Patent Application
20230168082
The present invention relates to a system for generating a signal representative of the profile of a surface moving relative to the system.
Knowing the profile of a surface of an object is useful for multiple types of application. First of all, when the object is moving relative to a second object, it is necessary to know this profile to ensure the relative mobility between the two objects. Indeed, it is possible to deduce, from the profile of the surface, the roughness thereof, making it possible to adapt for example the relative speed between the two surfaces so as to ensure a condition of adhesion. Second of all, when inspecting the quality of an object, the profilometry of the surface is a parameter subjected to surveillance since it is a sign of evolution of the manufacturing process, for example. Owing to manufacturing defects but also to ageing or wear in service of the surface of an object or to varying external conditions, it is useful to be able to have regular access to profile information. For example, in the field of land transport, knowing the profile of the ground at a millimetre scale is important for adapting active safety systems of the vehicle to the conditions of ground adhesion, which is dependent on the surface roughness of the ground. Similarly, in the field of object manufacturing, it is useful to measure the profile of the outer surface of manufactured objects to ensure compliance with specifications and to adapt the manufacturing process according to this parameter.
Thus, real-time knowledge of the profile of any surface of an object is important. However, there is a desire for profile-measuring devices that are physically non-intrusive with respect to the object to be evaluated or with respect to the device on which it is likely to be mounted. In addition, it should have little impact on the current operation of the devices so as not to affect the efficiency of these devices. Finally, it should be potentially economical both in terms of purchase price and in terms of use, and consume little energy.
Among devices for measuring the roughness of the ground, the rugolaser makes it possible to measure profiles dynamically. The measuring principle is based on the use of a pulsed laser source emitting vertically in the direction of the outer surface to be measured. The laser source is coupled to focusing optics and a CCD optical potentiometer or photoreceptor array. Depending on the laser/target distance, the optics focus the image of the point of impact of the laser beam at a given position on the sensor. By locating this position on the potentiometer, it is possible to arrive at the height of the profile of the target.
The drawback of such a device lies in the bulk embedded on the mobile device, its purchase cost and its low measurement flexibility in comparison with the use of the mobile device, since the system settings and the analysis of the measurements require time and post-processing is generally performed after the measurement. It is therefore not possible to obtain the information in real time.
Among devices for measuring surface roughness in terms of quality control for a manufactured object, mention may be made of imaging and lighting devices. However, this type of equipment is not suitable for live quality control since the processing time consumes flow time and the object is generally static or quasi-static. Although it makes it possible to access a three-dimensional observation of the outer surface of the object, this processing being well suited to a quality logic based on random sorting of objects, this is not the case for live quality control of all objects.
The present invention relates to a device for generating signals that are able to be used in real time and representative of the profile of a surface on a two-dimensional plane, simultaneously solving the problems encountered by devices from the prior art in terms of response time, embedded on devices while at the same time being non-intrusive and not having any impact on the operation of the object itself. Finally, some variants of the invention are also energy-efficient and inexpensive.
The invention relates to a system for generating at least one signal representative of the profile of an outer surface of a medium having a median plane and having a relative speed V with respect to the system in a direction U, in use thereof with an outer surface of a medium having a median plane, comprising:
The invention relates to a system for generating at least one signal representative of the profile of an outer surface of a medium having a median plane and having a relative speed V with respect to the system in a direction U, comprising:
Imposing that the optical paths are coplanar ensures that, during use thereof, the device will impose that the optical paths follow the same readout line on the outer surface. Imposing that the direction vectors of the optical paths define an angle θ greater than or equal to 3 degrees ensures a differentiation between the Doppler frequencies of each beam of light.
This device makes it possible to generate a first electrical signal at the output of the sensor that translates the effects of electromagnetic interference generated between the first incident beam of light from the first source and the beam of light backscattered by the outer surface resulting from the first incident beam. Here, the information relating to a variation in the emitting power of the first light source is sufficient. The term “backscattered beam” is understood here to mean that this corresponds to the incident beam that is reflected and/or scattered by the outer surface and that follows the same optical path as the incident beam in the opposite direction. These effects are also modulated by the distance between the geometric point d1 where a first portion of the Gaussian beam of light is focused and the outer surface of the medium due to the Gaussian propagation of the beam of light. The two beams interfere due to the spatial and temporal coherence between the incident beam and the backscattered beam. The observed variations represent a succession of phenomena driven by the harmonic frequencies related to the Doppler effect. The fundamental frequency of the Doppler effect depends simultaneously on the relative speed V between the system and the outer surface, on the angle of incidence θ of the beam on the outer surface with respect to the normal to the outer surface along the direction U and on the wavelength λ of the beam of light. It is thus necessary for the incident wave to generate an angle of at least 1 degree with respect to the normal to the outer surface along the direction U in order for the Doppler effect to be able to be observed in the signal. In practice, an angle of 5 degrees makes it possible to ensure observation of the Doppler effect beyond the geometric imperfections of any imperfect optical system and to obtain an easily usable signal. Taking a monochromatic light wave avoids interference between different wavelengths of a polychromatic light, making the harmonics of the monochromatic wavelength easily visible in the signal from the sensor. The same operation is performed with a second beam of monochromatic light the geometric position d2 of which with respect to the outer surface along the readout line described by the first optical path is different from the first geometric position d1. The two geometric positions d1 and d2 necessarily surround the median plane of the outer surface of the medium under observation in order to easily identify the distance from the outer surface. Two items of information proportional to the distance between a given geometric point d1 or d2 and the outer surface are thus retrieved. Of course, it is preferable for the distance between these two geometric points d1 and d2 to be greater than the greatest Rayleigh length of the Gaussian beam in order to obtain a high-quality signal with a spatial resolution adapted to what is sought in terms of discretization of the outer surface. Indeed, at the Rayleigh length, the shape of a beam of light is modified by increasing the radius of the beam by a factor of root 2. This modification of the beam will have a direct impact on electromagnetic interference between the emitted and backscattered beams and will therefore quantify the distance to the focal position of each of the first and second beams of light. If the two focal lengths differ by a distance less than the greatest Rayleigh length, the beams of light do not diverge enough for the information about the distance between the focal point and the outer surface to be perceived in the output signal from the sensor.
The distance information provided by the second beam of light is coded in the same first electrical signal, in which case the harmonics due to the Doppler effect should be dissociable between the two beams of light. This condition is complied with by imposing an angle θ of at least 3 degrees between the two optical paths, which will shift the frequency of the two fundamentals related to the Doppler effect, in the knowledge that, during use thereof with the outer surface, the device is arranged such that the optical paths are substantially perpendicular to the median plane. Importantly, the impact of the reflectivity of the outer surface should be similar between the two beams of light, which means either that they have the same wavelength or that their wavelength, although different, is insensitive to the optical reflectivity characteristics of the outer surface of the medium under observation.
By having information relating to two distances of the outer surface from two references d1 and d2 along the same readout line, it is possible to arrive, through signal processing, at a two-dimensional profile of the outer surface along the readout line regardless of the relative speed V of the system with respect to the medium. The spatial discretization of the profile is proportional to the sampling frequency of the signal and to the relative speed V. By adapting the sampling frequency as a function of the speed V, it is possible to obtain the desired spatial precision of the discretization of the readout line. The measurement is thus not impacted by the usage conditions between the medium under observation and the device receiving the system, and allows measurement at high relative speed V in the direction U.
The system preferably also comprises an electronic-type electrical signal amplifier when the amplitudes of the signals delivered by the sensors are low, in particular due to the width of the frequency band of the signal, it is necessary to amplify these signals without losing information.
Finally, the system does not require any complex adjustment since no precise orientation constraint on the beams of light with respect to the outer surface is required; only alignment in the direction U of the two incident beams is necessary, which is satisfied when the optical paths of the first and second beams of light are coplanar. The coplanarity of the optical paths ensures that the two optical paths follow the same readout line on any outer surface.
The light source may be for example a laser source, such as a laser diode for example. These laser-type sources generate coherent light, and the propagation of light is also naturally Gaussian. The sensor has to measure electromagnetic interference between the incident light and the light backscattered by the outer surface. This involves identifying a spatial area where the two beams are aligned, or at least a portion of them. The simplest technique is to place the sensor on the optical path between the light source and the surface. However, it is entirely possible to deflect a portion of one or both of these beams to make them coincide outside the optical path. The sensor may be for example a photodiode, a phototransistor or a current or voltage sensor for the power supply of the light source.
The first optical device and/or the means for routing the second beam of light includes for example means for orienting at least part of the system with respect to the outer surface so as to at least partially orient the angles of incidence of the first and/or second beam of light with respect to the outer surface.
The term “Gaussian beam” is understood to mean that the propagation of light in the direction of propagation of the beam is Gaussian.
The term “optical path” is understood to mean the succession of contiguous spatial positions followed by the beam of light between a light source or a means for generating a beam of light and the outer surface of the medium under observation.
The term “optical path” is understood to mean at least a portion of an optical path between the last focusing lens focusing the beam of light before the outer surface and the farthest point between the outer surface and the focal point of the beam of light.
According to a first embodiment of the system, the at least one first focusing lens being located upstream of the first optical device, the first optical device and the means for generating at least one second beam are pooled and comprises an optical splitter element located on the first optical path of the first beam of light, the at least one second beam of light is the other portion of the first beam of light, the means for routing at least one second beam of light comprises at least one mirror and the projections of the angles of incidence θ1 and θ2 onto said outer surface of the first and second optical paths are different. Regardless of the embodiment, the angle θ formed by the difference between the two angles of incidence θ1 and θ2 should be at least 3 degrees, thereby ensuring the ability to dissociate the fundamentals and potentially the harmonics in the single electrical signal.
The case in which the first light source is single is dealt with here. It is then necessary to split the light energy of the first beam of light into two beams of similar energy. For this purpose, a simple optical splitter element is sufficient for this function, which makes it possible to generate a second Gaussian beam of coherent and monochromatic light at the same wavelength λ1. Since the first focusing lens is located upstream of the optical splitter element, the two beams of light have the same focal length f1. In order to obtain two items of distance information, it is necessary to differentiate the length of the two optical paths between the focusing lens and the outer surface by a length d1-d2, such that the two geometric points are located on either side of the outer surface during use thereof with an outer surface. This is at least ensured by way of a set of mirrors that redirects the second beam of light towards the outer surface and generates a modification of the optical path of the light on the second optical path with respect to the first optical path. Finally, since the distance information is coded on the same electrical signal and the wavelength is unique, the angles of incidence of the two paths are different so as to dissociate, in the response spectrum of the electrical signal, the harmonics related to the Doppler effect corresponding to the first and second beams of light.
According to another preferred embodiment of the system, the at least one first focusing lens being located downstream of the first optical device on the first optical path, the first optical device and the means for generating at least one second beam are pooled and comprises an optical splitter element located on the first optical path of the first beam of light, the at least one second beam of light is the other portion of the first beam of light, the means for generating at least one second laser beam comprises at least one second focusing lens with a focusing distance f2 located downstream of the first optical device and the projections of the angles of incidence θ1 and θ2 onto said outer surface of the first and second optical paths are different. The angle θ formed by the difference between the two angles of incidence θ1 and θ2 should be at least 3 degrees, thereby ensuring the ability to dissociate the fundamentals and potentially the Doppler harmonics of each beam in the single electrical signal.
The case in which the first light source is single is again dealt with here. It is then necessary to split the light energy of the first beam of light into two beams of similar energy. For this purpose, a simple optical splitter element is sufficient for this function, which makes it possible to generate a second beam of light at the same wavelength λ1. If the first focusing lens is located downstream of the splitter element, that is to say the first beam of light is not focused before the optical splitter element, it is then necessary, having split the first beam of light in two to generate a second beam of light, to focus the second beam of light using a second focusing lens. The focusing distance f2 of this second focusing lens is possibly different from the first. If it is similar, it is necessary to differentiate the length of the first and second optical paths by modifying the optical path of the two beams so as to define two different geometric points d1 and d2. Finally, since the distance information is coded on the same electrical signal and the wavelength is unique, the angles of incidence of the two paths are different so as to dissociate, in the response spectrum of the electrical signal, the harmonics corresponding to the first and second beams of light.
Preferably, in the case of using the device with a medium to be studied having an outer surface having a median plane, the ratio between the projections in the plane defined by the direction U and the normal to the median plane of the angles of incidence θ1 and θ2 onto said outer surface of the first and second optical paths is between 1.2 and 1.8.
The angles of incidence of the first and second optical paths on the surface should be different in order to dissociate the information carried by each path when the information is coded on a single electrical signal. Moreover, if it is sought to follow multiple harmonics of each beam of light, it is necessary for the ratio of the frequencies generated by the angles, which is generally small, not to be an integer multiple of one another. Finally, in order for the analysis of the signals not to be too wideband, it is necessary for the frequency spacing between the harmonics generated by the angles of incidence to be sufficient to dissociate the frequencies without excessively fine discretization of the single signal. All of these constraints lead to the claimed range.
Preferably, in the case of using the device with a medium to be studied having an outer surface having a median plane, the first and second optical paths do not intersect before having reached the outer surface of the medium.
If a single light source is used, the first and second beams of light are mutually coherent. As a result, these first and second beams of light are likely to interfere electromagnetically with one another, be this on the incident or backscattered paths. This interference may alter the quality of the signal measured by the sensor and therefore lead to an error in the quality of the signals from the sensor. Ensuring the above condition minimizes the risk of creating parasitic electromagnetic interference between the first and second beams of light, thereby improving the quality of the output signals from the sensor and consequently the measurement of the profile of the outer surface of the medium. Finally, the improvement of the measured signal by this condition is potentially ensured on both optical paths.
Preferably, in the case of using the device with a medium to be studied having an outer surface having a median plane, the coherence length of the first laser beam and/or of the second laser beam is at least greater than twice the greatest length of the first and second incident optical paths to the outer surface of the medium.
The payload information is obtained by observing electromagnetic interference between the incident beam of light and the beam backscattered from the outer surface of the medium of the incident beam of light. The sensor should necessarily be located in a geographical area where the incident and backscattered beams are mutually coherent so as to interfere. Although the sensor may be remote from the incident optical path, it is often more convenient to position the sensor on the incident optical path. Since the backscattered beam has to pass through the incident optical path at least once in order to be generated, the mentioned length condition ensures the ability to position the sensor anywhere on the longest optical path.
Preferably, in the case of using the device with a medium to be studied having an outer surface having a median plane, the angles of incidence θ1 of the first beam and θ2 of the second beam of light are contained within a cone the axis of revolution of which is the normal to the median plane of the outer surface and the aperture angle of the cone is less than or equal to 45 degrees, preferably less than or equal to 30 degrees.
Regardless of the embodiment, it is desired to observe the profile of the outer surface of a medium. Ideally, the angle of incidence is along the normal to the median plane, thereby making it possible to observe any convex surface profile. In fact, the system imposes at least a certain inclination in the plane of relative movement of the system with respect to the outer surface so as to observe Doppler effects. This low inclination has little impact on the observation of a convex surface. However, if the inclination is greater and regardless of the orientation of the beam of light, some areas of the convex outer surface cannot be observed, and these will be areas masked by areas located above the masked area of the outer surface. Imposing a maximum inclination of 45° minimizes these masked areas of the readout surface. Preferably, the aperture angle of the cone should be less than 30 degrees, thereby making it possible to halve the masked surface while still allowing differentiated angles of inclination that are acceptable even in the case of a single first sensor.
Very preferably, the first sensor is contained within the group comprising a phototransistor, a photodiode, an ammeter and a voltmeter.
Multiple sensor technologies may be used to globally evaluate the temporal variations in the electromagnetic interference between the monochromatic coherent waves of the incident beam and the backscattered beam, such as for example those that evaluate light power. If the power supply for the light source is not fixed and the interference is generated as far as the light source, variations in the consumption of the source, for example a laser, may appear on the power supply signal. A high-resolution ammeter or voltmeter thus makes it possible to observe small electrical variations at the point of electric power supply to the light source. This will work well for specific electrical devices and when the monochromatic sources are distinct.
More conventionally, the system may be equipped as a sensor of a photodiode or a phototransistor. These types of light sensor will be capable of observing the temporal variations in monochromatic coherent light that are generated by interference, independently of the power supply circuit for the light source.
Moreover, these are standard elements of the packaging of laser diodes, thereby making these devices inexpensive and easy to implement. Finally, since it is desirable for a wave backscattered at the same wavelength to align with the wave of the incident beam, conventional laser diodes, due to their dedicated optical cavity for amplifying light, perform this function if the optical interface allows transmission of the reflected wave. This is the case with standard laser diodes, which also have a low cost price. Indeed, in sophisticated laser diodes, it is sought more to minimize or even prevent the reflected waves from penetrating into the optical cavity of the diode, requiring optical surface treatments at the external optical interfaces.
Advantageously, the wavelength of the first and of the second beams of light is between 200 and 2000 nanometres, preferably between 400 and 1600 nanometres.
The device according to the invention consists in focusing the first and second beams of light towards the outer surface of the medium to be observed. Due to the very nature of optical systems, what is known as an Airy spot of a certain dimension is generally obtained at the focusing distance due to the physical phenomena of light diffraction. However, the dimension of this concentric spot is directly proportional to the wavelength of the beam of light. Using wavelengths of visible light in general gives reasonable spot dimensions that allow a spatial resolution of the outer surface along the readout line that is suitable for the desired applications. Of course, the smaller the wavelength, the blue or the violet or the ultraviolet, the smaller the dimension of the Airy spot, thus increasing the resolution of the readout line, the thickness of which, along the direction perpendicular to the readout line, is also proportional to the wavelength. On the other hand, operating in the red or the infrareds generates metric precision that is lower but that may remain suitable depending on the application and depending on the desired resolution criteria.
Specifically, the first light source is a laser diode.
As already mentioned, this is by nature a source of monochromatic, coherent light for which the propagation of the beam of light is Gaussian when the laser diode operates at its threshold. In addition, the diode comprises an optical cavity, which may be the ideal location for observing electromagnetic interference between the incident beam and the backscattered beam if the external optical interface allows transmission of backscattered waves. At this time, with the optical cavity serving as an amplifying medium for the generation of light, any light power sensor will deliver a signal easily able to be used by the principle of self-mixing, which is another term for optical feedback. Finally, the packaging of a laser diode is compact and inexpensive. As a result, it is an economical solution that meets the need perfectly.
The invention also relates to a static or mobile device equipped with a system for generating at least one signal representative of the profile of an outer surface of a medium.
Indeed, the system covers only the essential aspects of the measurement. The system may therefore be installed on a device that allows the system to be integrated into the desired operating environment. This device may thus be mobile, such as for example a land vehicle moving relative to the ground. The ground is then the observation medium for the measurement system. The device may also be static, such as an industrial station for inspecting a linear appearance in relation to the scrolling of objects on a conveyor belt. The system installed on a mobile or static device may also observe the rotational movements of a medium relative to the device in order to analyse uniformity defects of the outer surface of this medium.
The invention also relates to a method for obtaining the profile of the outer surface of a medium, comprising the following steps:
Since the information is coded on the same electrical signal at the output of the sensor, it is necessary to extract, from this signal, the response of the fundamental, possibly those of the harmonics of the Doppler frequency, of each beam of light.
Here, it is first necessary to have two time signals corresponding, at all times, to the response of the same geometric point of the readout line on the outer surface of the medium under observation. The distance information is carried by the fundamental and the harmonics of the Doppler frequency. Consequently, it is necessary to determine the Doppler frequency on each signal. This depends on the relative speed V of movement in the direction U between the system and the medium under observation. However, it also depends on the angle of incidence θ with respect to the normal to the median plane of the outer surface. Finally, the Doppler frequency is also a function of the wavelength λ of the beam of light. Knowing all of these parameters, it is theoretically possible to determine the Doppler frequency, its fundamental. Another solution consists in frequency-analysing the time signal in order to determine the frequency and its harmonics, which should also emerge from the frequency analysis of the signal.
It is thus important that the sampling frequency of the time signals is at least greater than twice the Doppler frequency so that the payload signal carries information that is definitely reliable (signal processing—Shannon's theorem) on the fundamental of the Doppler frequency. However, depending on the application, the information may also be carried by the first harmonics of the Doppler effect; it is then preferable to perform enough sampling to have reliable information on the successive harmonics.
Next, to extract the distance information on the payload signal, it is first necessary to extract the envelope of the payload signal, which represents the extreme temporal variations of the recorded electromagnetic interference. This envelope may be constructed on the minimum value of the payload signal or on the maximum value of the payload signal. As a variant, it is also possible to take the maximum value of the absolute value of the envelope, which generally oscillates around the zero value. Here, it is the general information that carries the payload information, which justifies taking the envelope of the signal.
Finally, the last step is determining the profile along the readout line of the outer surface. For this purpose, a bijective function F is created, this being a relative mathematical combination of the envelopes of the signals resulting from the two optical paths. The advantage of the relative mathematical combination is that a calibration step may be performed a priori using a target representative of the nature of the media that it is desired to measure. The calibration then does not require the use of conditions similar to the desired measurement, but only requires ensuring the proportionality of the responses between the two signals. The result of this combination gives a quantity that, through a monotonic and bijective function F, translates one and only one distance E relative to a reference point through the step of calibrating the function F, despite this calibration not having not performed on the measurement medium. Real-time measurement of the profile of the outer surface of the medium is thus ensured. The variations in the distance E between the points of the readout line make it possible to automatically generate the profile of this same readout line, thereby allowing real-time processing of the profile, since this requires few computing resources.
According to one particular embodiment, the step of obtaining two time signals A and B for the same geometric target of the readout line comprises the following steps:
The system for generating signals representative of the profile of the outer surface does not necessarily phase the first and second optical paths. In this case, this preparatory step for the signals is essential for obtaining two time signals from the same point of the readout line of the outer surface in a robust and reliable manner.
Advantageously, the function F is calibrated using at least one white and rough target, the surface roughness of which is greater than the wavelength of the light of the first and second beams of light.
The use of the function F, a relative mathematical combination of the envelopes of the signals A and B, requires a calibration phase for calibrating this function F. This calibration may be performed using a specific target that is moved metrologically relative to the signal generation system such that the outer surface of the target remains between the geometric points d1 and d2 of the system for generating the representative signals. This target should advantageously be white and rough. The term “white” is understood to mean here that the outer surface of the target should backscatter more light, at the wavelength of the generation system, than it absorbs. And the amount that is backscattered should also advantageously be at least equal to or above the level of light backscattered by the outer surface of the medium that it is desired to measure, thereby guaranteeing the proportionality of the response regardless of the medium to be observed. It is also necessary for this target to have a rough outer surface in order to backscatter the light, and not just reflect it as an optical mirror would. Finally, the surface roughness of the target should be greater than the wavelength of the light of the first and second beams of light. Indeed, if the surface roughnesses are not greater than the wavelength, the surface will behave like a mirror at the wavelength in question, and therefore minimize backscattering. Moreover, the speckle phenomenon, that is to say the phenomenon of electromagnetic and in particular optical speckle, will not be present if the roughness condition is not complied with, and it has to be present in order to be able to optionally characterize speckle noise, that is to say noise generated by the phenomenon of electromagnetic speckle, while still being comparable with that of the targets that it is intended to measure.
Preferably, the step of generating the payload signal comprises the following step:
Very preferably, the step of generating the payload signal uses frequency windowing between 0.7 and 1.3 times the Doppler frequency.
It is possible, although not necessary, to filter the time signal around the Doppler frequency that carries the payload information in order to isolate a payload time signal that highlights electromagnetic interference. The uncertainty on the exact Doppler frequency naturally leads to selective filtering being carried out around the determined Doppler frequency. The potential uncertainty on the determination of the Doppler frequency leads to the signal being filtered over a width linked to the Doppler frequency, and the windowing between 0.7 and 1.3 times the determined Doppler frequency makes it possible both to cover the uncertainty on the Doppler frequency while focusing on the only fundamental, which generally carries sufficient information.
Preferably, the step of determining the Doppler frequency is performed through:
These three methods make it possible to determine a Doppler frequency of the time signal. The second method consists simply in theoretically evaluating the Doppler frequency, knowing the technical characteristics of the generation system. The first method performs a Fourier transform of the time signal in order to extract the fundamental frequency that emerges from the frequency spectrum. If the number of samples of the time signal is a multiple of 2, a fast Fourier transform may be performed, thereby allowing accelerated processing of the function. In the third method, it is necessary to analyse a temporal sample of the signal in order to detect the fringes and in particular the spacing between consecutive fringes in order to deduce the Doppler frequency therefrom. Of course, it is possible to use several of these methods jointly to converge rapidly on the Doppler frequency of the signal. All of these methods may be performed on board the device where the generation system is installed, minimizing computing or memory resources.
Preferably, the step of determining the envelope of the payload signal is performed on the absolute value of the payload signal.
The applicant has observed that determining the envelope on the payload signal when this corresponds to the absolute value of the payload signal provides an increase in robustness for the method for determining the profile of the outer surface. Indeed, the payload signal oscillates around the zero value, and taking the absolute value eliminates interference related to the phase positions of the two payload signals, thereby improving the prediction of the distance d from the outer surface of the medium.
Very preferably, the step of determining the envelope of each payload signal comprises a step of cleaning speckle noise on the determined envelope.
Indeed, it is often preferable to eliminate speckle effects from the determined envelope, that is to say the effects produced by speckle noise, generated by the Doppler effect and the parasitic signals from the backscattering of light. For this purpose, specific filters should be used, such as Lee Sigma or gamma Map filters, which are highly effective for cleaning the selected signature. This cleaning improves the prediction of the profile of the outer surface of the medium by minimizing noise on the payload signal. Cleaning on the envelope makes it possible to statically attenuate measurement noise while determining the parameters of this cleaning a priori, thereby allowing real-time and on-board use of the profile of the outer surface.
Preferably, the step of cleaning speckle noise on the envelope of the signal comprises the following steps:
In order to allow real-time use of the method for obtaining the profile of the outer surface, simple operations should be performed on the time signals resulting from the measurement. After having defined the filtering method used, which depends both on the nature of the outer surface of the medium to be measured and on the measurement system used, the envelope is divided into a multitude of windows the size of which is adapted to the speckle noise, with or without the windows overlapping with one another depending on the filtering method used. For each window that corresponds to the use of a limited memory space, an average of the values of the envelope of the window is computed. These values are potentially weighted in the event of overlap between contiguous windows. The signal reconstituted by the characteristic quantities of each window forms the envelope cleaned of speckle noise.
Very preferably, the filtering method for the step of cleaning the speckle noise is contained within the group comprising GammaMAP and Sigma.
These are filtering methods for which the effectiveness in terms of characterizing roughnesses of a millimetric outer surface is sufficient by using the signal generation systems disclosed in the invention.
Advantageously, the temporal size of the filtering windows and the level of overlap between contiguous windows are determined during a step of calibrating the speckle noise, comprising the following steps:
Indeed, the amplitude of the envelope is at the same time the combination of the reflectivity of the outer surface of the medium at the point of impact of the beams of light, of the distance between the point of impact on the outer surface and the focal point of the beam of light and speckle noise. This speckle noise is related both to the angles of incidence of the beams of light and the distance between the outer surface and the two geometric points d1 and d2. As a result, the speckle noise is related directly to the layout of the signal generation system. The mathematical model of the speckle noise may be the product of a Gaussian frequency distribution and a noise uniformly distributed between 0 and 1. This uniformly distributed noise is statically random, and it is therefore just necessary to identify the correct Gaussian distribution of the frequencies of the speckle noise by calibrating the generation system. Since the Gaussian distribution is of a certain frequency width, it is necessary to take a sufficient time window so as not to amputate the cleaning operation with an error linked to the time/frequency transformation of the signals. This is tantamount to averaging the time signal of the determined envelope over a time long enough to be statically representative of the speckle noise related to the generation system. The first phase is that of quantifying the speckle noise on the responses of the signals from the generation system. For this purpose, a single light path may be analysed on a single position of the target. However, it is preferable to increase the number of positions of the target and to analyse the various signals from the generation system. After having obtained a multitude of envelopes, a Gaussian distribution of these envelopes is determined through a technique of averaging the various distributions obtained. The second step consists in creating a multitude of noisy profiles from a theoretical profile by generating a multitude of speckle noise associated with the generation system. Finally, the optimum time windowing size and the level of overlap between the contiguous windows are determined using statistical analysis on all the noisy profiles in comparison with a known theoretical profile by choosing the parameters that minimize the differences on the entire population of noisy profiles.
Advantageously, the step of combining the cleaned envelopes comprises the difference between the cleaned envelopes expressed on a logarithmic scale.
The applicant has observed that this formatting of the mathematical combination of the cleaned envelopes made it possible to improve the sensitivity of the function F, thereby improving the precision on the evaluation of the distance d from the outer surface of the medium. Due to the Gaussian propagation of the beams of light and therefore the backscattered power, the logarithmic scale makes it possible to linearize the backscattered light power as a function of the distance from the focal point of the beam of light. The function F is linear in a manner more independent of the position of the geometric points and of the focusing of the beams of light at the geometric points.
The invention will be better understood on reading the following description in the case of an application involving a fixed device and moving medium under observation. This application is given solely by way of example and with reference to the appended figures, in which:
To implement the invention, it is first necessary to define an optical system that makes it possible to generate two beams of light the focal point of which is located on either side of the outer surface that it is desired to observe.
In this
The first optical path 11′ resulting from the first beam of light having passed through the first focusing lens 5 is then divided into 2 through a splitter cube 4, for example 50/50. The light power is thus split into two equal portions, and each half-power is directed in two different directions. The splitter cube 4 is the means for generating a second beam of light from the first beam of light along a second optical path 12, which is identical to the second optical path 12′. It therefore retains its characteristics of the first beam of light. The second beam of light is thus Gaussian, coherent and monochromatic at the same wavelength as the first beam of light. Moreover, it is generated from a focused beam of light, and the second optical path 12′ is thus created. The mirror 7 here acts as a means for routing along the second optical path 12′. Its first role is to direct the second beam of light towards the outer surface 22 at an angle of incidence the angular projection of which with respect to the normal to the median plane 23 in the plane defined by the normal to the median plane 23 and the direction U is equal to θ2, which is preferably less than 45 degrees while necessarily being non-zero. The second role of the mirror 7 is to define the position of the geometric point d2 such that this point is different from the geometric point d1 where the first optical path 11′ is focused. Therefore, the geometric point d2 corresponding to the focusing distance f2 of the second beam of light is located above the median plane 23 of the outer surface 22 of the medium 21. This second beam of light impacts the outer surface 22 at the point of impact 14 at an angle of incidence the projection of which, in the plane, is equal to θ2. The angles of incidence θ1 and θ2t are greater than one degree. Thus, at this incidence, the backscattered beam is subject to the Doppler effect caused by the speed of movement V between the device comprising the generation system 1 and the medium 21 in the direction U.
The other remaining half-power follows the first optical path 11′ and impacts the outer surface 22 at the point of impact 13 at an angle of incidence the projection of which with respect to the normal to the median plane 23 in the plane defined previously is equal to θ1, which is greater than 1 degree. For this portion of the first beam of light, the focusing distance f1 and the first optical path 11′ define a first geometric point d1 located below the median plane 23 of the outer surface 22. Here, the two points of impact 13 and 14 corresponding respectively to the meeting of the first optical path 11′ and, respectively, of the second optical path 12′ with the outer surface 22 are spaced apart by a distance X in the direction U.
A portion of the first and second incident beams of light are backscattered by the outer surface 22 at each of the points of impact 13 and 14. A portion of this backscattered light follows the path opposite to the incident optical path. In particular, the two beams of light recombine after the splitter cube 4 so as to jointly continue their route towards the light source 2 and necessarily passing through the light sensor 3. The meeting of the first incident beam and the backscattered beam generates electromagnetic interference as long as these two beams are mutually coherent.
This generation system 1 according to the first embodiment, using a single light source 2, but also equipped with a single light sensor 3, delivers only a single electrical signal at the output of the light sensor 3, which it directs towards the electronic device comprising the signal amplifier 9. The light sensor 3 should be capable of translating the electromagnetic interference that arises at this light sensor 3 into an electrical signal. The location of the light sensor 3 on the first optical path is unimportant, but it is preferable to position it where the sum of the two beams of light generates the smallest beam size in order to optimize interference.
In order to dissociate, on the single electrical signal from the light sensor 3, the information resulting from the interference of the first and second beams of light, it is preferable for the information to be easily detectable. However, the payload information is carried here in particular by the fundamental of the Doppler frequency and its harmonics. The Doppler frequency is theoretically dictated by three parameters. The first corresponds to the relative speed V in the direction U between the medium 21 and the device on which the generation system 1 is mounted. The second is the angle of incidence θ between the beam of light and the normal to the outer surface 22, which will be taken as being that of the median plane 23. The third is the wavelength λ of coherent light. Here, only the angle of incidence θ of the beam of light is able to generate an effective parameter to dissociate the two Doppler frequencies. Indeed, the other two parameters are potentially identical by design of this generation system 1. Therefore, the first and second angles of incidence with respect to the normal to the median plane 23, and reference is made here to their projection in the plane defined above, are different, thereby making it possible to dissociate the Doppler frequencies from one another. This difference should be at least 3 degrees in order to dissociate the signals from each beam of light.
Finally, in order for the signals from each beam of light to be able to be used to determine the distance d from the outer surface 22, and thereby allow the reconstruction of the profile of the outer surface 22 along the readout line, it is necessary to separate the two geometric points d1 and d2 by a length greater than the greatest Rayleigh length of the two Gaussian beams of light. Indeed, the Gaussian propagation of light ensures that the amount of light backscattered is proportional to the distance between the focal point of the beam of light and the outer surface 22 where the backscattering takes place. In fact, the energy will be at a maximum if the outer surface 22 is located at the focal point. The further one moves away from it, the more the backscattered light energy decreases following a Gaussian curve, thereby providing a measurable dynamic range on electromagnetic interference. Ensuring a sufficient spacing between the two focal points ensures that a combination of the electromagnetic interference measured on each channel makes it possible to deduce the distance d from the outer surface 22.
It is possible to observe electromagnetic interference through the self-mixing phenomenon or optical feedback phenomenon, by using a laser source as light source 2, equipped for example with a photodiode as light sensor 3. The temporal record of the output signal from the photodiode is then an image of the interference of light between the incident beam and the beam backscattered by the outer surface. The temporal variations in amplitude are due both to the distance between the focal point d and the outer surface 22 and the reflectivity of the outer surface 22. The information is partly carried by the Doppler frequency related to the relative speed V between the medium 21 and the generation system 1. By ensuring two measurement channels each corresponding to a beam of light pointing to the same readout line of the outer surface 22 the focal point of which is different, the two measurement channels are each the result of the reflectivity of the outer surface 22 and the distance between the focal point and the outer surface 22. The combination of the two channels makes it possible to make the result of the two channels insensitive to the reflectivity of the outer surface 22 and to be dependent only on the distance of the two focal or geometric points from the outer surface 22.
Similarly, the second beam of light, which is the other portion of the first beam of light, follows a second optical path towards the outer surface 22 of the medium 21. Here, the means for routing this second beam of light comprises a mirror 7 that redirects the second beam of light towards the outer surface 22. On the second optical path is located a second focusing lens 6, the focal length f2 of which focuses the second beam of light at the geometric point d2 located above the median plane 23. The point of impact 14 of the second beam of light along the second optical path is identical to the first point of impact 13 of a portion of the first beam of light. Only the projection θ2 of the angle of incidence of the second beam of light on the outer surface 22 with respect to the normal to the median plane 23 differs from the projection θ1.
In this
In order that the backscattered beam from the second optical path interferes as little as possible with the electromagnetic interference between the first beam of light and its backscattered beam at the first light sensor 3, it is preferable for the two beams of light downstream of their focusing lens, that is to say on their optical path, not to intersect before the outer surface. This is a precaution to be taken for any generation system 1 of the first embodiment having a single light source 2.
Here, the two points of impact 13 and 14 of the first 11 and second 12 optical paths are coincident. This avoids any time correction between the two coupled signals at the light sensor 3. The two signals will thus be able to be used directly in order to deduce therefrom the distance d from the outer surface 22 of the medium 21, thereby allowing faster real-time processing. In addition, the use of a single sensor does not require any synchronization of the signals, also limiting small errors, thereby reducing noise on the signals.
This is the preferred set-up of the first embodiment due to the absence of these time corrections on the signals, which speeds up the processing of the signals and limits noise on the signals. In addition, it is an inexpensive set-up since a single light source and a single sensor are used.
Of course, these examples are specific applications of the invention, which is not intended to be limited to these examples. In particular, any combination of the characteristics of these examples is conceivable and falls within the general scope of the invention.
The first phase comprises a first step 100 consisting in obtaining two time signals A and B representative of the profile of the outer surface with respect to a readout line. These may for example be the output of the electronic device of the generation system according to the invention. Of course, in this step, it is not certain that the two signals are temporally and spatially phased, which means having to go through the next step 1001. For example, these two points are separated along the readout line of the outer surface by a spacing X, as in the example of
The second step 1001 corresponds to the spatio-temporal correction to be applied to one and/or the other of the time signals A and B from step 1000. For this purpose, it is necessary to know the method for obtaining the two time signals, that is to say the spatio-temporal spacing between the two measurement points each corresponding to a time signal with respect to a common reference. The spatial position may be a metric position that is obtained visually, for example. The time offset may be the date of crossing in front of a reference point serving for example as a common reference, through a clock signal with a metric for each signal. In addition, it is useful to know the scrolling speed along the readout line of the outer surface associated with each time signal. All of these data make it possible to define a correction matrix to spatio-temporally recalibrate the two signals on one and the same geometric point of the readout line. Applying this correction to the time signals from step 1000 gives the result of step 1002, which ends the signal preparation phase.
The second phase corresponds to formatting of the measured data, which are represented by the time signals obtained in step 1002 from the first phase. The principle of the method according to the invention is that the payload information of the time signals is contained in the fundamental and the harmonics of the Doppler frequency associated with the relative speed V of the medium 21 with respect to the time signal measurement system. This is independent of the physical means for measuring the signals, whether this be light, sound or any other electromagnetic wave.
The first step 2001 consists in defining the Doppler frequency associated with the relative speed V. The Doppler frequency may be determined using a mathematical formula such as, in the case of light signals, the formula linking the relative speed V, the angle of incidence with respect to the normal to the outer surface and the wavelength of the light. It may also result from analysing the signals, whether this analysis be temporal or frequency analysis. Knowing this Doppler frequency, it is necessary for the sampling frequency of the time signals to be at least twice as great as the Doppler frequency, complying with the condition of Shannon's Theorem, in order to ensure that the information of the time signals is plausible and not induced by uncertainty related to the measurement conditions, this corresponding to step 2002. Optionally, it may prove useful to filter the time signals around the Doppler frequency identified in step 2001, and this may be carried out for example over a wide band of between 0.7 and 1.3 times the Doppler frequency. Thus, depending on the mode for identifying the Doppler frequency, theoretically or through frequency analysis of the signals or through temporal analysis of the signals, and also the potential slow evolution of the relative speed V, the Doppler frequency is not necessarily determined in absolute terms, and a wide window then makes it possible to cover all of these uncertainties by isolating the usable information, this corresponding to step 2003. Of course, if the frequency interference related to the signal measurement system is low, it is entirely conceivable to take the complete signal without selective filtering and move directly to step 2004.
Step 2004 consists in focusing on the general signal carrying the information through the envelope of the payload signal. It is expected that this will be an image of the events related to the Doppler frequency associated with the relative speed V. In step 2004, the envelope of the payload time signal is determined, potentially driven by a narrow frequency band around the Doppler frequency. Of course, the envelope of the payload signal may be constructed from the minima, the maxima or the absolute value of the payload signal. The choice of method depends on the nature of the measured signals with respect to the physical quantity under observation.
Optionally, in order to statically eliminate parasitic noise on the envelope of the measured time signal, speckle cleaning is carried out in order to extract the precise information therefrom in step 2005. This makes it possible to statistically eliminate measurement randoms caused by lack of compliance with the conditions for an ideal measurement. This is carried out through a learning campaign on a known target representative of the outer surface of the medium that it is desired to observe using the envisioned measurement system. This learning phase determines a Gaussian distribution of the measurement randoms, which should be coupled with an evenly distributed noise in order to determine a speckle noise. This makes it possible to determine the time windowing of the payload signal that should be taken into account in order to clean the speckle noise by applying the identified Gaussian distribution. Step 2005 consists in removing the determined speckle noise from the envelope signal in order to obtain a cleaned envelope on each measurement channel. This step ends the second data formatting phase.
The last phase is evaluating the variation in the distance d of the outer surface from a reference, making it possible to deduce the profile of the outer surface. This comprises a first step 3001, which consists in mathematically combining the envelopes obtained in steps 2004 or 2005 so as to define a function F that is bijective. The bijectivity of the function F makes it possible to guarantee the uniqueness of the distance d from the outer surface using the information from the two envelopes. In the case of self-mixing, or optical feedback, with Gaussian and coherent beams of light, defining the function F as being the difference between the envelopes expressed on a logarithmic scale ensures both monotony and good sensitivity of the function F over the distance range separating the two geometric points d1 and d2 of the generation system presented in the device invention. Precision is enhanced by taking the absolute value of the payload signal to construct the envelopes. Of course, the precision improves when taking the cleaned envelopes.
Finally, to arrive at a relative distance d between various points of the readout line of the outer surface with respect to a reference geometric position, it is necessary to establish a calibration between the result of the function F as defined in step 3001 and a target the position of which is known with respect to the geometric points of the measurement system, this corresponding to step 3002. This makes it possible to convert the response of the function F into a known metric quantity.
To this end, a calibration step should be undertaken using the measuring device, directly or indirectly delivering the time signals with respect to two different geometric points. In the case of the generation system of the device invention, the two geometric points are the points d1 and d2 where each of the beams of light are focused and located on either side of the median plane of the outer surface of the medium. Here, the calibration is performed using a target the physical response of which is at least as strong as the outer surface of the medium that it is desired to observe. In the case of the generation system of the device invention, it is necessary to use a white target, that is to say having very high reflectivity with respect to the observation medium. The majority of the incident light is thus backscattered by the surface, which absorbs a very small proportion thereof. In addition, in order to observe light scattering, the target should be rough. However, in order not to be penalized by a large degree of integration between the light from the generation system and the target, the surface roughness of the target should be greater than that of the medium under observation. It is then sufficient to calibrate the generation system by moving the target between the geometric points d1 and d2 in a known manner and to identify the value of the corresponding function F using the envelopes. This calibration will be used in step 3002 to obtain the distance d from the outer surface of the system. Here, the function F is insensitive to the backscattered power, since the function F is a relative combination of the signal envelopes, such as the linear scale ratio or the logarithmic scale difference. If the combination of the envelopes is absolute, it will be necessary to perform a more precise calibration using a target the physical properties of which are similar to those of the medium that it is desired to observe using the measuring device according to the metric used: light, sound or electromagnetic waves.
The optional speckle noise correction is also performed using the same target as in the signal calibration step. This time, the number of time measurements is increased by moving the target, knowing the result to be achieved in order to evaluate the distribution of the measurements around the reference value. For this purpose, the distribution of the measurements is evaluated in the frequency domain under diversified measurement conditions on a large time sample. This frequency distribution is modelled by a centred Gaussian. The width of this Gaussian determines the minimum size of the measurement time window, so that the Gaussian distribution is statistically representative. The speckle noise is then evaluated through the product of the Gaussian frequency distribution and a noise uniformly distributed between 0 and 1. This speckle noise is to be subtracted from the determined envelope in order to obtain a measurement that depends in the first order only on the reflectivity of the target or the outer surface of the medium under observation. Proportionality is assumed between the reflectivity of the target and that of the outer surface of the medium to be observed, which will be transparent due to the relative combination of the envelopes.
To apply the method, use was made of a preferred generation system, the principle of which is illustrated in
The generation system comprises, as light source, a laser diode equipped with a photodiode at the entrance of the amplifying cavity of the laser diode. The laser diode emits a beam of coherent, monochromatic light at the single wavelength and the propagation of which along the direction of the beam is Gaussian. Here, the wavelength of the first laser diode is centred on 1350 nanometres. The second, meanwhile, is centred on 1500 nanometres. The photodiode associated with each laser diode records the electromagnetic interference between the incident beam of light and the beam of light backscattered by the outer surface of the test specimen. The electromagnetic interference of the first optical path is mainly carried by the harmonics of the first Doppler frequency related directly to the first wavelength, which is inversely proportional to the wavelength. On the other hand, the electromagnetic interference of the second optical path is carried by the harmonics of the second Doppler frequency, the Doppler frequency of which is lower than the first Doppler frequency. The differentiation of the geometric points d1 and d2 where the first and second optical paths are collimated is defined by the length of the optical paths. Here, the first optical path towards the outer surface is defined by the first focusing lens through its position on the first optical path and its focal length. The second optical path comprises an optical splitter element in order to redirect the second beam of light towards the outer surface of the test specimen. The geometric point d2 is controlled directly by the positioning and the focal length of the second focusing lens of the generation system. Thus, at the output of the electronic device, an electrical signal associated with a photodiode is obtained, containing the payload information carried by the harmonics of each different Doppler frequency.
The measurement is carried out by fixing the generation system on a static device located in line with the test specimen such that the geometric points d1 and d2 are located on either side of the outer surface of the test specimen. In our case, they are equidistant from the median plane of the outer surface of the test specimen, at a distance of around 5 millimetres. The spacing between the geometric points is thus of the order of a centimetre, which is less than the variations in the profile of the outer surface of the test specimen, while being greater than the Rayleigh length of the first and second Gaussian beams of light.
Finally, a surface profile illustrated in
| Number | Date | Country | Kind |
|---|---|---|---|
| 2004162 | Apr 2020 | FR | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2021/060957 | 4/27/2021 | WO |
