regions in which material inclusions and / or air inclusions are arranged can be identified.

[0050] Material inclusions and / or air inclusions are often unknown beforehand and are identified only through appropriate analysis of the volume data from the measurement data or digital component representation. This could be, for example, an analysis to determine a defect mask. Advantageously, for example, only material inclusions can be used because they have better contrast than air inclusions. This is particularly applicable to metallic inclusions. In particular, this feature is advantageous when combined with the following sub-steps: determining the relative position of a reference region in the digital component representation, determining the projected position of the reference region in at least a portion of the determined measurement data, and using the reference region position and the projected position of the reference region to determine at least one geometric parameter.

[0051] According to another example, in the step of identifying regions in the digital component representation and / or component geometry as reference regions, regions with shapes that are spherical or classified as spherical while taking into account predetermined tolerance ranges can be identified.

[0052] This can be particularly advantageous regarding material inclusions and / or air inclusions, as inclusions have spherical geometries that can be easily located in projection or transmission radiographs. Furthermore, they are imaged very similarly in projections in different transmission directions. Analysis of the geometry of material inclusions and / or air inclusions is required to identify those inclusions with the corresponding geometries. This can be done, for example, based on a defect mask, which can be obtained, for example, from defect analysis. Numerous material inclusions may be present, especially in additive manufacturing. To determine the position of the spherical geometry in the projection as accurately as possible with subpixel precision, the centroid of the geometric image in the projection can be determined, for example. This feature is particularly advantageous when combined with the following sub-steps: determining the relative position of a reference region in the digital component representation, determining the projected position of the reference region in at least a portion of the determined measurement data, and using the reference region position and the projected position of the reference region to determine at least one geometric parameter.

[0053] In the step of identifying regions in the digital component representation and / or component geometry as reference regions, regions may also be identified, for example, as objects of a component within the component whose absorption coefficient is within a predetermined range.

[0054] These can be, for example, screws, nails, pins, rivets, metal balls, cables, or plug components. It is preferable to choose small objects within the assembly. A higher absorption coefficient for this object is advantageous. Features combined with the following sub-steps are particularly advantageous: determining the relative positions of reference regions in the digital component representation, determining the projected positions of the reference regions in at least a portion of the determined measurement data, and using the positions of the reference regions and the projected positions of the reference regions to determine at least one geometric parameter.

[0055] Furthermore, in the step of identifying regions in the digital component representation and / or component geometry as reference regions, regions arranged in the component geometry in the following volumes can be identified, where no reference regions have been identified or the number of identified reference regions is less than a predetermined limit value.

[0056] In this way, the achievable accuracy of geometric calibration can be improved because, for example, some errors in geometric calibration are manifested differently in different regions of the measurement volume containing the component. Especially when combined with sub-steps that determine the relative positions of reference regions in the digital component representation, determine the projected positions of the reference regions in at least a portion of the determined measurement data, and use the reference region positions and the projected positions of the reference regions to determine at least one geometric parameter, more stable results are obtained when the geometries are arranged as dispersed or far apart as possible.

[0057] According to the example, in the step of determining at least one geometric parameter required for the evaluation of measurement data with the help of a reference area, a safety parameter of the reference area may be considered.

[0058] A safety parameter can indicate how reliable the location of the reference region assigned to it is. For example, a reference region deemed unreliable due to its safety parameter might be given a smaller weight in a weighted adjustment. In this way, the risk of erroneous adjustment is minimized.

[0059] In all the examples above, a reference region with a surface facing all spatial directions can be selected. This ensures, for example, that grayscale transitions in all spatial directions can be present and analyzed in a transmission radiograph obtained by computed tomography. Examples of particularly advantageous geometries include: spheres with inclusions, such as hollow spheres with holes, or almost any fully enclosed geometry without edges. This is particularly advantageous in combination with the following sub-steps: changing at least one geometric parameter of the digital component representation in at least one projection of the digital component representation; determining at least one quality parameter of the component reconstruction using said at least one projection; repeating the changing of said at least one geometric parameter and the determination of said at least one quality parameter until the quality parameter possessed by at least one optimized geometric parameter is the maximum value of all determined quality parameters; and using the at least one optimized geometric parameter as the determined at least one geometric parameter.

[0060] Furthermore, when determining geometric parameters, other prior knowledge about component positioning can be considered: the positional relationships between the X-ray source, detector, and component under test. This can be considered in another example for individual projections or transmission radiographs. Thus, it can be known, for example, that the X-ray source and detector are in known "locked" or measured positions at the beginning and end of the measurement, positions known with high precision, where positioning in other projections is less reliable.

[0061] When determining geometric parameters, prior knowledge of the geometry can also be incorporated. For example, if standard geometries, such as threads or angular surfaces, are mapped in the measurement data, in some cases, it can be assumed that they are correctly manufactured without further reference measurements. For instance, knowledge about the good flatness of thread pitch or angular surfaces can be combined by identifying the geometric parameters that also satisfy this condition in the measurement data. In this way, additional geometric parameters, such as geometric magnification or voxel size, can be determined. Generally, dimensions or geometries derived from the component's CAD model can be used for this purpose. A catalog of standard geometries can be created or predefined, which is searched in the measurement data. This step is particularly advantageous when using dimensions or geometries that are typically machined with precision.

[0062] According to another example, the geometry can be selected from the central plane of the digital component representation, or the measurement data on which the geometry is based can have low unreliability or corresponding safety parameters, since the effect of cone-beam artifacts is minimized here.

[0063] Reference data for the region can come from another source, such as another measurement of the same component. In this case, imaging of the internal structure may be advantageous. Examples include high-quality or high-resolution computed tomography measurements or photomicrographs.

[0064] If a CAD model, preliminary reconstruction, or other information about the component geometry exists, then, according to another example, simulations can be calculated at a certain “density” across the entire angular space in the possible perspective directions, if necessary. The current real projection or transmitted ray photograph of the determined measurement data can be compared with the simulated projections to identify the best match. A rough photographic geometry can be determined through this matching. This example is particularly advantageous if prior knowledge of the geometric calibration is very imprecise. If necessary, in another example, if the “density” in the perspective directions is locally insufficient to determine the photographic geometry, other relevant simulated projections can also be recalculated specifically.

[0065] Another example that can be considered when implementing the method of the present invention is the use of computed tomography measurements where there is a spectrally correlated penetration depth of photons entering the detector. Components made of different materials can therefore be imaged at different effective magnifications in the case of a conical beam geometry, which can introduce measurement errors. Taking this effect into account can minimize these measurement errors.

[0066] If geometry is identified in a projection, it may be advantageous in another example to identify geometry individually for each projection.

[0067] When a component is being or will be produced using additive manufacturing, it may be advantageous, for example, to place a suitable geometry, such as a small spherical structure, inside the object so that it can be used in the method of the present invention.

[0068] In another respect, the present invention relates to a computer program product having instructions that can be run on a computer, which, when run on the computer, cause the computer to perform the method described above.

[0069] The advantages, functions, and improvement schemes of computer program products are derived from the advantages, functions, and improvement schemes of the methods described above. Therefore, in this regard, please refer to the above explanation. Attached Figure Description

[0070] Other features, details, and advantages of the invention are derived from the wording of the claims and the following description of embodiments in conjunction with the accompanying drawings, wherein:

[0071] Figure 1 A flowchart is shown of a computer-implemented method for determining the geometric parameters required for at least one measurement data evaluation. Detailed Implementation

[0072] In the following text, according to Figure 1 The computer implementation method for determining at least one geometric parameter required for the evaluation of measurement data is generally indicated by reference numeral 100 in the attached figure.

[0073] Using method 100, geometric parameters are determined to evaluate measurement data from transmitted light measurements of the component. The at least one geometric parameter may, for example, describe the imaging geometry used for the transmitted light measurements of the component.

[0074] In the first step 102, measurement data is determined using transmission radiography. Here, the measurement data can be obtained, for example, from computed tomography measurements and is a projection or transmission radiograph of the component having its geometry. The component geometry describes the true three-dimensional shape of the component. A digital representation of the component can be generated from the measurement data, for example, by reconstructing the volume data of the component from the projection.

[0075] In optional step 108, preliminary volume data can be obtained by preliminary reconstruction from at least a portion of the determined measurement data. Furthermore, a preliminary digital representation of the component is thus obtained. This preliminary reconstruction is performed using at least one predetermined initial geometric parameter. This predetermined initial geometric parameter can be predetermined by the user or by an alternative method, or it can be an estimate or empirical value determined earlier using method 100 than the final geometric parameters then used for the same or similar components.

[0076] Then, in step 104, at least one reference region in the digital component representation or component geometry is identified, for example, a region where material inclusions and / or air inclusions are arranged, and / or a region whose geometry is spherical or classified as spherical considering a predetermined tolerance range. Furthermore, for example, a region can be identified that is an object of a component within the component and has an absorption coefficient within a predetermined range, or that is arranged within the volume of a component geometry in which no reference region has been identified or the number of identified reference regions is less than a predetermined limit value. The at least one reference region here comprises only a portion of the digital component representation or component geometry.

[0077] Step 104 may have an optional sub-step 124, which is used to segment the boundary surfaces in the digital component representation.

[0078] Once step 108 has been performed, the identification of reference regions in the digital component representation obtained through preliminary reconstruction can be performed. Alternatively or additionally, the identification can be performed during the segmentation of the digital component representation.

[0079] The reference region and its location are determined based on measurements, in which existing errors or inaccuracies in the geometric parameters related to reconstruction still have an impact. Therefore, these errors are sometimes partially reflected in the reference region or its location and cannot be identified or corrected in the projection. To reconstruct the volume region, data from different regions of the detector are used to average out local biases. Therefore, the reconstruction of the volume data used to determine the reference region has a compensatory effect on the errors. Thus, the errors often have a negligible impact on the reference region or its location. Similarly, the segmentation or surface determination steps can also have a compensatory effect. Based on the information from the reference region, the errors can be determined and corrected in the next reconstruction.

[0080] When, for example, there is an unknown lateral misalignment of the detector about a few pixels, this will not cause any systematic distortion of the volume data in axial computed tomography. It will only cause blurring in the detector's xy-plane. The (centroid) position of the reference area (e.g., a defect) is unaffected, so the reference area is "correct" and can be used for correction.

[0081] Reference areas can be identified, for example, using a CAD model of the component. The CAD model is an example of prior knowledge about the component to be measured. Then, using the component geometry determined in the CAD model, suitable areas of the component are selected as reference areas to obtain geometric parameters for evaluating the measurement data. These can be, for example, areas that can be reliably and with high quality imaged in transmitted light measurements. They can be, for example, areas that are expected to be clearly visible in the measurement data due to their geometry or material composition.

[0082] Alternatively or additionally, the region can be selected from the digital component representation or its segmentation without using prior knowledge. For this purpose, for example, regions with high contrast and / or sharp edges are first selected as reference regions in a transmission radiograph. The transmission radiograph can then be evaluated individually for suitable regions. From these regions, regions imaged in the digital component representation and / or its segmentation are selected.

[0083] For example, identifying regions in a digital component representation might mean searching for reference regions particularly suitable for geometric calibration without specific prior knowledge of the component's geometry. Alternatively, this identification could, for example, mean determining the location of known or defined reference regions of the component within measurement data.

[0084] In step 106, at least one geometric parameter for evaluating the measurement data is determined using the reference region. For example, a safety parameter for the reference region may be considered. The evaluation may be, for example, a final reconstruction, i.e., generating volume data from the measurement data. Alternatively, the evaluation may be a direct geometric measurement in the projected data, where the geometric correspondence between the projections is taken into account.

[0085] Geometric parameters can describe, for example, the geometry of the image being captured, the deviation of the detector from the nominal shape, or the non-rigid behavior of a component during measurement, such as deformation.

[0086] Deviations from the nominal shape of the detector can cause detector deformation, especially in computed tomography (CT) robots. Furthermore, deviations in pixel distance, for example, may originate from detector structural tolerances.

[0087] In the following text, optional sub-steps 118, 120 and 122 of step 106 describe method steps for determining at least one geometric parameter that can be optionally combined with each other.

[0088] Therefore, in sub-step 118, the relative position of the reference region in the digital component representation can be determined. This can include not only the unreliability of determining or locating the geometry in three-dimensional space, but also the unreliability of identifying the geometry in a two-dimensional projection or determining its position and / or orientation. Furthermore, in this sub-step, the projected position of the reference region in at least a portion of the determined measurement data is determined and used to determine the at least one geometric parameter.

[0089] Optionally, the geometry of the reference region can be taken into account when determining the projection position in substep 118.

[0090] Using sub-step 120, X-ray perspective measurements can be simulated on a digital component representation and / or on segments based on the digital component representation. This simulation generates simulated measurement data, in which a reference region is also depicted. The reference region of the simulated measurement data is compared with the reference region of the determined measurement data. Regions of volume data and / or measurement data or projections that have higher noise levels are less considered in the evaluation of whether the simulated and determined measurement data are consistent, because these regions contain less information. The results of the above comparison are used to determine the at least one geometric parameter.

[0091] Optionally, in sub-step 120, simulated measurement data is generated only for the region with a reference area in the digital component representation.

[0092] Furthermore, substep 122 can be used to change at least one geometric parameter of the digital component representation in at least one projected or transmitted radiograph of the digital component representation. Component reconstruction based at least on the at least one projection is performed using at least one changed geometric parameter. For reconstruction, at least one quality parameter in at least one reference region is determined. The changing of the at least one geometric parameter and the determination of the at least one quality parameter in the at least one reference region are repeated until the quality parameter possessed by at least one optimized geometric parameter is the maximum value of all determined quality parameters. Here, a modified reconstruction is provided by changing at least one geometric parameter, and its quality is re-evaluated. Regions of volume data and / or measurement data or projections with, for example, increased noise are less considered when evaluating the quality of volume data because these regions contain less information. If an optimized geometric parameter is determined according to the above termination criteria, it is used as at least one geometric parameter for evaluating the measurement data.

[0093] Optionally, in sub-step 122, the region with the reference area of ​​the digital component representation can be reconstructed and analyzed using only the changing geometric parameters in the digital component representation.

[0094] Using reference regions, information about the possible local deviations of the detector from its nominal shape, i.e., deviations from the ideal detector geometry, can be derived. This is especially true if numerous high-quality reference regions exist and geometric calibration is known.

[0095] Real detectors are not perfect because the distance between pixels is usually not constant, and deformations perpendicular to the detector surface may occur. Therefore, there is a deviation between the detector and a perfect pixel matrix. These deviations (approximately three degrees of freedom per pixel) are approximately constant in many measurements. However, they can also vary between different projections, such as in detectors positioned by a robotic arm and deformed differently in different orientations under their own weight or due to thermal effects. These deviations can be described, for example, by a pixel-by-pixel definition of the deviation of each pixel from its nominal position. Deviations from the ideal detector geometry are generally not identifiable by projection analysis, provided the transmitted or imaged geometry is unknown. Because these reference areas and their relative positions to each other are known, some deviations that may be local can be determined by projection, and thus can be subsequently corrected or taken into account in reconstruction. There may also be a z-component, for example, when the detector is bent. This distortion may depend on temperature and therefore will differ for each projection. Nevertheless, the largest portion remains largely unchanged within the projection range. Advantageously, global distortion, for example in the form of a deviation vector, is then determined for each pixel.

[0096] To reduce the number of degrees of freedom, it may be advantageous to determine the bias vector only for a portion of the pixels and interpolate between them.

[0097] With the help of reference areas, non-rigid behavior of components during the measurement process, such as deformation, or geometric parameters can be determined for each projection, especially if there are many high-quality reference areas and geometric calibration is known.

[0098] Without correction, the measurement data are inconsistent as the measured component deforms. This is reflected in the image error in the volume data. In reconstruction, non-rigid behavior can be corrected or taken into account, thus allowing for the reconstruction of a consistent and error-free volume.

[0099] The non-rigid behavior of a component can be represented, for example, by a displacement vector field. In this case, various resolutions may be present in the field, resulting in a corresponding change in the total number of support points. Because the number of degrees of freedom is also relatively large when the detector deviates from the nominal shape, a large number of reference areas are needed to determine the geometric parameters.

[0100] Application examples include thin-walled components that deform due to their own weight during measurement, or deformation caused by thermal effects, such as heating a component in a limited area.

[0101] Advantageously, at the outset, at least one rough estimate of the geometric parameters being searched exists. However, knowing only a rough photographic geometry for a single projection is sufficient, as this is adequate for preliminary reconstruction. Alternatively, other methods can be used for the initial rough estimate of the parameters, for example, by using the consistency condition regarding line integrals in different projections or the so-called epipolar consistency condition, followed by precise determination using the present invention.

[0102] This invention is preferably used in measurements using surface detectors.

[0103] To accurately determine the location of the geometry of an identified component, the well-defined location of the geometry can be used, such as the centroid or the intersection of edges, or the corner of a cube.

[0104] Furthermore, with further methodological steps, it is possible to definitively determine the two angles at which the detector, positioned "out of plane," is tilted—that is, the two directions in which there is no detector normal as the axis of rotation. Therefore, the determination of geometric magnification or voxel size, the tilt of the detector out of plane, and / or other geometric parameters can be performed. For this purpose, for example, sensors such as laser interferometers can be used to determine the relative positions of the detector, X-ray source, and axis of rotation.

[0105] In another example, during the measurement process, another object with a known geometry, such as two spheres spaced a known distance apart, is located within the measurement range. In yet another example, when a reference measurement exists regarding the partial geometry of a component, knowledge of the geometry of certain components can be used to transfer that knowledge into the overall measurement.

[0106] Based on the example of geometric calibration using axial or spiral computed tomography (CT) scanners, global parameters can be determined first. Therefore, the number of degrees of freedom is initially limited in a reasonable manner. It is used as an initial value or starting point for the deviation of local geometric parameters or individual projections from the ideal motion trajectory in axial or spiral CT scanners.

[0107] Following step 106, the method may have an optional step 110. In step 110, optional sub-steps 112, 114, and 116 are repeated as long as a predetermined termination criterion is not met.

[0108] In sub-step 112, a preliminary reconstruction of another component, represented in digital component form, is determined. This is done using at least a portion of the determined measurement data. Here, the determination of measurement data is not necessarily isolated and can continue in parallel. Existing measurement data, or a portion of existing measurement data, can form at least a portion of the determined measurement data. Alternatively, after the determination of measurement data is complete, a portion of existing measurement data can be used for the preliminary reconstruction. This reconstruction is performed using the geometric parameters determined in step 106.

[0109] Then, in sub-step 114, the region in the digital component representation is re-identified as the reference region. Because the digital component representation is determined using geometric parameters, the digital component representation used in sub-step 114 is often more accurate than that in step 104. That is, the reference region determined in sub-step 114 also often has a higher quality than that determined in step 104.

[0110] In sub-step 116, at least one other geometric parameter is determined using the reference region from sub-step 114, replacing the previous at least one geometric parameter. Because the reference region is often better, the at least one other geometric parameter is often more accurate than the geometric parameter determined in step 106.

[0111] When substeps 112 to 116 are repeated, at least one other geometric parameter determined in substep 114 is considered as the at least one geometric parameter.

[0112] Predetermined termination criteria may include, for example, the condition that the difference between the newly determined at least one geometric parameter and the prior geometric parameter is less than a predetermined threshold, and therefore the newly determined at least one geometric parameter is not a significant improvement relative to the previous at least one geometric parameter.

[0113] The order of steps described in this example may be changed arbitrarily, provided that it can be reasonably performed.

[0114] The computer implementation method 100 can be executed on a computer using a computer program product. The computer program product here has instructions executable on a computer. When these instructions are executed on the computer, they cause the computer to perform the method.

[0115] This invention is not limited to any one of the embodiments described above, but can be modified in a variety of ways. All features and advantages derived from the claims, specification, and drawings (including structural details, spatial arrangements, and method steps) are important to this invention not only individually but also in various different combinations.

Claims

1. A computer-implemented method for determining at least one geometric parameter required for evaluating measurement data, wherein, The measurement data is determined by means of transmitted ray measurement of a component having a component geometry, wherein a digital component representation is generated from the measurement data, and wherein the method (100) includes the following steps: - Step (102) of determining the measurement data by means of transmitted rays measurement of the component; - Step (104) of identifying regions in the digital component representation and / or component geometry as reference regions. - Step (106) of determining at least one geometric parameter required to evaluate the determined measurement data using the reference area. The method (100) is performed in an online calibration manner. The step (106) of determining at least one geometric parameter has at least one of the following sub-steps: - The steps (118) of determining the relative positions of the reference regions to each other in the digital component representation, determining the projected positions of the reference regions in at least a portion of the determined measurement data, and determining at least one geometric parameter using the positions of the reference regions and the projected positions of the reference regions; and / or - Simulating transmitted ray measurements at the digital component representation and / or based on segments of the digital component representation, wherein the simulation generates simulated measurement data, and the step (120) of determining at least one geometric parameter using a comparison of a reference region in the determined measurement data with a reference region in the simulated measurement data; and / or - Change at least one geometric parameter of the digital component representation in at least one projection of the digital component representation, determine at least one quality parameter in at least one reference region reconstructed using the component of at least one projection, repeat the change of the at least one geometric parameter and the determination of at least one quality parameter in at least one reference region until at least one optimized geometric parameter has a quality parameter that is the maximum value of all determined quality parameters, and use the at least one optimized geometric parameter as the determined at least one geometric parameter (122), and In step (104), in identifying regions in the digital component representation and / or component geometry as reference regions, a sub-step is determined for each reference region that can be used to solve the geometric parameters.

2. The method according to claim 1, characterized in that, The at least one geometric parameter describes the photographic geometry used for the transmission ray measurement of the component.

3. The method according to claim 1 or 2, characterized in that, The method (100) includes the following steps between the step of determining the measurement data (102) and the step of identifying a region in the digital component representation and / or the component geometry as a reference region (104): - A step (108) to determine the initial reconstruction of the component in the form of a digital component representation from at least a portion of the determined measurement data using at least one predetermined initial geometric parameter. In step (104), in identifying regions in the digital component representation and / or component geometry as reference regions, regions in the digital component representation and / or segments based on the digital component representation are identified.

4. The method according to claim 3, characterized in that, The method (100) further includes the following steps after determining at least one geometric parameter (106): - Repeat the following steps as long as the predetermined termination criteria are not met: - The step of initially reconstructing another component in the form of a digital component representation from at least a portion of the determined measurement data using the at least one geometric parameter (112). - Step (114) of identifying a region in the digital component representation as a reference region. - The step (116) of determining at least one additional geometric parameter by means of the reference area, which replaces the at least one geometric parameter and is necessary for the evaluation of measurement data.

5. The method according to claim 1, characterized in that, The step (106) of determining at least one geometric parameter includes: determining the relative positions of the reference regions to each other in the digital component representation, determining the projected positions of the reference regions in at least a portion of the determined measurement data, and a sub-step (118) of determining at least one geometric parameter using the positions of the reference regions and the projected positions of the reference regions, wherein, in the sub-step (118), the geometry of the reference regions is taken into account when determining the projected positions.

6. The method according to claim 1, characterized in that, The step (106) of determining at least one geometric parameter includes: simulating transmitted ray measurements at segments of the digital component representation and / or based on segments of the digital component representation, wherein the simulation generates simulated measurement data, and a sub-step (120) of determining at least one geometric parameter using a comparison of a reference region in the determined measurement data with a reference region in the simulated measurement data, wherein in this sub-step (120), the simulated measurement data is generated only for regions in the digital component representation that have reference regions.

7. The method according to claim 1, characterized in that, The step (106) of determining at least one geometric parameter includes: changing at least one geometric parameter of the digital component representation in at least one projection of the digital component representation; determining at least one quality parameter in at least one reference region reconstructed using the component of at least one projection; repeating the change of the at least one geometric parameter and the determination of at least one quality parameter in at least one reference region until at least one optimized geometric parameter has a quality parameter that is the maximum value of all determined quality parameters; and using the at least one optimized geometric parameter as the determined at least one geometric parameter in a sub-step (122), wherein, in this sub-step (122), the region with the reference region in the digital component representation is reconstructed and analyzed only by means of the changed geometric parameter in the digital component representation.

8. The method according to claim 1, characterized in that, The step (104) of identifying a region in the digital component representation and / or the component geometry as a reference region has the following sub-step: the step (124) of segmenting the boundary surface of the digital component representation.

9. The method according to claim 1, characterized in that, In step (104), the region in the digital component representation and / or component geometry is identified as a reference region, and the region in which material inclusions and / or air inclusions are present is identified.

10. The method according to claim 1, characterized in that, In step (104), the region in the digital component representation and / or component geometry is identified as a reference region, and the region is identified as having a spherical geometry or is classified as spherical considering a predetermined tolerance range.

11. The method according to claim 1, characterized in that, In step (104), the region in the digital component representation and / or component geometry is identified as a reference region, and the region is identified as an object in the component of the component, the region having an absorption coefficient within a predetermined range.

12. The method according to claim 1, characterized in that, In step (104), the region in the digital component representation and / or component geometry is identified as a reference region, wherein the region is arranged in the volume of the component geometry in which the reference region has not been identified or the number of identified reference regions is less than a predetermined limit value.

13. The method according to claim 1, characterized in that, In step (106), when determining at least one geometric parameter required to evaluate the determined measurement data using the reference region, a safety parameter of the reference region is considered, wherein the safety parameter represents the reliability of the location corresponding to the reference region.

14. A computer program product having instructions executable on a computer, said instructions, when executed on said computer, causing said computer to perform the method (100) according to any one of claims 1 to 13.

Images