WO2023111426A1

WO2023111426A1 - Method for acquiring and correcting a tomographic reconstruction of an industrial part

Info

Publication number: WO2023111426A1
Application number: PCT/FR2022/052278
Authority: WO
Inventors: Clément REMACHA; Alexiane ARNAUD; Camille BILLON; Edward Romero
Original assignee: Safran
Priority date: 2021-12-16
Filing date: 2022-12-08
Publication date: 2023-06-22
Also published as: FR3131045A1; FR3131045B1

Abstract

The invention relates to a method for acquiring and correcting a tomographic reconstruction of an industrial part (4), comprising steps of identifying (34) noisy areas and of replacing (42), in a tomographic reconstruction, the noisy areas with corrected areas simulated on the basis of a theoretical model and a real digital volume made three-dimensionally.

Description

DESCRIPTION

TITLE: Process for acquiring and correcting a tomographic reconstruction of an industrial part

Technical area

The present invention relates to methods for imaging industrial parts.

In particular, the present invention relates to tomography imaging of industrial parts, and in particular to the elimination of defects in tomography imaging.

Prior techniques

In industry, and in particular during the manufacture of high pressure turbine blades, it is necessary to check the conformity of industrial parts leaving the factory.

For example, an industrial part such as a turbine blade is intended to be placed close to the combustion chamber of engines where it undergoes high thermomechanical stresses. To withstand this environment, the blades are made from a single-crystal cast alloy. This manufacturing process makes it possible to obtain complex morphologies of the blades, for example with internal cavities so as to create a passage for a cooling circuit.

In order to verify the conformity of the complex geometries with respect to the model produced by computer-aided design, the blades are measured using an X-ray tomography imaging system, non-destructive inspection imaging. The measurement is carried out by placing the industrial part on a turntable and by carrying out the acquisition of a radiographic image (two-dimensional image) every 0.1° for example. Said radiographic images are generally called projections, or else projections for tomography. One finally obtains a set of projections which are recombined using a mathematical process of tomographic reconstruction. Such a reconstruction is carried out digitally by implementing a reconstruction algorithm which makes it possible to obtain a volume in three dimensions corresponding to the tomographic image of the industrial part. This volume is usually called tomography or tomographic reconstruction.

This digital volume makes it possible to control industrial parts after manufacture, in particular the dimensions, the wall thicknesses, the sufficient presence of material, etc... In particular for turbine blades, six standard reference points are studied to verify their conformity. .

However, tomography imaging is subject to a combination trade-off of image acquisition parameters. In particular, the image acquisition parameters include the energy of the X-ray tube, the integration time per image, the magnification of the optical system. The image acquisition parameters are established according to the material composing the industrial parts as well as according to their shape and their thickness.

For example, if the x-ray tube energy and exposure time are set incorrectly, the x-ray image sensor will be underexposed or overexposed and the three-dimensional digital volume will not be usable.

For an industrial part, the adjustment of the acquisition parameters is often a compromise and the three-dimensional digital volume regularly includes underexposed zones or overexposed zones.

Additionally, inconsistencies, known as artifacts, may appear in the volume or tomography. They can come from the imaging system or from the three-dimensional reconstruction. There are several types of artifacts, including the metal artifact, or "metal artifact" in Anglo-Saxon terms. Inconsistencies can also be perceived as noise.

The metal artifact deteriorates the digital three-dimensional volume and appears as straight lines extending from areas of said digital three-dimensional volume corresponding to very dense parts of industrial parts, where X-rays have difficulty penetrating the material of industrial parts. These parasitic straight lines are the consequence of a lack of information in the projections of complex industrial parts.

Solutions for minimizing the impact of a metallic artefact are often implemented to correct medical images whose problems are different from industry.

In particular, the most used method consists in detecting the areas affected by a lack of information and completing them by interpolation with the nearby areas. However, this method requires that the nearby areas are similar to the areas affected by the lack of information. Otherwise, the three-dimensional digital volume remains largely noisy and distorted in areas lacking information.

Another method is an iterative method used by the commercial Orthopedic Metal Artifact Reduction algorithm that segments the uncorrected three-dimensional digital volume and simulates the expected digital volume. The comparison between the simulation and the digital volume makes it possible to calculate the error between the two and to reconstruct a corrected digital volume until an acceptable error is obtained. However, this method has a very high computational cost and is therefore unusable under industrial production conditions.

Finally, methods also make it possible to acquire images with different energies or integration times and to recombine them in order to capture all the details of the same object. However, this solution is time consuming and not industry compatible, as the images are often acquired with the energy already set to the maximum power possible for X-ray systems, and the integration time cannot be achieved. not be doubled.

Disclosure of Invention The object of the present invention is therefore to overcome the aforementioned drawbacks and to provide a fault correction method suitable for industrial problems.

The subject of the present invention is a method for acquiring and correcting a tomographic reconstruction of an industrial part, the industrial part being constructed from a model produced by computer-aided design, the method comprising the following steps:

- Determination of tomography acquisition parameters specific to the industrial part;

- Acquisition of a set of two-dimensional projections with the determined acquisition parameters applied to a tomographic imaging system comprising a sensor;

- Identification of the location of a subset of projections each comprising one or more noisy areas;

- Reconstruction of a digital volume in three dimensions from the set of projections;

- Registration of the model produced by computer-aided design with the three-dimensional digital volume so as to create a real model;

- Extraction of a subset of simulated projections from the real model, the subset of simulated projections being extracted from the location of the subset of projections including noisy areas; And

- Replacement in the three-dimensional digital volume of the subset of projections comprising noisy areas by the subset of simulated projections.

Thus, this method makes it possible to acquire and correct the defects of a tomographic reconstruction of an industrial part. In particular, this method allows a speed of measurement and correction advantageous to park the high rate of production. Moreover, and contrary to the medical field, an industrial part is reproducible since it is manufactured from a model produced by computer-aided design, thus allowing the registration step and the following steps.

In one mode of implementation, the acquisition parameters comprise an integration time and/or a magnification and/or an illumination power.

Advantageously, the step of acquiring a set of projections in two dimensions comprises the acquisition of at least two projections, preferably at least 1000 projections.

Advantageously, the step of identifying the location of a subset of projections comprises a determination of a noisy zone detection criterion and a digital segmentation processing making it possible to identify pixels of the projections comprising a noisy zone .

In one mode of implementation, the noisy zone detection criterion corresponds to 10% of the maximum dynamic range of the sensor.

Advantageously, the model produced by computer-aided design comprises several sub-models produced by computer-aided design.

Advantageously, the resetting step is performed from the resetting on six predetermined reference points.

Advantageously, the replacement step comprises the deletion of the noisy zones in the subset of projections comprising noisy zones, and the replacement of said deleted noisy zones by corrected zones corresponding to the same location of the subset of simulated projections.

The present invention also relates to a computer program configured to implement the method as defined previously when it is executed by the computer.

The invention also relates to a system for acquiring and correcting a tomographic reconstruction of an industrial part comprising a tomographic imaging system comprising a sensor, and a computer configured to implement the method as defined above. , or comprising the computer program as defined above. Brief description of the drawings

Other aims, characteristics and advantages of the invention will appear on reading the following description, given solely by way of non-limiting example, and made with reference to the appended drawings in which:

[Fig 1] schematically illustrates a system for acquiring and correcting a tomographic reconstruction of an industrial part according to the invention;

[Fig 2] is a flowchart schematically illustrating a method for acquiring and correcting a tomographic reconstruction of an industrial part according to the invention.

Detailed description of at least one embodiment

There is shown schematically in Figure 1 a system 2 for acquiring and correcting a tomographic reconstruction of an industrial part 4 allowing the implementation of the method as detailed below.

The acquisition and correction system 2 comprises a tomographic imaging system 6 and a computer 8 comprising means for calculating and processing tomographic images.

The tomographic imaging system 6 comprises an image sensor 10, an optical system 12 and a lighting system 14 of the industrial part 4. For example, the lighting system 14 is X-ray.

There is shown in Figure 2 an illustration in the form of a flowchart of the different steps of a process for acquiring and correcting a tomographic reconstruction of an industrial part 4.

In the mode of implementation detailed below, the aim of the method described is to create a tomographic reconstruction of an industrial part 4 by limiting the defects. In particular, the defects corrected by this method are metallic artefacts as well as missing information due to too great a part thickness or due to a loss of contrast. In particular, the industrial part can be a turbine blade whose production conformity is to be checked. This verification of conformity is carried out for example by the verification of six reference points.

Turbine blades, for example, are made in a foundry from a model 16 produced by computer-aided design. The blades, for example, incorporate one or more cores in the foundry molds so as to be able to generate a cooling circuit inside the industrial parts. The model 16 produced by computer-aided design comprises in this case several sub-models produced by computer-aided design, each sub-model corresponding for example to a core.

With reference to FIG. 2, a first step 18 of determining acquisition parameters 20 by tomography specific to the industrial part 4 is carried out. In particular, the acquisition parameters 20 can be an integration time and/or a magnification and/or illumination power. These are generally all the parameters necessary for the tomographic imaging system 6 to capture two-dimensional projections.

The acquisition parameters 20 are for example determined automatically by parameter determination means or entered manually by an operator.

In the following step 22, a set 24 of two-dimensional projections is acquired by tomography using the tomographic imaging system 6 adjusted with the acquisition parameters 20 determined previously.

The projections, also called projections for tomography, correspond to monochrome cross-sectional images of the industrial part 4. They may in particular be two-dimensional radiographic images.

From the set 24 of the projections for the tomography, a step 26 of reconstruction of a digital volume 28 in three dimensions of the industrial part 4 is carried out. the stacking of all the projections of the assembly 24 and makes it possible to reconstitute a three-dimensional image of the industrial part 4.

In order to obtain the digital volume 28 in three dimensions, it is necessary to acquire at least 2 projections during the step 22 of acquisition. However, in order to have a digital volume 28 that is as faithful as possible to the industrial part 4, it is preferable to carry out the acquisition of a large number of projections, for example at least 1000 projections.

Initially, the digital volume 28 in three dimensions makes it possible to rectify the model 16 produced by computer-aided design. Indeed, the model 16 produced by computer-aided design is a theoretical model of the industrial part 4 and does not correspond to the reality of the manufacture of the industrial part 4.

Particularly for a turbine blade including inner cores, the cores may shift during manufacture and thus not correspond to the exact theoretical location reported by the model 16 made by computer-aided design.

In order to take into account the reality of the industrial part 4, a step 30 of resetting the model 16 produced by computer-aided design with the digital volume 28 in three dimensions, which is a real image of the industrial part 4 is therefore carried out. , although with some flaws. In the case where the model 16 produced by computer-aided design comprises sub-models produced by computer-aided design, it is possible to readjust each sub-model independently of one another.

The registration step 30 thus makes it possible to create a real model 32. In a specific mode of implementation, the registration step 30 is performed from the registration of a number of predetermined reference points. These reference points, for example six in number, being essential for verifying the conformity of the industrial part 4, it is necessary that the resetting be carried out with respect to these reference points. In one mode of implementation, the registration step 30 is performed by 3D analysis software. In parallel with steps 26 and 30 of reconstruction and resetting, a step 34 of identification of noisy zones in the set 24 of projections acquired previously is carried out. In particular, it involves identifying the location of a subset 36 of tomographic projections each comprising one or more noisy zones. The subset 36 of projections can also be understood as comprising only the parts of the noisy zones in the projections, this in order to simplify the calculations.

To identify a noisy zone, digital segmentation processing is used which uses a noisy zone detection criterion. This detection criterion is determined beforehand. In one mode of implementation, the noisy zone detection criterion corresponds to 10% of the maximum dynamic range of the sensor 10. Thus, if the contrast of a zone is such that the differences in monochromatic levels are less than 10% of the dynamic of the sensor, it means that the contrast of the projection is bad and that information is missing. This is then a noisy area.

In a next step 38, from the real model 32 and the location of the subset 36 of projections comprising noisy areas, an extraction of a subset of simulated projections 40 is performed. projections, called here simulated projections, from the real model 32 to the location where the noisy areas are supposed to be on the digital volume 28 in three dimensions. The subset of simulated projections 40 is therefore a mixture of theoretical and real data making it possible to correct the noisy zones with a high degree of confidence and to replace them with corrected zones. As before, the subset of simulated projections 40 can also be understood as comprising only the parts of the zones corrected in the simulated projections, this in order to simplify the calculations.

A final step 42 of replacing the subset 36 of projections comprising noisy areas by the subset of Simulated projections 40 is performed in the digital volume 28 in three dimensions.

This step 42 is carried out for example with the deletion of the noisy zones in the subset 36 of projections comprising noisy zones, then by the replacement of said noisy zones by the corrected zones corresponding to the same location of the subset of simulated projections 40 In practice, it is a question of replacing the monochromatic levels of the pixels concerned.

Thus, one finally obtains a corrected digital volume 44 corresponding to a mixture between experimental images and simulated images but nevertheless readjusted with respect to the experimental images, so that the corrected digital volume 44 contains few imaging defects and is close to the actual geometry of the industrial part. All of the steps of the method according to the invention can be automated and implemented by a computer program executed by the computer 8.

Claims

1. Method for acquiring and correcting a tomographic reconstruction of an industrial part (4), the industrial part (4) being constructed from a model (16) produced by computer-aided design, characterized in that that it includes the following steps:

Determination (step 18) of acquisition parameters (20) by tomography specific to the industrial part (4);

Acquisition (step 22) of a set (24) of two-dimensional projections with the determined acquisition parameters (20) applied to a tomographic imaging system (6) comprising a sensor (10);

Identification (step 34) of the location of a subset (36) of projections each comprising one or more noisy areas;

Reconstruction (step 26) of a digital volume (28) in three dimensions from the set (24) of projections; Registration (step 30) of the model (16) produced by computer-aided design with the digital volume (28) in three dimensions so as to create a real model (32);

Extraction (step 38) of a subset of simulated projections (40) from the real model (32), the subset of simulated projections (40) being extracted from the location of the subset (36) projections including noisy areas; And

Replacement (step 42) in the digital volume (28) in three dimensions of the subset (36) of projections comprising noisy areas by the subset of simulated projections (40).

2. Method according to claim 1, in which the acquisition parameters (20) comprise an integration time and/or a magnification and/or an illumination power.

3. Method according to one of claims 1 and 2, wherein the step (22) of acquiring a set (24) of projections in two dimensions includes acquiring at least two projections, preferably at least 1000 projections.

4. Method according to any one of claims 1 to 3, in which the step (34) of identifying the location of a subset (36) of projections comprises a determination of a criterion for detecting noisy zone and digital segmentation processing making it possible to identify pixels of the projections comprising a noisy zone.

5. Method according to claim 4, in which the noisy zone detection criterion corresponds to 10% of the maximum dynamic range of the sensor (10).

6. Method according to any one of claims 1 to 5, in which the model (16) produced by computer-aided design comprises several sub-models produced by computer-aided design.

7. Method according to any one of claims 1 to 6, in which the step (30) of resetting is carried out from the resetting on six predetermined reference points.

8. Method according to any one of claims 1 to 7, in which the replacing step (42) comprises the deletion of the noisy zones in the subset (36) of projections comprising noisy zones, and the replacement of the said zones. noise suppressed by corrected areas corresponding to the same location of the subset of simulated projections (40).

9. Computer program comprising portions / means / program code instructions for the execution of the steps of the method according to any one of claims 1 to 8, when said program is executed by a computer.

10. System (2) for acquiring and correcting a tomographic reconstruction of an industrial part (4) comprising a tomographic imaging system (6) comprising a sensor (10), and a computer (8) configured to carrying out the method according to any one of claims 1 to 8 or comprising the computer program according to claim 9.