METHOD FOR DETERMINING THE DISPLACEMENT OF A TRANSLATOR BY ULTRASONIC IMAGING

The method determines the position change of ultrasonic translators using ultrasonic data alone, eliminating the need for additional instrumentation, thereby simplifying and reducing costs in ultrasonic imaging while maintaining image reconstruction accuracy.

FR3124014B1Active Publication Date: 2024-11-29COMMISSARIAT A LENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVES
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
FR2021006098
Authority / Receiving Office
FR · FR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2021-06-09
Publication Date
2024-11-29
Estimated Expiration
2041-06-09

AI Technical Summary

Technical Problem

Existing non-destructive testing methods for ultrasonic imaging require additional and expensive instrumentation, such as mechanical encoders or guides, to determine the mechanical movement of ultrasonic translators, which increases complexity and cost.

Method used

A method to determine the position change of an ultrasonic translator using ultrasonic data alone by applying image matching and transformation techniques to elementary images acquired at different positions, without the need for additional instrumentation.

Benefits of technology

Enables the reconstruction of an overall object image without additional instrumentation, simplifying and reducing costs while maintaining accuracy in ultrasonic imaging.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000014_0000
    Figure 00000014_0000
  • Figure 00000014_0001
    Figure 00000014_0001
  • Figure 00000015_0000
    Figure 00000015_0000
Patent Text Reader

Abstract

The present invention relates to a method for determining a change in position of an ultrasonic transducer. The method comprises the following steps: - obtaining, with the transducer, a first and a second elementary image of an object having a surface, the first elementary image being obtained with the transducer placed in a first position, the second elementary image being obtained with the transducer placed in a second position, different from the first position, such that the same part of the object is present in a common area of ​​the first and second elementary images; - highlighting the surface of the object on the elementary images; - determining the change in position by applying a mapping to the elementary images so that the surface of the object is covered in the common area of ​​the elementary images. The present invention also relates to a method for obtaining an overall image of the object.Figure for the abbreviation: figure 1.
Need to check novelty before this filing date? Find Prior Art

Description

Title of the invention: METHOD FOR DETERMINING THE DISPLACEMENT OF A TRANSLATOR BY ULTRASONIC IMAGING Technical field and state of the prior art

[0001] The present invention lies in the field of non-destructive testing by ultrasound, and in particular that of adaptive imaging allowing the inspection of components by surface learning based on the analysis of ultrasonic data. It relates to a method for acquiring an image of an object with an ultrasonic translator. The present invention also relates to a method for reconstructing the image of the object from a plurality of elementary ultrasonic images reconstructed at each position of the translator.

[0002] In order to be able to reconstruct the image of an object from several images acquired with a translator, it is necessary to know the mechanical displacement of the translator between its successive positions. The position of the translator can be measured, for example, with a mechanical encoder. Alternatively, it is possible to limit the freedom of movement of the translator by constraining its degrees of freedom of movement. For example, a guide or a robotic arm can maintain the orientation and the distance to the object of the translator during a movement to scan the object to be measured.

[0003] In both cases additional and sometimes expensive instrumentation, such as a mechanical encoder or a mechanical guide, is necessary.

[0004] To propose a simpler, less constrained and less expensive technique, the present invention proposes to determine the change in the position of the translator without resorting to additional instrumentation, based only on the acquired images themselves. Statement of the invention

[0005] It is therefore an object of the present invention to provide a method for determining a change in position of an ultrasonic transducer. The method comprises the following steps:

[0006] - obtain with the translator a first and a second elementary image of a object having a surface, the first elementary image being obtained with the translator placed at a first position, the second elementary image being obtained with the translator placed at a second position, different from the first position, so that the same part of the object is present in a common part of the first and second elementary images,

[0007] -highlight the surface of the object on the elementary images

[0008] - determine the change in position by applying a mapping to the elementary images so as to cover the surface of the object in the common part of the elementary images.

[0009] Preferably, the elementary images are obtained with a coupling medium arranged between the translator and the surface of the object, and the step of highlighting the surface of the object comprises the reconstruction of the elementary images in the coupling medium or in the volume of the object, preferably by a total focusing method.

[0010] Highlighting the surface of the object may include the steps of:

[0011] - defining a threshold for each column of each elementary image,

[0012] - highlight, as points of the surface, the pixels of each column having a value above said threshold,

[0013] - preferably apply a one-dimensional smoothing filter to each image element mental.

[0014] It is also possible to highlight the surface in the common part in each elementary image before applying the matching.

[0015] The matching of the elementary images may comprise the step of determining a transformation matrix (T) between the first and the second image, preferably by a closest point iteration (ICP) method.

[0016] A method for obtaining an overall image of an object may comprise the steps of:

[0017] - implement the method described previously to obtain a first and second elementary image at a first and second position and a change of position of the translator between the first and second position,

[0018] - use the determined position change to display the first image and the second elementary image in a common reference frame in order to obtain the overall image of the object.

[0019] The object may be immersed in a coupling medium, preferably water, or the translator may comprise a flexible membrane shoe filled with a coupling medium, preferably water, the flexible membrane shoe being positioned between the object and the translator.

[0020] It is possible to move the translator using a robotic arm between the first and second positions.

[0021] The translator can perform a scan of the object and a plurality of elementary ultrasonic images can be obtained at a plurality of positions of the translator. Then, the steps of determining the change of position and display in the common reference frame can be applied several times to pairs of images. Brief description of the drawings

[0022] The present invention will be better understood on the basis of the description which follows and the appended drawings in which:

[0023] [Fig-1] shows an object comprising an object volume and a surface,

[0024] [Fig.2] shows the ultrasonic translator at a first position and at a second position, different from the first position,

[0025] [Fig.3] shows a first image acquired at a first position and a second image, acquired at a second position,

[0026] [Fig.4a] shows the first and second images with the surface of the object,

[0027] [Fig.4b] shows an edge point rejection operation,

[0028] [Fig.5] shows the first and second images with their reference system,

[0029] [Fig.6] shows a preferred embodiment of the method of acquiring data with an ultrasonic translator and of reconstructing the unit images obtained at each position of the translator on the object, then of reconstructing the global image of the object from these unit images.

[0030] Detailed description of particular embodiments

[0031] [Fig.l] shows an object (10) comprising an object volume (180) and a surface (20). The figure also shows an ultrasonic transducer (50) comprising a flexible membrane shoe. The shoe is filled with a coupling medium (100), preferably with water, to provide coupling between a part to be examined and the transducer. The transducer is coupled to a robotic arm (130) which can move the transducer in a scanning direction (140). Said robotic arm may comprise a guide in which the transducer is guided to be moved in the scanning direction.

[0032] The presence of the coupling medium improves the passage of ultrasonic waves between the translator and the object. The flexible membrane shoe conforms to the surface of the object during movement of the translator so as to place the coupling medium between the translator and the surface of the object.

[0033] Alternatively, the translator may not be equipped with the flexible shoe, the object then being able to be immersed in the coupling medium, for example water. In this case, the active surface of the translator is also immersed and it is held at a predetermined distance from the surface of the object.

[0034] During the movement of the translator to perform a scan (140) of the object, the translator is moved in a common reference frame (170) while the object remains fixed in this same reference system. As illustrated in [Fig.2], the translator has its own reference system (190) which remains fixed relative to the translator and moves with it in the common reference frame (170). As illustrated in [Fig.2], the translator obtains an elementary image (30, 40) in its own reference system (190) which changes position with the translator. To obtain an overall image of the object, at least two elementary images are obtained at different positions by the translator. The elementary images are merged to obtain an overall image of the object.

[0035] By a "translator position" is meant the position of the translator's reference system (190) in the common reference system (170). Said translator's position in the common reference system can be characterized by a translation and a rotation of the translator's reference system (190) relative to the common reference system (170).

[0036] By a "change of position of the translator" is meant the translation and rotation of the reference system of the translator between a first position and a second position of the translator. During the movement of the translator, to perform a scan of the object, the translator undergoes a change of position, as illustrated in [Fig.2],

[0037] By "raw data" or "raw data set" we mean the data acquired by the translator at a fixed position relative to the object.

[0038] By "elementary image" we mean the image reconstructed from a set of raw data, acquired at said fixed position. An elementary image is said to be "obtained" at a fixed position by acquiring the raw data at said position and by reconstructing the elementary image from these raw data. As mentioned before, the elementary image is "obtained" in the translator's reference system.

[0039] The elementary image can be reconstructed "in the volume of the object". In this case, the elementary image shows an interior of the object as seen by the ultrasonic waves. The elementary image can also, from the same raw data, be reconstructed "in the coupling medium". In this case, the elementary image shows the coupling medium and a surface of the object which forms the boundary between the coupling medium and the volume of the object. The reconstruction of the elementary image in the coupling medium makes it possible to highlight the surface of the object particularly well.

[0040] The translator used for this purpose is a multi-element translator. With this translator, the raw data set is acquired using an acquisition technique called FMC (Full Matrix Capture). This acquisition technique allows the elementary image to be reconstructed using a technique called "all-point focusing" (Total Focusing Method, TFM) or a technique called "adaptive Total Focusing Method, ATFM). Both methods allow the elementary image to be reconstructed in the coupling medium or in the volume of the object, as described previously.

[0041] By a "fusion" of elementary images we understand the combination of several elementary images in order to obtain an "overall image". The said overall image shows a reconstruction of the object based on several elementary images.

[0042] In order to merge the elementary images it is necessary to determine the change in position of the translator which took place between the positions where the elementary images were obtained. Knowledge of the change in position of the translator then makes it possible to correctly display the elementary images in the common reference frame (170).

[0043] A method for determining the change in position of the ultrasonic transducer will be described below. Said method comprises the following steps:

[0044] With the translator (50) a first (30) and a second elementary image (40) of the object are obtained, as illustrated in [Fig.2]. In other words, a first and a second set of raw data are acquired at the first (60) and the second position (70). The data sets are then reconstructed to obtain the first and the second elementary image.

[0045] The first and second positions must be chosen such that the same part of the object is present in a common part (80) of the first and second elementary images.

[0046] Then, the surface of the object is highlighted on the elementary images.

[0047] The change in position is then determined by applying a matching to the elementary images so as to cover the surface of the object in the common part of the elementary images.

[0048] The steps of the method for determining the change in position of the translator will be detailed later.

[0049] [Fig.2] shows the object (10) with the volume of the object (180) and its surface area (20). The ultrasonic translator (50) is shown at a first position (60) and at a second position (70), different from the first position. [Fig.2] also shows the reference system (190) of the translator and the common reference system (170) of the object.

[0050] As described above, the ultrasonic translator acquires the raw data in its own reference system (190). For this reason, the corresponding elementary image is also reconstructed in the reference system of the translator (190). At the first position, a first elementary image (30) is obtained. At the second position, a second elementary image (40) is obtained. [Fig. 2] also shows a common part (80) of the first image and the second image. The common part of the images is the part in which the first image and the second image overlap. The surface of the object in the common part (90) is also highlighted. Advantageously, the common part (90) represents 70% or more of the elementary image.

[0051] The translator can be moved between the first position and the second position using the robotic arm (130). The translator can also be moved by any other means, for example manually.

[0052] [Fig. 3] shows the first elementary image (30) obtained at the first position and the second elementary image (40), obtained at the second position. In each of the two elementary images, the surface (20) of the object is shown as well as the common part (80) of the two images and the surface present in the common part of each image (90). The two images are shown with the reference system (190) of the translator aligned with the common reference system (170).

[0053] Then, the surface (20) of the object (10) is highlighted in the first elementary image and in the second elementary image.

[0054] For example, the first and second elementary images may have been reconstructed in the volume of the object. The surface can be identified by visual inspection in each image. To then highlight the surface, the identified pixels are marked as belonging to the surface. It is also possible to use an automated recognition algorithm to identify the surface in the images.

[0055] It is also possible to have reconstructed the first and second images in the coupling medium (100). Since the coupling medium closely matches the surface of the object, the elementary images that have been reconstructed in the coupling medium clearly show the surface of the object.

[0056] The highlighting of the surface of the object on the elementary images reconstructed in the coupling medium can be improved by the following steps:

[0057] For each column of each elementary image a threshold value is defined. Then, only the pixels of the column which have a value above said threshold value are retained as belonging to the surface of the object. Said threshold value can be set manually. It is also possible to determine the threshold value according to a maximum value determined for each column of the image. Advantageously, the threshold value is determined by a method described in the publication "Nobuyuki Otsu (1979). "A threshold selection method from gray-level histograms". IEEE Trans. Sys. Man. Cyber. 9 (1): 62-66" (hereinafter called "Otsu").

[0058] This method according to Otsu, assumes that the surface of the object is what has the highest amplitude on the image. According to Otsu, the threshold value is determined using a histogram of the gray level amplitudes of the image. Alternatively, it is possible to transform the image of different gray levels into a binary image. To calculate the threshold value according to Otsu, the algorithm assumes that the binary image contains only two classes of pixels. Then the algorithm calculates the value that separates said two classes so as to minimize the intra-class variance. This value is the threshold value.

[0059] It is also possible to apply a one-dimensional smoothing filter to the elementary images. This can be, for example, a Savitzky-Golay filter, an averaging filter or a Gaussian filter.

[0060] Advantageously, only the surface in the common part (90) is retained in the first image and in the second elementary image and the other details are deleted.

[0061] Then, an image registration is applied to the obtained elementary images as described above. The matching causes the surface of the object to be covered in the common part of the elementary images and thus determines the change in position of the translator between the first and second positions. Said matching is illustrated in [Fig.4a].

[0062] [Fig.4a] shows the first (30) and the second (40) elementary image with the surface (20) of the object highlighted. The surface of the object in the common part (90) has been highlighted in both images by a thick line. It can be seen that the surface present in the common part overlaps between the first and second elementary images. To overlap the surface in the common part, a rotation (150) and a translation (30), as shown in [Fig.4a], were applied to the second elementary image (40). This rotation (150) and translation (30) correspond to the change in position of the translator between the first and second positions.

[0063] The transformation necessary to overlap the first and second elementary images can be written using a transformation matrix (T) and a matrix equation: ^position i = T ■ XposiTion 2 in which Xposition i and XpOSition 2 represent the first and second elementary images obtained at the first and second positions and T represents the transformation matrix.

[0064] Different methods can be used to perform said matching and determine the transformation matrix (T), such as for example a least squares method (MMC), an Iterative Closet Point (ICP) type method, a Robust Point Matching (RPM) method, a Kernel correlation method, or algorithms based on the Fourier-Merllin transform.

[0065] Subsequently a preferred implementation of the ICP method is described.

[0066] At the end of the step described above, we obtain, for each elementary image, a set of points representing the surface of the object.

[0067] Let | be a first set of points representing the first surface of the first elementary image and j a second set of points re presenting the second surface of the second elementary image. In other words, by in relation to the previous paragraph, ^pmitionï is the first set of points representing the first surface of the first elementary image and Posit«m2 is the second set of points representing the second surface of the second elementary image.

[0068] ICP matching is applied to the point sets S1 and S2. For example, a "Point to Point ICP" method is used. Alternatively, a "Point to Plane ICP" or "Point to Line ICP" method can be used.

[0069] The ICP algorithm makes it possible to iteratively minimize the distance between the sets of points S1 and S2. The sets of points represent the same surface of the object or each represent a part of the same surface. An initial point matching is not necessary for an application of the ICP method.

[0070] The sets Sj and S2 are linked by the following relation:

[0071] [Math.l] q^Rpt + O (1)

[0072] where R is a rotation matrix and O is a translation. The ICP method makes it possible to determine the transformation T which is the set of the rotation transformation R and the translation O and which minimizes the distance between the two sets of points and S2. Said distance is defined by:

[0073] [Math.2] dist. = S"=1lli7É - RPt ~ Oll2 [2)

[0074] The determination of the minimum distance is iterative. An index "k" can be associated with each iteration, where k takes the values ​​of k = 1, ..., G and where G represents the total number of iterations performed.

[0075] Each iteration includes the following operations:

[0076] 1. Operation 1: Matching operation. For each point belonging to the first set of points we must find a corresponding point belonging to the second set of points. We thus determine for each point of the first set, P a, the point, of the second set which is the closest according to the definition finishing the chosen distance. This is therefore a "Point to point ICP" method.

[0077] 2. Operation 2: Edge point rejection operation. The points located on the edge of the surface and for which no corresponding point has been found are rejected. [Fig.4b] shows this operation: For points ql and q2 no corresponding point among the set of points p has been identified. Points ql and q2 are thus rejected.

[0078] 3. Operation 3: Transformation calculation operation. The transformation is calculated by minimizing the distance criterion, determined by equation (2). For example, a Singular Value Decomposition method can be used for this operation which aims to obtain the rotation R and the translation O.

[0079] 4. Operation 4: Operation of updating the positions of the points. By the mini By implementing the distance criterion, we obtained a transformation (T) which allows us to transform the points of the first set to obtain the points of the second set.

[0080] 5. Operation 5: Error calculation operation. At the end of each iteration we must evaluate the error between S2 and SI. If the error is less than a selected error threshold, the algorithm can be stopped. If the solution has not been found after a fixed number of iterations, the algorithm is stopped anyway.

[0081] When all the iterations have been completed, either because the error is less than the selected error threshold, or because the maximum number of iterations has been reached, we can say that we have found the optimal transformation T (the transformation T including rotation and translation).

[0082] A method for obtaining an overall image of the object comprises the method described above. At the end of this method we have a first set of raw data acquired with the translator placed at a first position and a second set of raw data acquired at the second position, as shown in [Fig.l]. We also have knowledge of the change in position of the translator between the first and second positions, for example in the form of the matrix T described previously.

[0083] The first and second sets of raw data can be reconstructed to obtain elementary images in the coupling medium, in the volume of the object or according to any other suitable reconstruction method. The elementary images thus reconstructed will be merged to obtain an overall image of the object. In order to correctly merge the elementary images in the common reference frame (170) it is necessary to compensate for the displacement of the reference system of the translator (190) between the first and second positions of the translator (see [Fig.2]). The compensation to be applied corresponds to the change in position of the translator between the first and second positions, for example in the form of the matrix T, determined previously.

[0084] [Fig.5] shows the first (30) and the second (40) elementary image with their reference system (190). The elementary images are displayed in the common reference system (170) in order to form an image of the whole object (10). The whole image of the object is thus obtained from the first and the second elementary image by repositioning the first elementary image and the second elementary image in a common reference system using the transformation matrix (T).

[0085] [Fig.6] shows a preferred embodiment of the method for obtaining an image of an object. Said method comprises the acquisition of raw data with the ultrasonic translator, the reconstruction of the elementary images based on the raw data obtained at each position of the translator on the object, then the fusion of the elementary images to obtain the image of the entire object.

[0086] In a first phase (210), the ultrasonic transducer is placed at a first position relative to the object. The transducer comprises the flexible membrane shoe which is placed between the transducer and the object. The membrane shoe is filled with the coupling medium and ensures contact between the transducer and the object. Alternatively, the object is immersed in the coupling medium and the transducer does not comprise the flexible membrane shoe.

[0087] In a second phase (220), the first raw data set is acquired with the translator at the first position. The first raw data set is acquired using an acquisition technique called FMC (Full Matrix Capture). This also remains the case for all subsequent acquisitions. The first raw data set corresponds to the first elementary image.

[0088] In a third phase (230), the translator is moved to the second position, different from the first position. The second position of the translator is chosen so that the image corresponding to the second position includes a part of the object also present in the image corresponding to the first position. The second set of raw data corresponding to the second elementary image is then acquired.

[0089] Advantageously, the translator is then moved to a third position, different from the second position in order to acquire a third set of raw data corresponding to a third elementary image having an overlap with the second elementary image. Said phases are repeated in order to scan the object and acquire a plurality of raw data at a plurality of positions. After reconstruction, each reconstructed elementary image will have a common part with at least one other elementary image.

[0090] In a fourth phase (240), the first elementary image and the second elementary image are reconstructed from the first raw data set and the second raw data set. The first elementary image and the second elementary image are reconstructed in the coupling medium. Then, a threshold for each column of each elementary image is defined. Preferably, said threshold is defined from the maximum of each column by the Otsu method [8]. Only the pixels having a value above said threshold value are retained in each column of the elementary image. Afterwards, a smoothing filter can be applied to the elementary images. Said smoothing filter can be used if the elementary images or the raw data are very noisy. Only the surface of the object is thus present in the two elementary images.

[0091] Optionally, it is possible to retain only the part of the surface which is present in the common part of each elementary image. Advantageously, it is not obligatory to retain only the part of the surface which is present in the common part of each elementary image if an ICP method is used.

[0092] In a fifth phase (250), the matching is applied to the first and second elementary images. The transformation matrix T is determined by the matching algorithm in order to cover the surface of the visible object in the first and second elementary images according to the equation:

[0093] ^position i — TXpositùm 2 in which position j and ^position 2 represent the first and second elementary images or, in other words, the point cloud of the surface of the first or second elementary images.

[0094] In a sixth phase (260), the first elementary image and the second elementary image are again reconstructed from the first and second raw data sets. This time, the first elementary image and the second elementary image are reconstructed in the volume of the object.

[0095] The transformation matrix T is applied to merge the first elementary image and the second elementary image. The two elementary images are displayed in a common frame of reference in order to obtain the image of the entire object as examined by the ultrasonic waves.

[0096] As described before, the fourth, fifth and sixth phases can be applied several times to pairs of elementary images in order to reconstruct the image of the whole object more complete.

Claims

Claims

1. Method for determining a change in position of an ultrasonic translator (50), the method comprising the following steps: - obtaining with the translator (50) a first (30) and a second elementary image (40) of an object having a surface (20), the first elementary image being obtained with the translator placed at a first position (60), the second elementary image being obtained with the translator placed at a second position (70), different from the first position, so that the same part of the object is present in a common part (80) of the first and second elementary images, - highlighting the surface of the object on the elementary images, - determining the change in position by applying a matching to the elementary images so as to cover the surface of the object in the common part of the elementary images,in which: - the elementary images are obtained with a coupling medium (100) arranged between the translator and the surface of the object, and - the step of highlighting the surface of the object comprises the reconstruction of the elementary images in the coupling medium by a method of focusing at all points, each elementary image showing the coupling medium and a surface of the object which forms the boundary between the coupling medium and the volume of the object.,

2. Method according to one of the preceding claims in which the highlighting of the surface of the object comprises the steps of: - defining a threshold for each column of each elementary image, - highlighting, as points of the surface, the pixels of each column (110) having a value above said threshold, - preferably applying a one-dimensional smoothing filter to each elementary image.

3. Method according to one of the preceding claims comprising the step of: - highlighting the surface in the common part in each elementary image before applying the matching.

4. Method according to one of the preceding claims in which the matching of the elementary images comprises the step of: - determining a transformation matrix (T) between the first and the second image, preferably by a closest point iteration (CPI) method.

5. Method for obtaining an overall image of an object, the method comprising the steps of: - implementing the method according to at least one of claims 1 to 4 to obtain a first and second elementary image at a first and second position and a change in position of the translator between the first and second position, - using the determined change in position to display the first image and the second elementary image in a common reference frame in order to obtain the overall image of the object.

6. A method according to claim 5 wherein the object is immersed in a coupling medium, preferably water.

7. The method of claim 5 wherein the translator comprises a flexible membrane shoe (120) filled with a coupling medium, preferably water, the flexible membrane shoe being positioned between the object and the translator.

8. Method according to one of claims 5 to 7 in which the translator is moved using a robotic arm (130) between the first and second position.

9. Method according to one of claims 7 to 8 in which - the translator performs a scan (140) of the object and - a plurality of elementary ultrasonic images is obtained at a plurality of positions of the translator and - the steps of determining the change of position and display in the common reference frame are applied several times to pairs of images.