Method for detecting particles in a medium, and equipment for implementing this method

Abstract

TITLE: Method for detecting particles in a medium, and equipment for implementing this method. The present invention relates to a method for the optical detection of particles in a medium, comprising capturing the light scattered by the medium flowing through an illuminated detection zone. According to the invention, the raw signal 31 from the sensors is processed before being analyzed. This processing includes identifying, over a reference period, the minimum value (32) of the raw signal (31), determining the amplitude (34) of the noise in the raw signal (31) by calculating the derivative of order greater than or equal to 1 of this raw signal (31) and multiplying this derivative by a predetermined constant value, and correcting the raw signal (31) by subtracting the minimum value (32) and half the amplitude (34) of the noise from the value of the raw signal (31). Abstract figure: Figure 3

Description

Title of the invention: Method for detecting particles in a medium, and equipment for implementing this method. Field of the invention

[0001] The present invention relates to methods for detecting particles in a medium. It relates in particular to such optical methods, involving illumination of the medium and capture of the light scattered by the particles present in that medium.

[0002] The invention also relates to optical detection equipment enabling the implementation of such a particle detection method in a medium. Previous art

[0003] It is known to use optical methods to detect particles in a transparent medium such as, for example, air, water, or any liquid or gas.

[0004] Such optical particle detection methods involve illuminating the medium in question with a light source, which is generally chosen so that the medium is relatively transparent to the light source and the particles of interest are relatively opaque. Capturing and measuring the light scattered by the particles present in this medium, in response to the illumination, allows for the detection of the particles present in the medium.

[0005] In this application, the expression "detection of particles in a medium" is used generically to designate a set of operations which may include, without limitation, the detection, counting, measurement, identification, and characterization of particles in that medium. Similarly, the expression "light scattering" is used generically to designate the emission of light produced by an illuminated particle, this light emission being caused by any known optical phenomenon such as diffraction, reflection, scattering, radiation, phosphorescence, etc.

[0006] As an example, it is known from document FR3105829A1 to detect the presence in the air of relatively large particles, such as pollen grains, by illuminating them with an appropriate laser source and measuring the light scattered by the illuminated air, according to several measurement angles.

[0007] For particle detection to be effective, it is important that the sensors used for measurement be capable of measuring the light actually scattered by the particles present in the illuminated medium, and distinguishing it from light that may come from other light sources. For this reason, optical devices for detecting particles in a medium include typically a dark chamber in which the analyzed medium circulates and optical measurement devices, including sensors, which are configured to avoid any illumination of the sensors by light sources other than the illuminated particles that one seeks to detect.

[0008] Despite all efforts, the signal measured by the sensors generally includes, in addition to the signal representing the light scattered by the illuminated particles being sought, a quasi-continuous component (i.e., continuous or varying very slowly compared to the unit measurement) and noise that are not representative of this light. This quasi-continuous component and this noise can originate from several causes. In particular, they may include the detection of stray light, not originating from the particles being sought, which is received by the sensors. For example, they may include light scattered by a collection of particles that generate stray light detected by the sensors. This quasi-continuous component and this noise may also include electronic noise resulting from the operation of the electronic measuring instruments.They may also include a fault in the calibration of the sensors, or a variation in this calibration during the measurement.

[0009] When optical detection devices are used for the optical detection of relatively large particles or particles that scatter light well, the signal representing the light scattered by the illuminated particles is relatively stronger than the quasi-continuous component and the noise interfering with this signal, and particle detection is effective. However, using optical detection devices for the optical detection of small particles, or particles that scatter light less well, is more difficult. Indeed, the signal representing the light scattered by the illuminated particles can be at least partially masked by the quasi-continuous component and by the noise interfering with this signal, which can prevent the correct detection of the particles.

[0010] Researchers attempting to detect small particles or particles that scatter light very little try to limit the quasi-continuous component and noise by using more sophisticated measuring devices, high-performance and expensive electronic components, and by performing very frequent calibrations. These precautions are costly, complicated to implement, and only allow for an imperfect reduction of the quasi-continuous component and noise.

[0011] It is thus known to carry out, on a regular basis, a calibration of the detection device, aimed at determining the quasi-continuous component of the measured signal, appearing outside of any detection of light from the particles sought present in the detection zone, and subtracting the value of this quasi-continuous component from the value of the measured signal.

[0012] Such calibration generally involves an interruption of detection, stopping the illumination or introducing into the detection area a pure medium, containing no visible particles at the wavelength of the illumination light beam, and measuring the signal captured by the photosensitive sensors under these calibration conditions.

[0013] The signal measured under these calibration conditions corresponds to a quasi-continuous component of the signal measured during particle detection, which is not representative of the presence of the particles sought in the detection zone. However, it is observed that this signal measured under the calibration conditions is systematically lower than the quasi-continuous component actually appearing in the signal measured during particle detection. Therefore, a correction is commonly applied to the signal measured under the calibration conditions. However, applying this correction can introduce bias into the measurements performed to detect particles.

[0014] There are also methods for continuously suppressing the quasi-continuous component of the signal measured during particle detection, which consists of filtering the measured signal with a high-pass filter, either digital or analog. Such filtering removes the quasi-continuous component of the measured signal. However, such filtering also distorts the measurement. Thus, for example, an increase in the measured signal corresponding to the detection of a large particle leads to a bias in the filtered value, which only becomes representative of the presence of particles in the detection zone after a certain time.

[0015] There is therefore a need for particle detection methods in a medium which allow a better measurement of the signal representative of the light scattered by the illuminated particles, by better distinguishing it from the quasi-continuous component of the raw measured signal. Description of the invention

[0016] The present invention aims to overcome these drawbacks of the prior art.

[0017] In particular, the present invention aims to provide a method of particle detection in a medium, which allows for more reliable and more precise detection of particles whose signal level is on the order of magnitude of electronic noise and / or scattering little light.

[0018] A particular objective of the invention is to provide such a method which can be implemented in a simple and inexpensive way.

[0019] Another objective of the invention is to provide such a method which consumes relatively little computing power, and therefore little energy.

[0020] Another objective of the invention is to provide such a method which can be implemented without requiring interruption of particle detection.

[0021] Yet another objective of the invention is to provide such a method which can be implemented in real time, in a particle detection equipment.

[0022] These objectives, as well as others which will become clearer later, are achieved using an optical particle detection method in a medium, comprising the following steps: - a step of setting the medium in motion through a detection zone, - a step of illuminating the medium within this detection zone, - a stage of capturing light by at least one photosensitive sensor diffused through the medium in the detection zone, - a step of processing the raw signal produced by the photosensitive sensors, - a step of analyzing the processed signal to detect particles present in the medium, within the detection zone, the raw signal processing stage comprising, according to the invention: - a sub-step for identifying, over a reference period longer than the maximum time a particle can spend in the detection zone, the minimum value of the raw signal, - a sub-step of determining the noise amplitude of the raw signal, by calculating the derivative of order greater than or equal to 1 of the raw signal, and multiplying this derivative by a predetermined constant value, - a sub-step of correction of the raw signal, by subtracting the minimum value of the raw signal and half the amplitude of the noise from the value of the raw signal.

[0023] Such a method advantageously allows for more efficient processing of the raw signal than prior art methods. It enables the precise calculation, in real time and without interrupting detection, of the exact value of the quasi-continuous component of the raw signal, which is then subtracted from the raw signal to calibrate it. This improved processing of the raw signal advantageously facilitates the detection of particles, particularly small particles that generate only a relatively weak signal compared to the noise.

[0024] Advantageously, the determination of the noise amplitude of the raw signal is carried out by calculating the derivative of order greater than or equal to 2 of the raw signal, and multiplying this derivative by a predetermined constant value.

[0025] The use of the derivative of order greater than or equal to 2 allows for a more precise determination of the noise amplitude of the raw signal, the component of this derivative originating from the significant signal becoming much smaller than the component of this derivative originating from the noise.

[0026] Preferably, the determination of the noise amplitude of the raw signal is carried out by calculating the derivative of order greater than or equal to 3 of the raw signal, and multiplying this derivative by a predetermined constant value.

[0027] The use of the third-order derivative, and even more preferably the fourth or higher order, allows for an even more precise determination of the noise amplitude of the raw signal, even in situations where the amplitude of the significant signal is large relative to the noise amplitude. Indeed, for such derivatives, the component of the derivative originating from the significant signal becomes negligible compared to the component of this derivative originating from the noise.

[0028] Preferably, the identification of the minimum value of the raw signal is repeated at regular intervals, the duration of these intervals being equal to the duration of said reference period.

[0029] This regular repetition of the search for the minimum value in a time window advantageously allows all minimum values ​​of the raw signal to be detected, while considerably reducing the calculations of comparisons of the signal values.

[0030] Advantageously, the duration of the reference period is greater than 100 times the maximum duration of passage of a particle in the detection zone.

[0031] This duration advantageously ensures that there will be values ​​of the measurement which will correspond to times when no particle is detected.

[0032] The invention also relates to optical detection equipment, capable of implementing the method described above, and / or configured to implement this method. Description of the figures

[0033] The invention will be better understood upon reading the following description of preferred embodiments, given by way of simple figurative and non-limiting example, and accompanied by the figures, among which: - Fig. 1 is a schematic representation of the steps of an optical particle detection process in a medium, according to an embodiment of the invention. - Fig. 2 is a schematic representation of the sub-steps implemented during the raw signal processing step, according to one embodiment of the invention. - Fig. 3 is a graphical representation of a portion of the raw signal representative of the light coming from the illuminated detection area, when a particle crosses this area. Description of the implementation methods

[0034] Fig. 1 schematically represents the steps of an optical particle detection process in a medium, according to an embodiment of the invention.

[0035] The steps are represented sequentially in this diagram. However, in a preferred embodiment of the invention, they are not carried out sequentially but simultaneously, continuously.

[0036] The first step 11 represents the movement of the medium in which particles are sought, so as to make it circulate through an area subsequently called the "detection zone".

[0037] The medium may consist of a gas or a liquid, capable of containing the particles of interest. For example, it may be air, in which the presence of dust grains, pollen grains, fine particles, etc., is being sought. It may also be a liquid such as water. Its movement can be achieved by means of a pump well known to those skilled in the art.

[0038] It is necessary that this medium in which particles are sought be at least partly transparent to light, at least at one wavelength, of the visible spectrum or outside the visible spectrum.

[0039] The detection zone through which the medium in which particles are sought circulates is advantageously a zone in which the light is controlled. It is generally a dark chamber of a detection device, isolated from all external light.

[0040] The second step 12 represents the illumination of the detection area by a light beam. The light source producing this light beam is advantageously chosen so that the medium is at least partially transparent to the light beam, and so that the particles of interest are not transparent to this light beam. The illumination can, for example, be caused by a directional light beam such as a laser beam.

[0041] The third step 13 represents the capture, by suitable photosensitive sensors, of a raw signal representative of the light coming from the illuminated detection area. By way of example, [Fig. 3] shows a curve, with arbitrary units, representing the evolution of such a raw signal 31, over a short period of time hereafter referred to as the "time window".

[0042] The detection device is preferably constructed such that, as far as possible, only the light from the detection zone is received by the photosensitive sensors, and that the light from the light beam providing the illumination, either directly or through reflection or diffusion on other surfaces other than those of illuminated particles in the detection area, are not received by the photosensitive sensors, or are received only very little.

[0043] In a manner known per se, this capture can be carried out simultaneously by several photosensitive sensors. These sensors can, in particular, measure the light scattered by the particles present in the detection zone in several directions, forming different angles with the direction of the light beam illuminating this detection zone.

[0044] Step 14 relates to the processing of the raw signal captured by the photosensitive sensors. This processing may include, in a known manner, a calibration of this signal intended to remove the quasi-continuous component, a filtering intended to exclude irrelevant sequential components, and / or a digitization of the signal.

[0045] Step 15 concerns the analysis of the processed signal in order to detect the presence of a particle in the illuminated medium within the detection zone. Since the medium does not normally emit any light, even when illuminated by the light source, the detection of light from the detection zone means that the medium present in this detection zone contains at least one particle.

[0046] Calculating the intensity of the light, its color, the duration of its emission, etc., can provide information to characterize the particle that caused the light scattering, and thus to identify it.

[0047] The process as described above is known, in itself, to a person skilled in the art.

[0048] The inventors have developed an improvement to step 14 of raw signal processing, described above, making it possible to substantially improve the accuracy of the suppression of the quasi-continuous component of the raw signal captured by the photosensitive sensors, and thus to facilitate the detection of particles during step 15 of analysis of the processed signal.

[0049] Figure 2 schematically represents the substeps that can be implemented during this step 14 of raw signal processing, according to one embodiment of the invention. These substeps are preferably implemented simultaneously and continuously.

[0050] Fig. 2 is a graphical representation of a portion of the raw signal from a sensor measuring the light from the illuminated detection area when a particle crosses this area.

[0051] This raw signal can be understood as the sum of several distinct sources. It thus comprises: - the representative signal of the light coming from the illuminated particle (hereafter called the "significant signal", which is normally a bell-shaped signal of Gaussian type); - a quasi-continuous component, which may originate from several distinct sources; - a noise, oscillating randomly or pseudo-randomly around a neutral value, which may come from several distinct sources.

[0052] In this description, the term "quasi-continuous component" refers to the component of the signal that is continuous or that exhibits such a small variation, relative to the variations of the significant signal and the noise, that it can be considered continuous for the purpose of analyzing the raw signal to detect the significant signal representative of particle detection. One of the objectives of step 14 of raw signal processing is to determine, as precisely as possible, the amplitude values ​​of the quasi-continuous component and the noise, in order to determine the value of the significant signal as precisely as possible.

[0053] According to an advantageous embodiment of the invention, a substep 21 of step 14 of raw signal processing consists of identifying a minimum value of this raw signal.

[0054] For this purpose, it can be determined, depending on the medium studied, the volume of the search area and the flow rate of the medium in this search area, a reference time window during which it is statistically certain that the medium will, at least at one time, be devoid of particles in the search area.

[0055] This time window must be longer than the transit time of a particle in the search zone. In the preferred embodiment of the invention, the time window is chosen to be at least 100 times greater than the transit time of a particle in the search zone. Thus, in this embodiment, the transit time of a particle being estimated to be between 100 µs and 1 ms, for particles sought with a size between 100 nm and 200 µm, the chosen time window lasts 500 ms.

[0056] The time window represented by [Fig. 3] is, however, much shorter, to facilitate the visualization of the solution of the invention in this figure. Thus, in this example, the duration of the time window shown is approximately twice the time it takes for the particle to pass through the search area.

[0057] Substep 21 of identifying the minimum value consists of searching, within this time window, for the minimum value of the raw signal captured by the photosensitive sensors. When the raw signal is a digital and discretized signal, this search amounts to selecting the minimum value from among the values ​​composing the raw signal within the time window.

[0058] In the example of the signal time window represented by [Fig.3], the minimum value of the raw signal 31 is represented by the horizontal line 32, and is equal to 50 (arbitrary unit).

[0059] This minimum value is close to, but not exactly equal to, the value of the quasi-continuous component of the raw signal. Indeed, the noise causes the value of the raw signal to vary, positively or negatively, around the value of the quasi-continuous component. The value of the quasi-continuous component can therefore be considered as the minimum value of the raw signal plus half the amplitude of the noise.

[0060] To determine the value of this quasi-continuous component, it is therefore necessary to also know the value of the noise amplitude.

[0061] Substep 22 of determining the noise amplitude comprises, according to the invention, calculating the derivative of order greater than or equal to 1 of the raw signal value. The raw signal 31 can be considered as the sum of the quasi-continuous component, the noise, and the significant signal; the derivative of this raw signal is the sum of the derivative of the quasi-continuous component, the derivative of the noise, and the derivative of the significant signal.

[0062] Due to the very small variations in this component, the derivative of the quasi-continuous component has a very small value. If this component is continuous, its derivative is, by definition, zero. If it varies slowly and slowly, its variation during the analysis time window is extremely small, and the value of its derivative can be considered negligible compared to the derivatives of the other components of the raw signal.

[0063] The inventors have identified that the noise appearing in the raw signal captured by the photosensitive sensors behaves approximately like white noise, in a random manner. When the raw signal is a discretized digital signal, this noise is thus close to electronic white noise, with no average value, zero expectation, constant power spectral density, and is stationary (i.e., these properties are invariant under time translation). The derivative of such random, time-invariant electronic white noise is also random, time-invariant electronic noise. Denoting P as the value of perfect white noise and x as any discrete value, calculating the derivative P'(x) is equivalent to adding two noises: P'(x) = P(x) - P(x) Since the statistical properties of P(x) are invariant over time, the maximum value of this derivative P'(x) can be written as: Since noise has no average value, the maximum value of this derivative P'(x) can be written as: max P'00 = (P 00 — P(> — 1)) max ( r 1)) = max (P(x)) -min (P(x)) ___ v- <? '■ __-L. s-’."- ' ' •' The maximum value of the derivative P'(x) can therefore be written as: u. Pf(x) = max FhJ n- f' ' -a> <xsf« max F(xl) (x) = 2x max P(x) The maximum value of the multiple derivative, of order n, of this noise P(x), can also be written as: max PSB-^(x)=2xnx max (F(x)) For such white noise, the amplitude of the nth order derivative of P(x) is therefore equal to the amplitude of P(x) multiplied by a multiplicative constant a = 2 x n.

[0064] The inventors observed that the noise appearing in the raw signal captured by the photosensitive sensors does not behave exactly like white noise, but that, nevertheless, the amplitude of the nth-order derivative of this noise is equal to the amplitude of this noise multiplied by a multiplicative constant b. Since the noise is not perfectly random, this constant b is different from the constant a defined above, but it can be expressed in the form b = cxn, in which "c" is a constant less than 2, invariant on a given piece of equipment, which can be measured by an initial calibration on the equipment or on identical equipment.

[0065] For example, on the equipment implemented in an embodiment of the invention, a constant c = 0.5 has been determined, such that, on this equipment, the amplitude of the nth order derivative of the noise P(x) can be written: IliaX Dn-ieme m aX . . (X)=0.5X]]X co< < +oo (P(X))

[0066] In the raw signal captured by the photosensitive sensors, the significant signal, representative of the detection of light scattered by a particle crossing the detection zone, is normally a Gaussian signal whose formula is:

[0067] In the example of the signal time window represented by [Fig.3], this significant signal is represented by curve 33.

[0068] The derivative of such a Gaussian signal has a much smaller amplitude than the amplitude of the signal itself. Thus, the amplitude of the first-order derivative of a Gaussian signal is approximately 15,000 times smaller than the amplitude of the original Gaussian signal.

[0069] This attenuation of the derivative's amplitude, relative to the original Gaussian signal, is much greater for higher-order derivatives. Thus, the amplitude of the fourth-order derivative of a Gaussian signal is approximately 1014 times smaller than the amplitude of the original Gaussian signal.

[0070] When calculating the amplitude of the nth-order derivative of the raw signal captured by the photosensitive sensors, this derivative is therefore the sum of: - the amplitude of the noise, multiplied by a known multiplicative factor, and - a fraction of the amplitude of the significant signal, representative of the detection of light scattered by a particle crossing the detection zone.

[0071] When the derivative is a first-order derivative, the fraction of the significant signal amplitude can be considered negligible compared to the noise amplitude in certain situations. This may be the case, for example, when the medium in which the particles are being sought has been pre-filtered and contains only small particles, generating only a significant signal of low amplitude.

[0072] When the derivative is a second-order derivative, the fraction of the amplitude of the significant signal can be considered negligible compared to the amplitude of the noise, in many situations.

[0073] When the derivative is a derivative of order 3 or 4, the fraction of the amplitude of the significant signal can be considered negligible compared to the amplitude of the noise, in almost all situations.

[0074] In all these situations, the nth-order derivative of the raw signal captured by the photosensitive sensors can therefore be considered equal to the noise amplitude multiplied by a known multiplicative factor b = c x n. It is thus possible to directly obtain the value of the noise amplitude by calculating this nth-order derivative of the raw signal captured by the photosensitive sensors. This calculation therefore allows the noise amplitude to be determined with high precision.

[0075] Substep 22 of determining the noise amplitude therefore includes, according to the invention, calculating the derivative of order greater than or equal to 1 (and preferably of order greater than or equal to 2, preferably of order greater than or equal to 3, and even more preferably greater than or equal to 4) of the value of the raw signal, and multiplying this derivative by a known multiplicative factor.

[0076] In the example of the signal time window represented by [Fig.3], the amplitude of the noise of the raw signal 31 is represented by the arrow 34, and is substantially equal to 2 (arbitrary unit).

[0077] Substep 23 of the raw signal correction consists of subtracting from the raw signal value: - the minimum value of this raw signal, and - half the value of the noise amplitude.

[0078] This subtraction makes it possible to remove from this raw signal, measured by the photosensitive sensors, the quasi-continuous component of this raw signal, while retaining the noise and the significant signal, representative of the detection of light scattered by a particle crossing the detection zone.

[0079] In the example of the signal time window represented by [Fig.3], subtracting the raw signal value 31 from the minimum value 32 and half the noise amplitude 34 allows this raw signal to be placed relative to a reference value 35, here equal to 51, which becomes the new value "0", representing the absence of a significant signal.

[0080] Implementing such a correction to the raw signal value, either continuously or regularly, allows for greater accuracy in calibrating this raw signal. It can thus be continuously adapted to the operating conditions of the detection device, which may vary over time, for example, depending on characteristics such as the temperature of the various components, their aging, the lighting conditions in the detection area, etc.

[0081] Furthermore, implementing such a correction to the raw signal value also advantageously avoids the detection interruptions required by occasional calibration operations. It therefore allows for easier and less expensive calibration of the raw signal than prior art solutions, requiring only moderate computing power and consuming very little energy.

[0082] The identification of the minimum value of the raw signal, the calculation of the noise amplitude, and the correction of the raw signal value can be performed continuously on the signal. In such a case, the identification of the minimum value of the raw signal can be done over a sliding time window.

[0083] It is also possible for some of the steps to be performed regularly, without being performed continuously. Thus, the identification of the minimum value of the raw signal can be done over a sliding time window and can be carried out at regular intervals which can advantageously be of a duration substantially equal to the duration of the time window. Such a solution makes it possible to limit the computing power required by searching for the minimum value only once in each time window.

[0084] Similarly, the calculation of the noise amplitude can be carried out continuously or at regular intervals.

[0085] The correction of the raw signal value can be performed in real time, even if the identification of the minimum value of the raw signal and / or the calculation of the noise amplitude are carried out at regular intervals. In such a case, the correction of the raw signal value can be based on the latest values ​​determined for the minimum value of the raw signal and for the noise amplitude.

[0086] It is also possible, in another embodiment, that the identification of the minimum value of the raw signal, the calculation of the noise amplitude and the correction of the value of the raw signal are carried out in a delayed manner, on values ​​of the raw signal which were recorded during an optical detection of particles in this medium, and before the analysis of this signal to detect the particles which were present in this medium.

[0087] It should be noted that determining the noise amplitude can, in some cases, be useful independently of evaluating the quasi-continuous component of the raw signal. Thus, for example, knowing the noise amplitude can be useful for determining more precisely the value of a maximum in the signal representing the detection of a particle. By way of illustration, knowing the noise amplitude can allow us, for the signal represented by [Fig. 3], to determine the maximum value of the significant signal by detecting the maximum value of the raw signal and subtracting half the value of the noise amplitude.

[0088] According to one aspect of the invention, it can therefore be defined as a method for determining the amplitude of the noise in the raw signal produced by a photosensitive sensor of an optical particle detection device in a medium, this raw signal being produced during the capture, by the photosensitive sensor, of the light scattered by a moving medium through a detection zone, in which it is illuminated, the method for determining the amplitude of the noise comprising the calculation of the derivative of order greater than or equal to 1 of said raw signal, and the multiplication of this derivative by a predetermined constant value.

[0089] The derivative is, preferably, of order greater than or equal to 2, preferably of order greater than or equal to 3, and even more preferably, of order greater than or equal to 4.

[0090] More generally, the invention can be defined as a method for determining the amplitude of the noise in a raw signal produced by a sensor, this raw signal being the sum of a quasi-continuous component, random noise and a significant signal of Gaussian type, the method for determining the amplitude of the noise comprising calculating the derivative of order greater than or equal to 1 of said raw signal, and multiplying this derivative by a predetermined constant value.

[0091] The derivative is, preferably, of order greater than or equal to 2, preferably of order greater than or equal to 3, and even more preferably, of order greater than or equal to 4.

Claims

Demands

1. A method for the optical detection of particles in a medium, comprising the following steps: - a step (11) of setting said medium in motion through a detection zone, - a step (12) of illuminating said medium in said detection zone, - a step (13) of capturing, by at least one photosensitive sensor, the light scattered by said medium in said detection zone, - a step (14) of processing the raw signal (31) produced by said photosensitive sensors, - a step (15) of analyzing the processed signal to detect the particles present in the medium, in said detection zone, characterized in that said step (14) of processing said raw signal (31) comprises: - a substep of identifying (21), over a reference period whose duration is greater than the maximum passage time of a particle in said detection zone, the minimum value (32) of said raw signal (31),- a substep (22) for determining the amplitude (34) of the noise in the raw signal (31), by calculating the derivative of order greater than or equal to 1 of said raw signal (31), and multiplying this derivative by a predetermined constant value, - a substep (23) for correcting said raw signal (31), by subtracting said minimum value (32) of said raw signal and half of said amplitude (34) of the noise from the value of said raw signal (31).

2. Optical detection method according to the preceding claim, characterized in that said determination of the amplitude (34) of the noise of the raw signal (31) is carried out by calculating the derivative of order greater than or equal to 2 of said raw signal (31), and multiplying this derivative by a predetermined constant value.

3. Optical detection method according to the preceding claim, characterized in that said determination of the amplitude (34) of the noise of the raw signal (31) is carried out by calculating the derivative of order greater than or equal to 3 of said raw signal (31), and multiplying this derivative by a predetermined constant value.

4. Optical detection method according to any one of the preceding claims, characterized in that said substep of identification (21) of the minimum value (32) of said raw signal (31) is repeated at regular intervals, the duration of these intervals being equal to the duration of said reference period.

5. Optical detection method according to any one of the preceding claims, characterized in that the duration of said reference period is greater than 100 times the maximum passage time of a particle in said detection zone.

Images