Velocimetry method using image processing of a fluid flow

Abstract

Title: Velocimetry Method by Image Processing of Fluid Flow The invention relates to a velocimetry method by image processing of fluid flow (10) in a pipe (20), in which tracers (30) are present in the fluid. A source (40) located outside the pipe emits acoustic signals incident on an inspection volume (100), which are then reflected by the tracers and by interfaces between the fluid and media exhibiting different acoustic impedances. The reflected signals are received by a sensor (50) to constitute a plurality of signals from the inspection volume.A first and second plurality of signals from the inspection volume are acquired in the presence of the tracers at two different times, and are then subtracted to eliminate signals from the interface and obtain residual signals forming a residue, which is then processed to obtain a final image where each tracer is identified at a given time. Figure for the abstract: Fig. 10.

Description

Title of the invention: Velocimetry method by image processing of a fluid flow. Technical field

[0001] The invention relates to the field of fluid flow measurement. More particularly, the invention relates to ultrasonic imaging. Its particularly advantageous applications include the development of passive heat extraction systems, industrial processes requiring the characterization of flow velocities (mixing zone, unsteady flow, etc.), and the medical, geophysical, and nuclear fields. STATE OF THE ART

[0002] A velocity field is a 2D image representing, at each point, the intensity and direction of the velocity of the fluid under study. This measurement allows for the precise study of a fluid's behavior, as it enables the visualization of its movements, such as turbulence, the appearance of eddies, recirculation zones, etc. It also allows for the deduction of a pressure field or dynamic viscosity. However, obtaining direct information about the flow of a fluid in a precise manner is quite complex. Therefore, an approach exists for performing such velocimetry measurements, which relies on the use and imaging of tracers in the fluid under study. Tracers, such as microparticles, are introduced into the flow and carried along by it. These tracers describe a motion similar to that of the fluid. It is then possible to perform measurements on these tracers to characterize the fluid's motion.

[0003] Conventional methods for making such measurements are optical methods, which are limited by the need for transparency of both the fluid and the conduit through which the fluid flows. To overcome this limitation, acoustic measurement methods can be used. These are primarily developed in the medical field, but in most cases, they prove unsuitable for industrial applications.

[0004] One method based on acoustic inspection is echo-PIV (Particle Image Velocimetry). It relies on the precise calculation of acoustic images of tracer clouds present in a flow, the motion of which is analyzed over time. Based on the overall motion of these tracer clouds, it is possible to deduce the motion and velocity of the fluid at each point in an image. Echo-PIV can be performed using a topological energy approach. It is an innovative method, enabling... Precise measurements with a high acquisition rate. Their ability to perform image reconstructions for echo-PIV was demonstrated by Rodriguez S. (doi.org / 10.1121 / 1.4799639). In this document, the acoustic probe must be immersed in the fluid, which is not feasible for an industrial application requiring a non-intrusive probe in the flow.

[0005] The present invention proposes to partially overcome the drawbacks of known methods. In particular, an objective of the present invention is to improve the velocimetry method using echo-PIV.

[0006] Another objective of the present invention is to propose a velocimetry solution that does not require the introduction of a probe through a pipe and that is suitable for non-intrusive industrial applications. SUMMARY

[0007] To achieve this objective, according to one embodiment, a velocimetry method is provided according to a first aspect of the invention, by image processing of a fluid flow inside a pipe, tracers being present in the fluid, the method comprising: • two emissions at two different times by an incident acoustic signal source configured to propagate at least part of the acoustic signals into a volume of the pipe called the inspection volume in which the incident acoustic signals are reflected by the tracers present in the inspection volume and by at least one interface between the fluid and a medium having an acoustic impedance different from that of the fluid, to form reflected acoustic signals, • reception by a sensor of so-called reflected acoustic signals resulting from the reflection of incident acoustic signals within the inspection volume, so as to acquire a first plurality of reflected signals and then a second plurality of reflected signals, • a subtraction using a processor of the first and second pluralities of reflected signals in order to eliminate the part of the reflected acoustic signals coming from at least one interface and obtain residual acoustic signals forming a third plurality of signals called the residue, • a processing of the residual by the processor to obtain at least one final image of the inspection volume in which each tracer of the inspection volume is identified at two given times.

[0008] The source is located outside the conduit.

[0009] The first plurality of reflected signals is acquired at a time t1 and the second plurality of reflected signals is acquired at a time t2 different from time t1, the tracers being present in the fluid, at least in the inspection volume, at times t1 and t2. Thus, the inspection volume in each of the first and second plurality of reflected signals includes tracers.

[0010] The method described above makes it possible to study the flow of a fluid in a pipe by an acoustic approach, in a non-intrusive manner by taking into account the presence of walls or static obstacles in the flow containing tracers.

[0011] Classically, topological energy acoustic imaging is performed by subtracting two plurality of signals: reference signals without inclusions and signals to be studied. This approach works well when the probe is immersed in the fluid where there is no change of medium. However, during the development of the present invention, it was observed that if the probe is not immersed in the fluid, this approach does not effectively eliminate, in the generated images, direct reflections of acoustic signals at the interfaces between the fluid and a different medium such as the pipe wall, particularly when the fluid contains tracers such as microbubbles of gas.

[0012] Indeed, the presence of tracers in the flow modifies the acoustic characteristics of the fluid, particularly its velocity. This change in velocity induces a variation in the acoustic impedance of the fluid, which in turn influences the intensity of the signal reflected by the interface between the pipe and the fluid. Thus, the amplitude of the signal reflected by the pipe / fluid interface differs between measurements with and without tracers. Subtracting the two measurements does not eliminate this acoustic contribution, as it differs between the two measurements. This prevents the use of this type of image for VIP (Visual Imaging). This problem, identified during the development of the present invention in non-intrusive measurements of a fluid containing tracers, is solved by the method of the present invention.This method proposes to effectively eliminate the contribution of interfaces of media changes, by measuring two successive states of the flow in the inspection volume containing tracers, without resorting to reference signals devoid of tracers.

[0013] This method is suitable for ultrasonic imaging of two-phase flows behind a potentially opaque wall. For example, it allows for non-intrusive acoustic PIV measurements. This is of major interest for potential industrial applications that cannot benefit from advances in acoustic velocity field measurement techniques in the medical field because these techniques are not suitable for non-physiological fluid flows confined behind opaque walls. Indeed, in the medical field, the propagation media are made of biological tissues and fluids whose acoustic behaviors are similar to each other.

[0014] A second aspect of the invention relates to a computer program product, preferably recorded on a non-transient medium, comprising instructions which, when executed by a digital data processing device, for example a processor, perform at least the following steps: • a subtraction of the first and second pluralities of reflected signals so as to eliminate the part of the reflected acoustic signals coming from at least one interface and to obtain residual acoustic signals forming a third plurality of signals called the residue, the first plurality of reflected signals coming from the inspection volume being acquired at a time t1 and the second plurality of reflected signals coming from the inspection volume being acquired at a time t2 different from time t1, • a processing of the residual to obtain at least one final image of the inspection volume in which each tracer of the inspection volume is identified at a given time.

[0015] A third aspect of the invention relates to a velocimetry system by image processing of a fluid flow in a pipe, the system comprising: • an acoustic signal source configured to emit incident acoustic signals onto a volume of the pipe called the inspection volume, and configured such that, in the presence of fluid in the pipe and tracers in the fluid, the incident acoustic signals are reflected by the tracers present in the inspection volume and by at least one interface between the fluid and a medium having a different acoustic impedance than that of the fluid, to form reflected acoustic signals, • at least one sensor configured to receive acoustic signals, known as reflected signals, resulting from the reflection of incident acoustic signals within the inspection volume, • at least one processor, the system being configured to perform the following steps: • an acquisition by the sensor of a first and second plurality of reflected signals originating from the inspection volume, • a subtraction using the processor of the first and second pluralities of reflected signals in order to eliminate the part of the reflected acoustic signals coming from at least one interface and to obtain residual acoustic signals forming a third plurality of signals called the residue, • a processing of the residual by the processor to obtain at least one final image of the inspection volume in which each tracer of the inspection volume is identified at a given time.

[0016] The source is located outside the pipe. The system is configured so that the first plurality of reflected signals is acquired at a time t1 and the second plurality of reflected signals is acquired at a time t2 different from time t1, the tracers being present in the inspection volume at times t1 and t2. BRIEF DESCRIPTION OF THE FIGURES

[0017] The aims, objects, features and advantages of the invention will become clearer from the detailed description of an embodiment thereof, which is illustrated by the following accompanying drawings in which:

[0018] [Fig.1] Fig.1 illustrates a flowchart showing steps in the velocimetry process by image processing of a fluid flow according to an example.

[0019] [Fig.2] [Fig.3] Figures 2 and 3 schematically illustrate the flow in a conducting a fluid containing tracers.

[0020] [Fig.4A] [Fig.4B] [Fig.4C] Figures 4A to 4C represent acoustic measurements according to a classical approach.

[0021] [Fig.5A] [Fig.5B] [Fig.5C] Figures 5A to 5C represent acoustic measurements according to an example of the method of the present invention.

[0022] [Fig. 6] Figure 6 illustrates an image of an inspection volume obtained by subtraction of two pluralities of reflected signals according to the classical approach.

[0023] [Fig.7] Figure [Fig.7] illustrates a residue obtained by subtracting two pluralities of signals reflected according to the method of the present invention.

[0024] [Fig. 8] Figure 8 illustrates a flowchart showing sub-steps of the processing of the residue according to an example of the present invention.

[0025] [Fig.9] Fig.9 illustrates a numerical model of the inspection volume according to an example.

[0026] [Fig. 10] The [Fig. 10] schematically illustrates a velocimetry system by image processing of a fluid flow according to an example.

[0027] The drawings are given by way of example and are not limiting of the invention. They constitute schematic representations of principle intended to facilitate understanding of the invention and are not necessarily to scale with practical applications. DETAILED DESCRIPTION

[0028] Before proceeding to a detailed review of embodiments of the invention, optional features that may be used in combination or alternatively are listed below:

[0029] According to one example, the inspection volume has a concentration Cl of tracers at time t1, and the inspection volume has a concentration C2 of tracers at time t2, the times t1 and t2 being chosen so that a relative difference between Cl and C2 is less than or equal to 30%, preferably less than or equal to 10%.

[0030] According to one example, the first plurality of signals is acquired at time t1 and the second plurality of signals is acquired at time t2, different from time t1, such that the time separation At = t2 - t1 is less than or equal to 150 ms. This value is determined as follows: in order for the tracer concentration between times t1 and t2 to be sufficiently close, at least half of the tracers present at time t1 must also be present in the inspection volume at time t2. Considering L to be the smaller of the two dimensions of the study area and vmax to be the maximum velocity of the tracers in the area, At must be less than or equal to 0.5 L / vmax. The study area can be projected onto a two-dimensional plane. Typically, this plane contains the principal axis of the pipe's extension at the level of the inspection volume.

[0031] The maximum speed of movement of the tracers, vmax, corresponds to the observed speed of the fastest tracers within the observation volume. This speed can, for example, be estimated by Computed Fluid Dynamics (CFD) simulation.

[0032] The smaller of the two dimensions of the study area, denoted L, corresponds to the smaller dimension between the length and the width of the image.

[0033] L corresponds to the smaller of the two dimensions of the volume under study reduced to a two-dimensional 2D plane. This 2D plane is, for example, a plane containing the axis of revolution of the pipe if the latter has a circular cross-section.

[0034] According to a non-limiting example, At is less than or equal to 5000 ms (103 seconds), preferably less than or equal to 150 ms.

[0035] This temporal separation value At between the first and second plurality of signals is chosen so that the acoustic properties of the inspection volume do not change, or change only slightly, between times t1 and t2. When the first and second plurality of reflected signals are close in time, the inspection volume is characterized at times t1 and t2 by similar populations, or concentrations, of tracers. During the development of the present invention, it was found that the change in acoustic speed at times t1 and t2 is sufficiently small to provide a perfectly satisfactory result. Thus, the acoustic impedance of the inspection volume at times t1 and t2 will be similar, and the reflection of the acoustic signals at the interfaces between the fluid and another medium will also have a similar intensity.In this way, subtracting the two images will greatly reduce, or even eliminate, the acoustic contribution of the objects. static, given that the acoustic propagation of the incident signals takes place under similar conditions at times t1 and t2 of the two measurements.

[0036] According to one example, a minimum value of the temporal separation At depends on the minimum speed of movement of the tracers v0 and the size d of the tracers. According to one example, At is greater than or equal to 0.25d / vo, According to another example, At is greater than or equal to 0.5 ms, preferably greater than or equal to 3 ms.

[0037] According to one example, At > 0.5 ms and preferably At > 3 ms and preferably At > 10 ms. According to one example, At > 100 ms. According to one example, At > 1 s.

[0038] According to an example, At < 5000 ms and preferably At < 1000 ms and preferably At < 1000 ms.

[0039] Typically, the size d of a tracer is its maximum dimension. This can be a diameter if the tracer is a sphere, or a large diameter if its surface forms an ellipsoid. It can be a length if the tracer has an elongated shape, such as a fiber.

[0040] The minimum speed of movement of the tracers v0 corresponds to the observed speed of the slowest tracers within the observation volume. This speed can, for example, be estimated by Computed Fluid Dynamics (CFD) simulation.

[0041] Although the temporal separation At between the first and second pluralities of signals is advantageously minimized, it is important that At be large enough to allow minimal displacement of the tracers from one image to the other. These minimal values ​​of At allow a tracer to appear at two different locations within the inspection volume at times t1 and t2. This optimizes the performance of the image processing algorithm.

[0042] According to one example, the treatment of the residue includes: • a temporal reversal of residual acoustic signals, • a digital model of the inspection volume to form a virtual domain, • a backpropagation of residual acoustic signals, temporally returned into the virtual domain, to generate an adjoint field, • generation in the same virtual domain of a direct field by modeling the propagation of incident acoustic signals in the virtual domain devoid of tracers, • a convolution of the adjoint field and the direct field to form the final image.

[0043] This approach makes it possible to efficiently detect impedance contrasts in an inspected medium. With good spatial and temporal discretization, it can accurately locate and measure impedance variations, even in the presence of low-contrast heterogeneities. Furthermore, this approach makes it possible to obtain a high signal-to-noise ratio thanks to the combined effect of re-focusing of the adjoint field (time reversal) and spatio-temporal correlation via the direct field.

[0044] According to one example, the treatment of the residue includes: • a digital model of the inspection volume to form a virtual domain, • the generation of a radiation pattern of the source in the virtual domain, • a calculation of the Fourier transforms of residual acoustic signals and incident acoustic signals in the virtual domain, • a multiplication of the Fourier transform of the residual acoustic signals by the radiation pattern of the source to generate an adjoint field, • a multiplication of the Fourier transform of the incident acoustic signals by the radiation pattern of the source to generate a direct field, • a multiplication of the adjoint field by the direct field to form the final image.

[0045] This approach makes it possible to considerably accelerate image processing. As the range of frequencies of interest is reduced compared to the sampling frequency, the useful data is smaller than in the time domain.

[0046] According to one example, tracers are microbubbles formed by injecting a gas into the fluid.

[0047] The use of gaseous tracers offers several advantages. Due to their small size, the gas microbubbles accurately track fluid movements without disturbing the flow. Their high acoustic reflectivity, resulting from the high impedance contrast between the gas and the surrounding liquid, allows for the acquisition of clear and detailed acoustic images. This is advantageous for applications in the nuclear field. For example, liquid sodium is being considered for cooling Generation IV nuclear reactors. Liquid sodium normally and continuously contains argon microbubbles.

[0048] According to other examples, tracers are solid microparticles introduced into the fluid, such as mineral microspheres, polystyrene, glass, metal, carbon, etc.

[0049] A microbubble or microparticle has a maximum dimension cross-section Sm less than or equal to 500 pm (10 6 meters), and preferably greater than or equal to 5 pm (106 m).

[0050] In one example, the velocimetry system includes a device for introducing tracers into the fluid. In another example, all or at least some of the tracers are already present in the fluid.

[0051] According to one example, the fluid is a metal or a mixture of metals in liquid phase.

[0052] According to one example, the fluid is liquid metallic sodium.

[0053] The outside of the pipe is separated from the inside of the pipe by a wall.

[0054] This method makes it possible to study liquids that are opaque to visible light or that flow behind walls opaque to visible light, which is not possible with conventional optical methods requiring transparency of the fluid and the pipe.

[0055] For example, the fluid can be water, blood, oil, saline solutions, gels, emulsions, coolants, chemical mixtures, etc. The process is applicable to a wide range of fluids as long as they allow the propagation of acoustic waves and the circulation of tracers.

[0056] According to one example, the emission of incident acoustic signals is carried out by the source which comprises a plurality of components based on a piezoelectric material.

[0057] According to one example, the reception of reflected acoustic signals is achieved by the sensor which comprises a plurality of components based on a piezoelectric material.

[0058] According to one example, the emission of incident acoustic signals and the reception of reflected acoustic signals are carried out by the same transducer which is both the source and the sensor.

[0059] Alternatively, the emission of incident acoustic signals and the reception of reflected acoustic signals are carried out by two separate transducers. One thus acts as a source of acoustic signals and the other as a sensor.

[0060] According to one example, the method includes obtaining several successive final images of the fluid flow over time and reconstructing a velocity field of the tracers using a PIV algorithm.

[0061] This makes it possible to measure fluid flow velocity fields, by precisely correlating, using the method of the invention, from the final images obtained, the positions of the tracers at different times in the pipe in order to reconstruct their movement over time.

[0062] According to one example, the pipe has a wall separating the inside from the outside of the pipe, the wall being opaque to visible light.

[0063] According to one example, the conduit is based on a metal, plastic or glass.

[0064] Several embodiments of the invention implementing successive steps of the manufacturing process are described below. Unless explicitly stated, the adjective "successive" does not necessarily imply, even though this is generally preferred, that the steps follow each other immediately, with intermediate steps potentially separating them.

[0065] Furthermore, the term "step" refers to the execution of a part of the process, and can designate a set of sub-steps.

[0066] Furthermore, the term "step" does not necessarily mean that the actions carried out during a step are simultaneous or immediately successive. Certain actions of a first step may, in particular, be followed by actions related to a different step, and other actions from the first step may be repeated later. Thus, the term "step" does not necessarily imply unitary actions that are inseparable in time and in the sequence of phases of the process.

[0067] The terms "approximately", "around", "in the order of" mean to the nearest 10%, and preferably to the nearest 5%. Furthermore, the terms "between ... and ..." and equivalents mean that the bounds are inclusive, unless otherwise stated.

[0068] A "tracer" is defined as a solid particle, gas bubble, or liquid droplet introduced into a fluid to track and visualize the movement of that fluid. Tracers move in synchrony with the fluid, without disturbing its physical properties. Their movement is then captured and analyzed using imaging techniques to measure the speed, direction, and other characteristics of the fluid flow.

[0069] The term "conduit" means a tube, channel, or closed or semi-closed structure containing a fluid in motion. Thus, the term conduit is equivalent to the term channel.

[0070] A "static" object is defined as an element or structure that is fixed and stationary relative to the fluid. This object can influence the flow by changing its direction or speed, or by creating areas of turbulence or separation. A static object could be, for example, an obstacle, a pipe wall, or a structural element within a conduit.

[0071] An "acoustic signal" is understood to be a mechanical oscillation or wave that propagates through a medium (such as air, water, or solids) in the form of longitudinal or transverse pressure waves. These waves are generated by the vibration of a sound source, creating successive zones of compression and rarefaction in the medium.

[0072] The term "PIV algorithm" refers to a mathematical method commonly used to reconstruct the velocity field of a fluid by analyzing successive images showing the position of tracers at different times. The algorithm can then divide these images into small areas, called correlation windows, and compare the displacements of the tracers between two times using cross-correlation. The displacement of the tracers allows the velocity vectors of the fluid in each analyzed area, and this information is then combined to reconstruct an overall fluid velocity field.

[0073] The velocimetry method by image processing of a flow of a fluid 10 in a pipe 20 is now described with reference to Figures 1 and 2. [Fig.1] illustrates a flowchart representing different stages of the velocimetry method.

[0074] As illustrated in Figures 1 and 2, the method comprises introducing a plurality of tracers 30 into a fluid 10 flowing in a pipe 20. The pipe 20 includes a wall 25 enclosing the fluid 10. The tracers 30 are randomly dispersed in the fluid 10 and follow the flow and movements of the fluid 10 in the pipe 20. The tracers 30 thus form an imprint of the dynamics of the fluid 10 within the pipe 20. In order to probe these tracers 30, inspection sound waves are used.

[0075] The method comprises an emission E2, by an acoustic source 40, of acoustic signals Si incident on the fluid 10. This emission E2 is carried out in a non-intrusive manner, that is to say, the source 40 is located outside the pipe 20, without being in direct contact with the fluid 10. This avoids modifying the pipe 20 and disrupting the flow of the fluid 10 or contaminating it. Consequently, the incident acoustic signals Si first encounter the wall 25 outside the pipe 20. At least a portion of these incident acoustic signals Si propagates into a volume of the pipe 20, referred to as the inspection volume 100. These acoustic signals Si, propagating within the inspection volume 100, pass through the wall 25 of the pipe 20 and are then transmitted to the fluid 10.

[0076] During a passage from a first medium to a second medium, the acoustic signals Si can be partially transmitted into the second medium and partially reflected back into the first. This reflection depends on the acoustic properties of the two media, in particular on a physical quantity called acoustic impedance. An interface 60 between the fluid 10, which has a certain acoustic impedance, and another medium with a different acoustic impedance than that of the fluid 10 can therefore generate a partial reflection of the acoustic signals Si. Thus, when the acoustic signals Si propagate through the wall 25 and then arrive at a first interface 60 between the wall 25 and the fluid 10, a portion of these signals can be reflected back into the wall 25 and / or onto the first interface 60.

[0077] The incident acoustic signals Si then propagate through the fluid 10 and encounter tracers 30, for example a cloud of tracers 30, present in the inspection volume 100. These tracers 30 also have a different acoustic impedance than the fluid 10. Consequently, the acoustic signals Si incident on the tracers 30 are reflected by them. The inspection volume 100 may also include static objects, other than the wall 25, such as an obstacle 27 as illustrated in [Fig. 2]. This obstacle 27, having a different acoustic impedance than the fluid 10, forms a second interface 60 with the fluid. This can generate reflections of the incident acoustic signals Si on this second interface 60.

[0078] The method further includes a reception E3 of all the acoustic signals Sr reflected by the interfaces 60 and the tracers 30 using a sensor 50 located outside the pipe 20. This reception E3 is therefore carried out in a non-intrusive manner. For example, the inspection volume 100 corresponds to the portion of space from which the reflected signals Sr originate and reach the sensor. This portion of space includes fluid, tracers, possibly a section of wall, or even an obstacle.

[0079] These reflected acoustic signals Sr received by the sensor 50 form an acoustic image of the inspection volume 100. This acoustic image includes echoes from the tracers 30, which allows the movements of the fluid 100 to be inspected, as well as echoes coming directly from the interfaces 60 between the fluid 10 and static objects such as the wall 25 or the obstacle 27.

[0080] The Sr acoustic signals from the interfaces 60 between the fluid and the static objects are generally more intense than the Sr acoustic signals from the tracers 30. This is because the tracers 30 are typically much smaller than the static objects. Thus, the echoes from the tracers 30 can be masked by the acoustic signals reflected by the static objects in the inspection volume, particularly the Sr acoustic signals reflected at the first interface 60 between the wall 25 and the fluid 10. These signals are relatively strong and can sometimes be subject to multiple reflections and artifacts.

[0081] As illustrated in [Fig. 1], the method further comprises an acquisition step E4 by the sensor. This step comprises or consists of two acquisitions. A first plurality of reflected signals Sri of the inspection volume 100 is acquired at a time t1, as illustrated, for example, in [Fig. 2]. In this first plurality of reflected signals Sri, tracers 30 present in the inspection volume 100 are captured at a time t1 during the time evolution of the flow of the fluid 10 in the pipe 20. After a certain time At, a second plurality of reflected signals Sr2 of the inspection volume 100 is then acquired by the sensor 50 at a time t2 different from t1, such that At = t2 - t1. [Fig. 3] illustrates an example of the second plurality of reflected signals Sr2 acquired at t2, in which at least some of the same tracers 30 that were captured in the first plurality of reflected signals Sri are present.The temporal separation At between the acquisition of the first and second pluralities of reflected signals Sri, Sr2 allows the tracers 30. which reflect these signals move between times t1 and t2 to reflect signals from different locations among the pluralities of reflected signals Sri and Sr2, as illustrated in [Fig. 3]. Thus, if these pluralities of reflected signals Sri and Sr2 are transposed into images, the tracers 30 that reflect these signals appear at different positions in the images taken at times t1 and t2.

[0082] In order to reduce, or even eliminate, the acoustic contribution of static objects in the acquired images, the method includes a subtraction E5 of the first and second pluralities of reflected signals Sri, Sr2. This subtraction E5 is performed digitally using a processor 1100. It eliminates the portion of the reflected acoustic signals Sr originating from the interfaces 60 between the fluid and static objects such as the wall 25 and the obstacle 27. This subtraction E5 results in residual acoustic signals originating primarily from the tracers 30 present in the inspection volume. All of these residual acoustic signals form a third plurality of reflected signals Sr3 called the "residue".

[0083] The subtraction E5 of the two pluralities of reflected signals Sri, Sr2 from the inspection volume containing tracers 30 within the framework of non-intrusive velocimetry makes it possible to circumvent the problem of the evolution of the acoustic properties of the interfaces 60 over time. Indeed, this evolution of the acoustic properties of the interfaces 60 is observed in particular when the source 40 and / or the sensor 50 are located outside the pipe and are separated from the fluid 10 by a wall 25. This evolution is mainly due to a modification of the acoustic speed of the fluid 10 in the presence of the cloud of tracers 30 circulating near the interfaces 60. The acquisition E4 of the first and second pluralities of reflected signals Sri, Sr2 is therefore carried out in the continuous presence of the tracers 30 in the inspection volume 100 without the need for a reference image devoid of tracers.Since the times t1 and t2 are advantageously close in time, the acoustic properties of the interfaces 60 do not change, or change very little, from one image to the next. The two plurality of reflected signals Sri and Sr2 are therefore acquired under conditions that are close or almost identical. This makes it possible to effectively reduce or eliminate the acoustic contribution of static objects in the residual and to highlight the acoustic contribution of the tracers 30 in the residual.

[0084] The method then includes an E6 processing of the residual using the processor 1100. Substeps of the E6 processing of the final image will be described later. This E6 processing makes it possible to obtain a final image of the inspection volume 100 in which each tracer 30 of the inspection volume 100 is identified at a given instant. This makes it possible to reconstruct a velocity field which requires the identification and precise positioning of each tracer in the images acquired over time. The velocimetry method according to the present invention makes it possible to obtain Images of sufficient quality to allow the reconstruction of the velocity field. To perform this reconstruction, the process may include obtaining several successive final images of the fluid flow. These images can feed a PIV algorithm to reconstruct a velocity field of 200 tracers.

[0085] In order to improve the quality of the obtained residual and that of the residual processing E6, the choice of times t1 and t2 is optimized. These times can be chosen so that the temporal separation At between the pluralities of reflected signals Sri and Sr2 is sufficiently large to allow detectable displacement of the tracers in the inspection volume 100, thereby improving the processing of the residuals by the algorithm. The temporal separation At can, for example, be greater than or equal to 3 ms. Considering L to be the smaller of the two dimensions of the study area and vmax to be the maximum velocity of the tracers in the area, At must be greater than or equal to 0.25 d / v0.

[0086] Furthermore, the times t1 and t2 can be chosen such that the time separation At between the pluralities of reflected signals Sri and Sr2 is small enough to avoid changes in measurement conditions, in particular changes in the acoustic impedance of the fluid 10 at the interfaces 60, in order to ensure effective elimination of the acoustic contribution from static objects. As an example, the time separation At can be less than or equal to 150 ms. The upper bound of the time separation depends on the minimum travel speed of the tracers v0 and the size of the tracers d. At must be less than or equal to 0.5 L / vmax.

[0087] Furthermore, the times t1 and t2 can be chosen according to the evolution of the tracer population in the inspection zone. At time t1, the inspection volume 100 has a concentration Cl, and at time t2, the inspection volume 100 has a concentration C2. The times t1 and t2 can be chosen so that the tracer cloud present in the inspection volume 100 does not change significantly, allowing the tracers 30 to be identified from one image to the next. In other words, this choice of times is optimized so that the tracer population does not have time to completely leave the inspection volume 100, thus enabling the overall movement of the tracer cloud to be tracked. This also preserves the acoustic properties of the fluid at the interfaces 60.Precise control of the tracer concentration 30 in the inspection volume 100 allows measurements to be taken at later times t1 and t2 with quite satisfactory results. For example, t1 and t2 can be chosen so that the relative difference between Cl and C2 is less than or equal to 30%, preferably less than or equal to 10%.

[0088] Figures 4A to 4C and 5A to 5C illustrate, respectively, examples of non-intrusive measurements of acoustic signals Sr according to a conventional approach and according to the method of the present invention. The x-axis represents time measured in seconds. The ordinate axis 302 represents the amplitude of the acoustic signals Sr.

[0089] As illustrated in Figures 4A and 4B, according to the conventional approach, an acoustic signal 401 from the inspection volume 100 including tracers, and a reference acoustic signal 402 from the inspection volume 100 without tracers 30 are measured. The frames 310 in Figures 4A and 4B indicate the portions of the signals 401, 402 originating from the first interface 60 between the fluid and the wall 25 of the pipe 20. The frames 320 in Figures 4A and 4B indicate the portions of the signals 401, 402 originating from the second interface 60 between the fluid 10 and the obstacle 27. These framed portions of the signals 401, 402 are intense enough to mask the contribution of the tracers to the acoustic signal 401. In order to highlight the acoustic contribution of the tracers 30, these two signals 401, 402 are subtracted.

[0090] The residual acoustic signal 403 resulting from the subtraction is illustrated in [Fig. 4C]. This result shows that subtracting the signals 401, 402 according to the classical approach does not eliminate the acoustic contribution of static objects as explained previously. Indeed, due to the variation in the acoustic impedance of the fluid 10 in the presence of tracers 30, the amplitude of the acoustic signal reflected by the interface 60 between the fluid and the wall 25 differs between the two measurements with and without tracers. Consequently, subtracting the signals 401, 402 does not eliminate this contribution. Moreover, the variation in the speed also varies the propagation time of the acoustic signals in the inspection volume, and therefore the time it takes for the acoustic wave to reach the static obstacle 27 located in the fluid.The contribution of this obstacle 27 to the acoustic signal, which is represented by frame 320, will therefore be detected at two different times on the signals 401, 402 measured with and without tracers 30.

[0091] As illustrated in Figures 5A and 5B, according to the method of the present invention, two acoustic signals 501, 502 originating from the inspection volume 100 containing tracers are measured at two different times t1 and t2. In order to highlight the acoustic contribution of the tracers 30, these two signals 501, 502 are subtracted (E5). The residual acoustic signal 503 resulting from the subtraction E5 is illustrated in [Fig. 5C]. This result shows that the subtraction of the signals 501, 502 according to the present invention makes it possible to effectively eliminate the acoustic contributions of the static objects indicated by the frames 310 and 320 in Figures 5A and 5B, which mask the signal from the tracers, which is generally less intense. The residual acoustic signal 503 originates mainly, or even solely, from the tracers 30 present in the inspection volume 100.

[0092] Figure 6 illustrates an example of an image resulting from the subtraction of a plurality of reference signals acquired according to the conventional approach, from the volume of inspection volume 100 devoid of tracers 30, and of another plurality of signals from inspection volume 100 in the presence of tracers. Figure 7 illustrates an example of a residual Sr3, resulting from a subtraction E5 according to the present invention of two plurality of signals Sri, Sr2 acquired at two different times t1 and t2 of the inspection volume 100 including tracers 30 in each of the two images Sri, Sr2.

[0093] The experimental measurements in Figures 6 and 7 were carried out in a non-intrusive manner. For example, each of the source 40 and / or the sensor 50 can comprise a plurality of components based on a piezoelectric material. These piezoelectric components can be arranged side-by-side. A piezoelectric component consists of a material that transforms an electrical signal into an acoustic signal and vice versa. This will allow the generation of acoustic fields in the inspection volume and the measurement of their evolution. For example, the source 40 and the sensor 50 can be a single entity, namely a transducer that serves both to emit the incident acoustic signals Si, to receive the reflected acoustic signals Sr, and to measure the variations in acoustic pressure over time. Line AA indicates the position of the acoustic source 40 and the sensor 50, or the position of the transducer.

[0094] Line BB indicates the position of the first interface 60 between the wall 25 of the conduit 20 and the fluid 10. In these measurements, a sharp conical obstacle 27 is placed within the fluid 10. The second interface 60 between the obstacle 27 and the fluid is represented by a triangle CC, the apex of which represents the tip 27' of the obstacle 27. Axes 601 and 602 represent distances in millimeters (mm) along the horizontal and vertical directions. The color scale 603 represents the intensity of the acoustic signals Sr measured in decibels (dB). The color gradation in these images allows visualization of the static objects 25, 27 and the tracers 30, as well as the significance of their contributions. The wall 25 of the pipe according to this example is made of plexiglass and has a thickness of 3 mm, and the base of the obstacle 27 is placed 30 mm from the wall 25.

[0095] According to the classical approach, as shown in [Fig. 6], signal subtraction does not eliminate acoustic signals from static objects 25, 27, which are more intense than those from tracers 30. Thus, signals reflected by interfaces 60 and within wall 25 are still present in the image after subtraction and mask the tracers 30. Indeed, the number of tracers visible in the image is smaller than the number of tracers actually present in the inspection volume; consequently, a large amount of information about the fluid flow is lost.

[0096] Artifacts are also present in this image. These artifacts may arise from multiple reflections, for example, or from the appearance of interfaces 60 and the The 27' point is located at different points due to variations in acoustic impedance, temperature, or other conditions from one measurement to another. These signals from static objects make image processing difficult using PIV algorithms. This leads to poor reconstruction of the velocity field through image processing. A person skilled in the art could use conventional image processing methods such as contrast correction and noise filtering, or more advanced methods like machine learning algorithms to detect and remove noise or acoustic artifacts. However, these methods complicate image processing and do not allow for the recovery of all the information lost due to the presence of static objects.

[0097] In contrast, the present invention makes it possible to solve these problems, as demonstrated in [Fig. 7]. The residual Sr3 obtained by the present process essentially comprises signals from the tracers 30 present in the inspection volume 100. The spurious signals from static objects and the artifacts seen in [Fig. 6] are effectively eliminated in the residual Sr3 of [Fig. 7] by subtracting E5 two pluralities of Sri, Sr2 signals acquired under very similar measurement conditions and in the presence of the tracers 30 in the inspection zone. Such a residual Sr3 can be processed without complications, and a high-quality final image containing a maximum of information on the flow of the fluid 10 can be obtained.The invention makes it possible to eliminate not only the parasitic acoustic signals arising from the presence of tracers near interfaces, but also the parasitic acoustic signals arising from variations in the temperature of the inspected medium, the accumulation of free gases at interfaces, and non-homogeneous fluid densities. All these parameters contribute to the modification of the acoustic properties of the inspected medium and consequently to the production of parasitic signals.

[0098] The substeps of the E6 processing of the Sr3 residue according to an example, to obtain at least one final image of the inspection volume 100 are now described.

[0099] The E6 processing of the residual can be carried out in the time domain according to a method known as "Time-Domain Topological Energy" (TDTE). As illustrated in [Fig. 8], according to the TDTE method, the E6 processing can include a time-reversal E61 of the residual acoustic signals resulting from the subtraction E5. By way of example, the transducer 40, 50 can comprise 64 piezoelectric components. During the reception step E3, each piezoelectric component receives a reflected acoustic signal Sr, so 64 acoustic signals are recorded. For a signal Sn(t), with T being the final acquisition time, the time-reversed residual Sr3 will consist of 64 Sn(Tt) signals.

[0100] The E6 processing may then include a digital model E62 of the inspection volume 100 to form a virtual domain. This virtual domain may constitute a digital model representing the inspection volume without tracers, including a model of the transducer 40, 50, and each time-returned signal is emitted into this digital medium from the corresponding piezoelectric element.

[0101] Processing E6 includes backpropagation E63 of the residual acoustic signals returned in time to the virtual domain to generate an adjoint field. During this backpropagation E63, the residual acoustic signals are returned in time and fed back into the digital model of the inspection volume 100 in the absence of tracers. This step allows the residual acoustic signals to be focused on the position of the echo sources and thus the positions of the tracers 30 to be located.

[0102] On an indexed piezoelectric component n, the signal Sn(Tt) is generated. This allows the propagation process to be reversed. The signals emitted in this way will generate, in the virtual domain, an acoustic field consisting of several wavefronts that will focus on the position of each tracer 30 that was the source of an echo measured by the sensor 50. From this simulation, a time-varying acoustic pressure field over the entire inspection volume 100 can be exported. This data is known as the "adjoint field" and can be three-dimensional (3D): two spatial dimensions, since the acquired images and simulations are generally two-dimensional (2D), and one temporal dimension.

[0103] Processing E6 can then include generating a "direct field" E64 in the same virtual domain. For this purpose, in the same virtual domain, the same acoustic signal as that generated in the experimental medium can be generated, in the absence of tracers 30. For example, if a plane wave was used as the incident acoustic signal Si to excite the experimental inspection volume 100, then the same plane wave will be generated in the virtual domain. The direct field represents the export of the variation of acoustic pressure over time in this numerical model. This data has the same dimensions as the adjoint field.

[0104] The E6 processing can then include an E65 convolution of the adjoint and direct fields to form the final image. The synthesized acoustic field, i.e. the adjoint field, is convolved with the direct field, making it possible to retain only the instant of focus of the signals, and thus to obtain an image on which areas of high intensity are present at the locations where the tracers 30 are located.

[0105] An example of the numerical modeling is illustrated in [Fig. 9], in which the positions of the tracers 30 are located. Axes 701 and 702 represent the distance in millimeters along the horizontal and vertical directions respectively, and the color scale represents the acoustic pressure intensity expressed in pascals (Pa). Line AA represents the position of the modeled transducer 40, 50. In this example, a plane wave is propagated through the digital medium by the modeled transducer. The transducer then measures the echoes returned by the tracers 30 present in the medium.

[0106] According to another example, the E6 processing of the residual can be carried out in the frequency domain using a method known as Fast Topological Imaging (FTIM). According to this example, the E6 processing substeps are carried out in Fourier space using multiplications. The E6 processing also includes the digital modeling of the inspection volume 100 to form a virtual domain.It can also include generating a radiation pattern of the source 40 (frequency response over the entire inspection area of ​​the source excitation by a Dirac delta function) in the virtual domain. Next, the Fourier transforms of the residual acoustic signals and incident acoustic signals Si are calculated in this same virtual domain, followed by the multiplication of the Fourier transform of the residual acoustic signals by the source radiation pattern to generate the adjoint field. The E6 processing can further include multiplying the Fourier transform of the incident acoustic signals Si by the source radiation pattern to generate a direct field. Finally, the adjoint field is multiplied by the direct field to form the final image. One advantage of the E6 processing using the FTIM method is its processing speed compared to the TDTE method.

[0107] By following the steps of the process, a first final image without perturbations induced by reflections from static objects will be generated. The same process can then be carried out to obtain a second final image. Since the final images are close in time, it is therefore possible to visualize the overall displacement of the tracer cloud. By obtaining at least two consecutive final images, it is possible to apply PIV correlation algorithms to reconstruct a velocity field 200.

[0108] Figure 10 illustrates a system 1000 for performing the image-processing velocimetry method of the flow of fluid 10, described previously. In this example, the system 1000 may include a device 1200 for introducing tracers 30 into the fluid 10. It should be noted that if the fluid contains tracers, for example naturally and preferably in sufficient quantity, it is not necessary for the system 1000 to include a device 1200. The system also includes an acoustic source 40 configured to emit (E2) the incident acoustic signals Si onto the inspection volume 100. The system 1000 also includes a sensor 50 configured to receive the acoustic signals reflected Sr by the tracers 30 and the interfaces 60 between the fluid 10 and static objects such as the wall 25 of the pipe 20. As described previously, the source 40 and the sensor 50 can comprise juxtaposed piezoelectric components and can form a single entity as illustrated in [Fig. 10]. The system 1000 further includes at least one processor 1100 enabling, for example, the acquisition E4 of the Sri, Sr2 images, the subtraction E5 of the images, and the processing E6 of the residual.

[0109] This system 1000 is configured to perform steps E4 to E6 described above in a non-intrusive manner. In other words, in this system, the source and the sensor are advantageously located outside the pipe 20 as illustrated in [Fig. 10]. This system is also configured to allow the acquisition E4 of the pluralities of reflected signals Sri and Sr2 at different times t1 and t2 with an optimized temporal separation At as described previously, following the introduction El of the tracers 30 by the device 1200 into the fluid 10. In this way, both images Sri and Sr2 of the inspection volume include tracers 30 and are acquired under similar or even identical conditions to the experimental medium.

[0110] The system 1000 may further include a memory 1300. This memory 1300 is used to store the acquired images and data. The memory may be, for example, a hard drive. The system may also include electronic components connected to the processor 1100 and the transducer 40, 50. The processor 1100 controls its electronic components to acquire images at precise times by inputting the desired parameter At, and to transmit and receive acoustic signals.

[0111] A computer program product can be implemented on memory 1300. This program product includes instructions that can be executed by the processor 1100 to subtract the first and second plurality of signals Sri, Sr2 in order to eliminate the acoustic signals Sr from static objects such as the wall 25 or the obstacle 27. The program product also allows, by executing the instructions, the processing E6 of the residual to obtain the final image. The program product may further include instructions allowing, for example, the definition of image acquisition parameters, such as the parameter At, the intensities of the acoustic signals, the resolution of the acquired image, etc.

[0112] A PIV algorithm can also be implemented on memory 1300. The successive final images of the inspection volume, obtained through data processing by the program product, can feed the PIV algorithm in order to reconstruct a velocity field 200 as illustrated in [Fig. 10].

[0113] The tracers 30 can be, for example, microbubbles of gas formed by injecting a gas into the fluid using the device 1200. According to another example, the tracers can be solid microparticles, or microdroplets of a liquid introduced using device 1200 in fluid 10, this liquid being different from fluid 10. According to another example, these microbubbles or microparticles may be naturally present in the fluid. A microbubble, microparticle, or microdroplet may have a maximum cross-sectional area Sm less than or equal to 500 pm, and preferably greater than or equal to 5 pm.

[0114] Furthermore, the fluid may be a metal or a mixture of metals in liquid form. For example, the fluid may be liquid metallic sodium, which is used to cool certain so-called fourth-generation nuclear reactors. The fluid may also be a physiological fluid such as water or blood, or an oil-based fluid, saline solutions, gels, emulsions, coolants, chemical mixtures, etc.

[0115] The fluid can be opaque or transparent to visible light. One advantage of this method is the possibility of studying opaque industrial fluids, thanks to its non-intrusive nature, which is often necessary in industrial applications. This also applies to the nature of the pipe wall, which can be opaque, as is often the case in industrial applications. For example, the pipe 20 can be made of metal, plastic, or glass.

[0116] This invention is not limited to 2D velocity field measurement applications. It is applicable to acoustic imaging applications, as well as to 3D velocity field measurements. Furthermore, the invention allows the topological energy imaging method to be applied to all cases where variations in the acoustic properties of the acoustic propagation media are present.

[0117] The invention is not limited to particular operating frequencies, a particular number of piezoelectric components in the transducer, or particular transducer dimensions. Any type of inspection waveform can be used. Furthermore, the invention is not limited to a restricted inspection volume.

[0118] The invention is not limited to the embodiments described above. Various specific examples of the velocimetry method and the associated program product system have been described. Other embodiments are possible, for example, by combining features described above, without departing from the principle of the present invention. Furthermore, the features described with respect to one aspect of the invention can be combined with another aspect of the invention.

Claims

1. Demands A velocimetry method by image processing of a flow, inside a pipe (20), of a fluid (10) in which tracers (30) are present, the method comprising: • an emission (E2) by a source (40) of incident acoustic signals (Si) configured to propagate at least part of the acoustic signals into a volume of the pipe (20) called the inspection volume (100) in which the incident acoustic signals (Si) are reflected by the tracers (30) present in the inspection volume (100) and by at least one interface (60) between the fluid (10) and a medium, for example a wall of the pipe, having an acoustic impedance different from that of the fluid (10), to form reflected acoustic signals (Sr), • a reception (E3) by a sensor (50) of acoustic signals, called reflected signals (Sr), resulting from the reflection of incident acoustic signals (Si) within the inspection volume (100) so as to acquire (E4) a first plurality of reflected signals (Sri) and a second plurality of reflected signals (Sr2), • a subtraction (E5) using a processor (1100) of the first and second pluralities of reflected signals (Sri, Sr2) so as to eliminate the part of the reflected acoustic signals coming from at least one interface (60) and obtain residual acoustic signals forming a third plurality of signals called residue (Sr3), • a processing (E6) of the residue (Sr3) by the processor (1100) to obtain at least one final image of the inspection volume (100) in which each tracer (30) of the inspection volume (100) is identified at a given time, the method being characterized in that the source (40) is disposed outside the conduit (20) and in that the acquisition (E4) of the first and second plurality of signals (Sri, Sr2) is carried out such that the first plurality of signals (Sri) is acquired at a instant tl and the second plurality of signals (Sr2) is acquired at an instant t2 different from instant tl, and that the tracers are present in the inspection volume (100) at instants tl and t2.

2. A method according to the preceding claim, wherein the inspection volume (100) has a concentration Cl in tracers (30) at time t1, and the inspection volume (100) has a concentration C2 in tracers (30) at time t2, the times t1 and t2 being chosen so that a relative difference between Cl and C2 is less than or equal to 30%, preferably less than or equal to 10%.

3. A method according to any one of the preceding claims, wherein the acquisition (E4) of the first and second plurality of signals (Sri, Sr2) is carried out such that At=t2-tl is less than or equal to 0.5 L / vmax with L the smaller of the two dimensions of the inspection volume (100) reported in a plane and vmax the maximum speed of the tracers in the inspection volume (100).

4. Method according to the preceding claim, wherein At is greater than or equal to 0.25 d / v0 with v0 the minimum velocity of movement of the tracers in the inspection volume (100) and d the size of the tracers.

5. A method according to any one of the preceding claims, wherein the acquisition (E4) of the first and second plurality of signals (Sri, Sr2) is carried out such that At=t2-tl with At < 5000 ms and preferably At < 1000 ms.

6. A method according to any one of the preceding claims, wherein the acquisition (E4) of the first and second plurality of signals (Sri, Sr2) is carried out such that At=t2-tl with At > 0.5 ms and preferably At > 3 ms and preferably At > 1 s.

7. A process according to any one of the preceding claims, wherein the treatment (E6) of the residue (Sr3) comprises: • a time reversal (E61) of the residual acoustic signals, • a numerical modeling (E62) of the inspection volume (100) to form a virtual domain, • a backpropagation (E63) of the time-reversed residual acoustic signals in the virtual domain to generate an adjoint field, • a generation (E64) in the same virtual domain of a direct field by modeling the propagation of the incident acoustic signals (Si) in the tracer-free virtual domain (30), • a convolution (E65) of the adjoint field and the direct field to form the final image.

8. A method according to any one of claims 1 to 4, wherein the treatment (E6) of the residue (Sr3) comprises: • a numerical modeling of the inspection volume (100) to form a virtual domain, • a generation of a radiation pattern of the source (40) in the virtual domain, • a calculation of the Fourier transforms of the residual acoustic signals and incident acoustic signals (Si) in the virtual domain, • a multiplication of the Fourier transform of the residual acoustic signals by the radiation pattern of the source to generate an adjoint field, • a multiplication of the Fourier transform of the incident acoustic signals (Si) by the radiation pattern of the source to generate a direct field, • a multiplication of the adjoint field by the direct field to form the final image.

9. A method according to any one of the preceding claims, wherein the fluid is opaque to visible light.

10. A method according to any one of the preceding claims, wherein the conduit (20) has a wall separating the inside from the outside of the conduit (20), the wall being opaque to visible light.

11. A method according to any one of the preceding claims, wherein the tracers (30) are microbubbles formed by injecting a gas into the fluid.

12. A method according to any one of the preceding claims, wherein the fluid (10) is a metal or a mixture of metals in liquid phase.

13. A method according to the preceding claim, wherein the fluid (10) is liquid metallic sodium.

14. A method according to any one of the preceding claims, wherein the emission (E2) of incident acoustic signals (Si) is carried out by the source (40) which comprises a plurality of components based on a piezoelectric material.

15. A method according to any one of the preceding claims, wherein the reception (E3) of reflected acoustic signals (Sr) is achieved by the sensor (60) which comprises a plurality of components based on a piezoelectric material.

16. A method according to any one of the preceding claims, wherein the emission (E2) of incident acoustic signals (Si) and the reception (E3) of reflected acoustic signals (Sr) are carried out by the same transducer which is both the source (40) and the sensor (60).

17. A method according to any one of the preceding claims, comprising obtaining several successive final images (Sr3) of the fluid flow (10) over time and reconstructing a velocity field (200) of the tracers (30) using a PIV algorithm.

18. A method according to any one of the preceding claims, wherein the conduit (20) is based on a metal, plastic or glass.

19. A method according to any one of the preceding claims, comprising, prior to acquisition (E4), an introduction (E1) of the tracers (30) into the fluid (10).

20. Product computer program comprising instructions which, when executed by a digital data processing device, for example a processor (1100), perform at least the following steps of claim 1:

21. • a subtraction (E5) using the first and second plurality of reflected signals (Sri, Sr2) so as to eliminate the part of the reflected acoustic signals (Sr) coming from at least one interface (60) and to obtain residual acoustic signals forming a third plurality of signals called residual (Sr3), the first plurality of reflected signals (Sri) of the inspection volume (100) being acquired at a time t1 and the second plurality of reflected signals (Sr2) of the inspection volume (100) being acquired at a time t2 different from time t1, • a treatment (E6) of the residue (Sr3) to obtain at least one final image of the inspection volume (100) in which each tracer (30) of the inspection volume (100) is identified at a given time. System 1000 for velocimetry by image processing of a fluid flow (10) in a pipe (20), the system comprising: • a source (40) of acoustic signals configured to emit (E2) incident acoustic signals (Si) on a volume of the pipe (20) called the inspection volume (100), and configured such that in the presence of fluid in the pipe and tracers (30) in the fluid the incident acoustic signals (Si) are reflected by the tracers (30) present in the inspection volume (100) and by at least one interface (60) between the fluid (10) and a medium having an acoustic impedance different from that of the fluid (10), to form reflected acoustic signals (Sr), • at least one sensor (50) configured to receive so-called reflected acoustic signals (Sr), resulting from the reflection of incident acoustic signals (Si) within the inspection volume (100), • at least one processor (1100), the system being configured to perform the following steps: • an acquisition (E4) by the sensor (50) of a first plurality of reflected signals (Sri) and a second plurality of reflected signals (Sr2) of the inspection volume (100), • a subtraction (E5) using the processor (1100) of the first and second pluralities of reflected signals (Sri, Sr2) so as to eliminate the part of the reflected acoustic signals (Sr) coming from at least one interface (60) and obtain residual acoustic signals forming a third plurality of signals called residue (Sr3), • a processing (E6) of the residue (Sr3) by the processor (1100) to obtain at least one final image of the inspection volume (100) in which each tracer (30) of the inspection volume (100) is identified at a given time, the system being characterized in that the source (40) is disposed outside the conduit (20) and in that it is configured so that the first plurality of reflected signals (Sri) is acquired at a time tl and the second plurality of reflected signals (Sr2) is acquired at a time t2 different from the time tl.

Images