Process for rectifying the brightness of a digital image

Abstract

The invention relates to a method for rectifying the brightness of a digital image. The rectification method implements the following steps: a) acquiring a digital image of an object (14) by means of an optical detection device comprising photosensitive sensors (10), the digital image being formed from measurements of variations in light intensity at an observation wavelength, the reflectance of which varies depending on the chemical composition of the surface of the object (14); b) measuring the distance between the surface of the object (14) and the photosensitive sensors (10), the distance measurements being made at a telemetric wavelength, the reflectance of which does not depend on the chemical composition of the surface of the object (14); c) weighting the light intensity of the image acquired by the optical detection device in step a), according to the distances measured in step b).

Description

Description Title of the invention: Method for correcting the brightness of a digital image Technical field to which the invention relates

[0001] The present invention relates to the technical field of digital image processing. More particularly, the invention relates to a method for rectifying the brightness of a digital image, in order to allow a better evaluation of the surface condition of a food product, for example, by means of a hyperspectral imaging method performed from one or more rectified digital images. Technological background

[0002] Hyperspectral imaging is a well-established technique for identifying the physical characteristics of an object's surface in a non-destructive manner. This is achieved by decomposing the light reflected by an object into several spectral bands. The light intensity of each spectral band is then measured using photosensitive sensors arranged in a matrix.

[0003] By comparing the light spectrum recorded by each pixel to known light spectra, it is then possible to locally identify the chemical composition of the object's surface.

[0004] This analytical technique is commonly used in the food industry, particularly on high-speed production lines, to perform quantitative and qualitative measurements of the food produced. For example, hyperspectral imaging allows the detection of foreign bodies such as shells or other residues present in pasta.

[0005] In another example, hyperspectral imaging also allows for the automated assessment of apple harvest quality. More specifically, the ripeness of the apples is quantified based on their reflectance properties.

[0006] In another example, hyperspetral imaging can detect greening on the surface of potatoes. More specifically, the greening is quantified to prevent abnormal amounts of alkaloids that deteriorate the flavor of the potato and can also be toxic when their concentration exceeds the threshold of 200 mg per kilogram.

[0007] Hyperspectral imaging is a well-known and recognized technique in the food industry for quickly and accurately assessing the quality of food production. However, the accuracy of the measurements depends directly on the reflectance properties of the food being studied. These reflectance properties vary according to the food's nature, surface condition, and the angle at which it is illuminated.

[0008] In the food industry, it is common to move food on conveyor belts, where the food is then randomly presented to hyperspectral imaging devices positioned above said belts.

[0009] It has been observed that measurements taken by this type of device are incomplete, or even inaccurate, at the edges of foods with complex shapes. By "complex shape," we mean a food with a three-dimensional form, delimited by one or more rounded or substantially rounded edges, such as an apple, a potato, a grain of rice, or others.

[0010] The invention aims to solve this technical problem, by proposing a method for rectifying the brightness of a digital image of at least one food of complex shape, a digital image acquired by a hyperspectral imaging device. Object of the invention

[0011] To this end, the invention proposes a method for rectifying the brightness of a digital image, implementing the following steps:

[0012] a) Acquisition of a digital image of an object, by means of an optical detection device comprising photosensitive sensors, the image digital is formed from measurements of variations in light intensity, made by the photosensitive sensors of the optical identification device, at an observation wavelength whose reflectance varies according to the chemical composition of the surface of said object;

[0013] b) measurement of the distance between the surface of the object and the photosensitive sensors of the optical detection device, the distance measurements are carried out at a rangefinding wavelength whose reflectance does not depend or depends very little on the chemical composition of the surface of the object;

[0014] c) weighting of the light intensity of the image acquired by the optical detection device during step a), as a function of the distances measured during step b).

[0015] The term "digital image" refers to an image composed of a multitude of pixels, the light intensity of each pixel being proportional to the amount of light perceived by a photosensitive sensor of the optical detection device.

[0016] Thus, during step a) described above, the digital image is formed from the light intensities measured by the photosensitive sensors, at a specific wavelength called the observation wavelength.

[0017] In step b) of the process according to the invention, the distance between each photosensitive sensor and the object observed by the optical detection device is measured by means of a rangefinding process, at a wavelength also specific and called the rangefinding wavelength.

[0018] Remarkably, the observation wavelength is chosen so that its reflectance properties depend on, or vary with, the chemical composition of the observed object's surface. The term "depend" means that the reflectance properties of the observation wave vary significantly, allowing the optical detection device to detect these variations.

[0019] According to the invention, variations in the reflectance properties of the observation wavelength are considered significant when the luminous intensity of said observation wavelength reflected by the observed object varies positively or negatively by more than 10%, 20%, 30%, or 40%. relative to the light intensity of the observation wavelength illuminating said object.

[0020] For example, when the object observed by the optical detection device is food, the observation wavelength is selected so that the amount of light at that wavelength reflected by the surface of the food varies according to its ripeness, state of decomposition, or the presence of a pathogen on its surface. In other words, the observation wavelength aims to allow for the optical assessment of food quality based on its surface condition.

[0021] Preferably, the observation wavelength is chosen according to the nature of the food, so that the intensity value of said wavelength varies positively or negatively by more than 10%, 20%, 30%, or 40% after the surface of the food is covered by a fungus or after the appearance of a surface defect. The term "surface defect" means an appearance or color of the food that does not conform to a production standard.

[0022] According to another notable feature of the invention, the telemetry wavelength is chosen according to the nature of the food, such that the reflectance properties of the telemetry wavelength do not depend on, or vary significantly from, the surface condition of the observed object. The term "significantly" means a variation in the reflectance properties of the telemetry wavelength that is less than 5% of the luminous intensity of the telemetry wavelength illuminating the object.

[0023] Preferably, the telemetry wavelength is chosen so that its intensity does not vary by more than 2%, 5%, or 10% after the food surface is covered by a parasitic fungus or has a surface defect as defined above. In other words, whether or not the parasitic fungus or surface defect is present on the food, the intensity of the telemetry wavelength reflected by said surface does not vary by more than 2%, 5%, or 10%.

[0024] By making these remarkable choices of wavelengths in steps a) and b), the intensity of the pixels composing the digital image, acquired in step a), is rectified or corrected from the distance measurements taken in step b), so that the variations in intensity of the wavelength observations on the surface of the observed object, due to the reliefs of said object, are corrected or compensated from the measurements taken during step b). In this way, a more precise and faithful digital image of the surface state of the observed object is obtained at the end of step c).

[0025] Thus, the rectification process according to the invention makes it possible to obtain a digital image that more faithfully reflects the chemical composition of the surface state of an object or food product when said object has a pronounced geometry. The term "geometry" refers to a non-planar surface. The process according to the invention therefore offers a digital image processing solution that is particularly well-suited and efficient for quickly and non-invasively identifying the quality of a food product, for example.

[0026] According to one embodiment, in step a), an analyzer is present in front of the photosensitive sensors of the optical detection device, the analyzer is capable of polarizing the light reflected by the object in at least four different directions, and at least one photosensitive sensor of the optical detection device is illuminated by light polarized by the analyzer at a single angle, and the Stokes parameters of the light reflected by the object are calculated from the rectified digital image as a result in step c).

[0027] In other words, the analyzer allows the optical detection device to simultaneously acquire four polarized images of the same object reflecting light in the direction of said optical detection device, each image being polarized in a single direction.

[0028] In other words, the Stokes parameters are calculated from a digital image that has been previously corrected in step c) of the rectification process according to the invention. By proceeding in this manner, it was surprisingly found that the calculated Stokes parameter values ​​are more accurate or more faithfully reflect the surface condition of the food, particularly at its edges visible in the digital image.

[0029] Advantageously, using the Stokes parameters calculated above, a person skilled in the art can then perform known digital processing on the image acquired by the optical detection device, in order to make primarily or solely the diffuse reflection of the observation wavelength at the surface of the food apparent. In other words, using the Stokes parameters, a person skilled in the art can process the digital image in order to The invention minimizes or subtracts from the image the light rays reflected from the surface of the food through specular reflection. The invention thus enables the acquisition of higher-quality digital images of food moving, for example, on a conveyor belt. From these higher-quality digital images, a person skilled in the art can naturally perform a more detailed and comprehensive analysis of the food moving on the conveyor, thereby enabling a more efficient subsequent food selection stage.

[0030] According to another advantage, the invention allows the calculation of Stokes parameters, from digital images of foods of complex shape and / or unknown nature, obtained during step c) of weighting of the rectification process according to the invention.

[0031] By "complex shape" we mean a food with a three-dimensional shape, delimited by one or more rounded or substantially rounded edges, such as an apple, a potato, a grain of rice or others.

[0032] In particular, the calculation of stock parameters makes it possible to detect the nature or change in composition of the surface of the food, regardless of the direction of incidence of the observation wavelength on the food.

[0033] Preferably, the observation wavelength is between 350 nm and 2300 nm or between 400 nm and 1900 nm.

[0034] Preferably, the rangefinding wavelength is between 500 nm and 1600 nm or between 500 nm and 1000 nm when modeling the surface geometry of the observed object. The term "modeling" here refers to measuring the surface geometry of the observed object.

[0035] In one embodiment, the observation wavelength is around 850 nm when the goal is to identify the ripeness of an apple's surface, for example. The telemetry wavelength is preferably around 750 nm when the goal is to identify surface features on the apple.

[0036] According to one embodiment, the observation wavelength is on the order of 1550 nm, when it is desired to identify the germination level on the surface of a potato, for example. The telemetry wavelength is of the order of 650 nm, when one wishes to identify the reliefs on the surface of the potato.

[0037] According to one embodiment, the observation wavelength is between 1300 nm and 1500 nm or between 550 nm and 800 nm, preferably between 800 nm and 900 nm, preferably in the order of 850 nm.

[0038] In one embodiment, the observation wavelength is approximately 750 nm, for example, when identifying the presence of bacteria on the surface of a tomato. The telemetry wavelength is approximately 532 nm, when identifying surface features on the tomato.

[0039] In one embodiment, the observation wavelength is approximately 1550 nm when identifying the presence of rice blast on the surface of a grain of rice, for example. The telemetry wavelength is approximately 900 nm when identifying the surface features of the rice grain.

[0040] In one embodiment, the observation wavelength is approximately 532 nm when identifying the presence of Penicillium digitatum on the surface of a lemon, for example. The telemetry wavelength is approximately 640 nm when identifying the surface texture of the lemon.

[0041] The examples mentioned above are not exhaustive. A person skilled in the art may choose other observation wavelengths depending on the condition or defects they wish to identify on an object or food.

[0042] It is clear from these different examples that the choice of observation wavelength depends on the food being observed in order to identify the state of ripeness of the food and / or the presence of a pathogen on the surface of said food.

[0043] According to another embodiment of the invention, the reflectance of the observation wavelength varies according to the state of organic decomposition of the object and / or the presence of one or more species of pathogens on the surface of said object, and the reflectance of the telemetry wavelength does not vary according to the state of organic decomposition of the object and / or the presence of one or more species of pathogens on the surface of said object.

[0044] According to another embodiment of the invention, in step c), the light intensity of a pixel constituting the digital image is modified proportionally to the measured distance between the photosensitive sensor associated with said pixel and the surface of the object illuminating said photosensitive sensor. Advantageously, by thus correcting the light intensity of the pixels constituting the digital image, the surface of the observed object is virtually flattened so that the reflectance properties of the observation wavelength depend little or not at all on the geometry of the surface of the object illuminated by said observation wavelength. This limits or eliminates artifacts in measurements made from the observation wavelength, intended to estimate the surface quality of an object, preferably a food item, artifacts due to the particular or irregular shape of the food.

[0045] According to another embodiment of the invention, in step c), the light intensity of a pixel constituting the digital image is modified proportionally to the value of the angle of inclination between a photosensitive sensor associated with said pixel and the surface of the object illuminating the photosensitive sensor. This variant of the invention allows for more precise correction of the brightness of the digital image so that it depends little or not at all on the geometry of the observed object.

[0046] According to another embodiment of the invention, in step c), the weighting of the digital image's light intensity also depends on the angle of incidence of the light coming from the object. This variant of the invention also allows for more precise correction of the digital image's brightness so that it depends as little as possible on the geometry of the observed object.

[0047] According to another embodiment of the invention, in step b), the distances between the surface of the object and the photosensitive sensors are measured using a laser beam. Preferably, the digital image is acquired over a length range excluding the wavelength range of highest intensity of the laser beam.

[0048] According to one embodiment, steps a) and b) of the process described above can be reversed or carried out simultaneously.

[0049] According to another embodiment, several processes identical to that of the invention can be implemented so as to allow rectification of a hyperspectral digital image. In other words, each method according to the invention is implemented by selecting a different observation wavelength, so as to be able to produce a hyperspectral digital image, each spectrum being associated with a specific observation wavelength or a single range of observation wavelengths. This embodiment can be particularly relevant when it is necessary to identify different surface states of the observed object or food.

[0050] Of course, the different characteristics, variants and forms of realization mentioned above can be associated with each other in various combinations, insofar as they are not incompatible or exclusive of each other.

[0051] The invention also relates to a hyperspectral digital imaging device of an object, comprising a device for acquiring a digital image of an object by said acquisition device, the acquisition device comprising several photosensitive sensors capable of forming a digital image of said object, and a telemetry measuring device for the distance between the object and several photosensitive sensors of the acquisition device, a control unit composed of a storage module and a calculation module, the control unit being connected to the digital image acquisition device as well as to the telemetry measuring device, the storage module comprising code instructions for the execution of a method for rectifying the brightness of a digital image as described above.

[0052] According to one embodiment, an analyzer is present in front of the photosensitive sensors of the optical detection device, the analyzer is capable of polarizing the light reflected by the object, in at least four different directions, and at least one photosensitive sensor is illuminated by light polarized by the analyzer at a single angle, and the storage module includes code instructions for executing a method for rectifying the brightness of a digital image as described above.

[0053] Preferably, the telemetry measurement device includes an optoelectronic system of the lidar type or similar. Description of the figures

[0054] The invention will be better understood from the following description, which relates to preferred embodiments, given by way of non-limiting examples, and explained with reference to the accompanying schematic drawings, in which:

[0055] [Fig. 1] illustrates a longitudinal section of a hyperspectral digital imaging device according to the invention, observing a potato from a first angle of observation;

[0056] [Fig. 2] illustrates a longitudinal section of a hyperspectral digital imaging device according to the invention, observing a potato from a second viewing angle;

[0057] [Fig. 3] illustrates a variant embodiment of the hyperspectral digital imaging device shown in Figure 1, comprising an analyzer placed in front of the photosensitive sensors of the optical detection device. Detailed description of the invention

[0058] As a reminder, the invention proposes a method for rectifying the brightness of digital images, making it possible to significantly reduce measurement artifacts related to the geometry of an object, in order to increase the quality and precision of a hyperspectral imaging process carried out from said digital images.

[0059] Figure 1 illustrates a hyperspectral digital imaging device 2 according to the invention. The hyperspectral digital imaging device comprises a spectrograph 4 consisting of an optical diffraction grating 6 positioned upstream of a digital image acquisition device 8. The digital image acquisition device 8 includes several photosensitive sensors arranged in an array. Each photosensitive sensor 10 is capable of measuring the variations in light intensity of a light spectrum from the optical diffraction grating 6.

[0060] The hyperspectral digital imaging device 2 also includes a rangefinding device 12, associated with the spectrograph 4, so as to be able to measure the distance between each photosensitive sensor 10 and the surface of an object positioned in front of the spectrograph 4. According to the present For example, the object is a food, more precisely a potato 14. Preferably, the telemetry measuring device 12 and the spectrograph 4 are attached to the same support 16 in order to facilitate distance measurements between the potato 14 and the photosensitive sensors.

[0061] The hyperspectral digital imaging device 2 also includes a control unit 18, consisting of a storage module 20 and a computing module 22. The control unit 18 is connected to the spectrograph 4 as well as to the telemetry measuring device 12. By the term "connected" is meant the possibility for the control unit to exchange information with the spectrograph 4 and the telemetry measuring device 12.

[0062] According to this example, the spectrograph 4 includes a photodetector marketed by the company Sony, under the reference SWIR IMX990. The photodetector is characterized by a detection range between 400 nm and 1750 nm.

[0063] The rangefinder 12 is also a device marketed by Sony, under the reference IMX556 "Time of Flight Sensor". The sensitivity of this sensor is within a wavelength range between 500 nm and 940 nm.

[0064] Remarkably, the memory module 20 includes code instructions for executing a process for rectifying the brightness of a digital image, acquired via the spectrograph 4 described above.

[0065] The rectification process according to the invention implements a first step of acquiring a digital image of an object positioned opposite the spectrograph 4.

[0066] The digital image consists of a multitude of light intensity measurements, taken by the photosensitive sensors 10, at an observation wavelength selected by the spectrograph. In this example, the observation wavelength is selected such that the reflectance of said observation wavelength, at the surface of the potato 14, depends on the chemical composition of the potato skin.

[0067] The term "depend" refers to the fact that the reflectance properties of the observed light wave vary significantly, so that the An optical detection device can detect these variations. According to this example, variations in light intensity are considered significant when the value of these variations is equal to or greater than 1% or greater than 5%.

[0068] According to a second step of the rectification process according to the invention, the rangefinder measuring device 12 measures the distance between the surface of the observed potato 14 and the photosensitive sensors 10 of the spectrograph 4. The distance measurements are carried out at a rangefinder wavelength, the reflectance of which does not vary significantly according to the chemical composition of the potato surface.

[0069] The term "significantly" means a variation in the intensity of the telemetry wavelength of less than 1% or less than 5%.

[0070] According to a third step of the rectification process, the light intensity of the pixels composing the digital image, acquired by the spectrograph 4 during the first step of the rectification process, is weighted according to the distances measured between the photosensitive sensors and the surface of the observed potato, distances measured by the telemetric measuring device 12 during the second step of the process described above.

[0071] By making these remarkable choices of observation wavelength and telemetry wavelength, during the first two steps of the process according to the invention, the intensity of the pixels composing the digital image acquired during the first step is rectified during the third step of the process, so that the variations in intensity of the observation wavelength on the surface of the observed potato, due to the reliefs of said potato, are corrected or compensated from the measurements made during the second step of the process.

[0072] At the end of the third stage of the rectification process, a more precise and accurate digital image of the observed potato's surface is obtained, particularly when the potato's surface is textured. In other words, the rectification process according to the invention significantly reduces measurement artifacts related to the potato's irregular surface.

[0073] It should be noted that the first and second steps can be reversed or carried out simultaneously.

[0074] The rectified image can then be analyzed using a known state-of-the-art hyperspectral imaging method to identify the chemical composition on the surface of the observed potato more precisely than with conventional hyperspectral imaging techniques.

[0075] The invention also relates to a method for calibrating the hyperspectral digital imaging device 2 described above. The calibration method implements a first step of acquiring a digital image of a flat-surfaced object, in this example a potato, positioned opposite the spectrograph 4 as illustrated in Figure 1, such that the surface of the potato is perpendicular to the optical axis 24 of the spectrograph 4. During the acquisition of the digital image, the object is illuminated by a light beam at an observation wavelength according to the invention.

[0076] In a second step illustrated in Figure 2, the potato 14 is tilted at a known angle relative to the optical axis 24 of the spectrograph. Without changing the observation parameters of the spectrograph 4, a second digital image of the potato is acquired by the spectrograph.

[0077] In a third step, a correlation is established between the variation in light intensity of the digital images acquired by the spectrograph 4, as a function of the variation in the angle of inclination of the potato with respect to the optical axis 24 of the spectrograph.

[0078] The three steps described above can be reproduced for different angles of inclination between the potato 14 and the optical axis 24 of the spectrograph, so as to obtain a set of correlations between the light intensity measured by the photosensitive sensors of the spectrograph and the inclination of the surface of the object with respect to the spectrograph.

[0079] These correlations are subsequently recorded in the memory module of the hyperspectral digital imaging device 2 according to the invention, so as to allow a more precise rectification or correction of the brightness of a digital image during the third step of the rectification process described above.

[0080] According to one alternative embodiment, the calibration process can also be implemented by modifying the distance between the potato 14 and the spectrograph 4, in order to allow for finer rectification of the brightness of a digital image according to the invention depending on its position relative to the spectrograph.

[0081] As a non-limiting example, Table 1 below offers wavelength choices, depending on the object being observed and the surface condition to be detected:

[0082] [Tables 1]

[0083] According to another embodiment of the invention, illustrated by Figure 3, an analyzer 26 is positioned in front of the photosensitive sensors 10 of the digital image acquisition device 8 which is visible in Figure 1. Preferably, the analyzer 26 is interposed between the optical diffraction grating 6 and the photosensitive sensors 10.

[0084] The analyzer 26 is capable of polarizing the light reflected by the potato 14, along at least four different directions, and at least one photosensitive sensor 10 is illuminated by light polarized by the analyzer 26, along a single angle of polarization.

[0085] By way of non-limiting example, the four polarization directions mentioned above may be separated by an angle between 30° and 50°, preferably on the order of 45°

[0086] The spectrograph 4 illustrated in Figure 3 can, for example, be marketed by the company SONY, under the reference "Pregieus IMX253MZR".

[0087] The invention also relates to a calibration method for the hyperspectral digital imaging device 2, as illustrated in Figure 3, reproducing the calibration method described above.

[0088] This calibration method differs from the previous one in that, during the first stage of digital image acquisition, the Stokes parameter values ​​are calculated from the acquired digital image of the potato, for a target area of ​​the potato's surface. The target area corresponds to a surface of the potato that intersects the optical axis of the digital image acquisition device 8 and forms a known angle with said axis, in this case an angle of 90°.

[0089] The values ​​of the Stokes parameters thus calculated are recorded in a database, so as to be correlated to the value of the angle formed between the optical axis of the digital image acquisition device 8 and the surface of the potato 14.

[0090] The second and third steps described above are subsequently implemented in the same way.

[0091] This calibration process allows for the creation of databases specific to each type of food. These databases can then be used as reference values ​​to assess the quality and / or conformity of a food product.

Claims

DEMANDS

1. A method for rectifying the brightness of a digital image, characterized in that the following steps are implemented: a) acquiring a digital image of a food (14) by means of an optical detection device comprising photosensitive sensors (10), the digital image is formed from measurements of variations in light intensity, made by the photosensitive sensors (10), at an observation wavelength whose reflectance varies according to the state of decomposition and / or maturation of the surface of said food (14); b) measuring the distance between the surface of the food (14) and the photosensitive sensors (10) of the optical detection device, the distance measurements are made at a telemetry wavelength whose reflectance does not depend on the state of decomposition and / or maturation of the surface of said food (14);c) weighting of the light intensity of the image acquired by the optical detection device during step a), as a function of the distances measured during step b).;

2. A method for rectifying the brightness of a digital image according to claim 1, characterized in that the observation wavelength is chosen according to the nature of the food (14), so that the intensity of said wavelength decreases by more than 10%, when the surface of the food is covered by a fungus or when the surface of the food has an appearance defect.

3. Method for rectifying the brightness of a digital image according to claim 1 or 2, characterized in that the telemetry wavelength is chosen according to the nature of the food (14), so that its intensity does not vary by more than 2%, when the surface of the food is covered by a parasitic fungus or when the surface of the food has an appearance defect.

4. A method for rectifying the brightness of a digital image according to any one of claims 1 to 3, characterized in that during step a), an analyzer (26) is present in front of the photosensitive sensors (10) of the optical detection device, the analyzer is capable of polarizing the light reflected by the food (14) in at least four different directions, and at least one photosensitive sensor is illuminated by light polarized by the analyzer at a single angle, and in that the Stokes parameters of the light reflected by the food (14) are calculated from the rectified digital image following step c).

5. Method for rectifying the brightness of a digital image according to any one of claims 1 to 4, characterized in that during steps c), the light intensity of a pixel constituting the digital image is modified proportionally to the distance measured between the photosensitive sensor (10) associated with said pixel and the surface of the food (14) illuminating said photosensitive sensor.

6. Method for rectifying the brightness of a digital image according to claim 5, characterized in that during step c), the light intensity of a pixel constituting the digital image is modified proportionally to the value of the angle of inclination between a photosensitive sensor (10) associated with said pixel and the surface of the food (14) illuminating the photosensitive sensor, the angle of inclination being calculated from the distance values ​​achieved during step b).

7. Method for rectifying the brightness of a digital image according to any one of claims 1 to 6, characterized in that during step c), the weighting of the light intensity of the digital image also depends on the angle of incidence of the light coming from the food (14).

8. A method for rectifying the brightness of a digital image according to any one of claims 1 to 7, characterized in that during In step b), the distances between the surface of the food (14) and the photosensitive sensors (10) are measured using a laser beam.

9. A method for rectifying the brightness of a digital image according to any one of claims 1 to 8, characterized in that the observation wavelength is between 1300 nm and 1500 nm or between 550 nm and 800 nm.

10. A method for rectifying the brightness of a digital image according to any one of claims 1 to 9, characterized in that the telemetry wavelength is between 800 nm and 900 nm, preferably in the order of 850 nm.

11. A hyperspectral digital imaging device for a food (14), comprising a device for acquiring a digital image (8) of a food (14) by said acquisition device, the acquisition device comprising several photosensitive sensors (10) capable of forming a digital image of said food (14), and a telemetry measuring device (12) for the distance between the food (14) and several photosensitive sensors (10) of the acquisition device, a control unit (18) comprising a storage module (20) and a computing module (22), the control unit being connected to the acquisition device as well as to the measuring device, characterized in that the storage module (20) comprises code instructions for executing a method for rectifying the brightness of a digital image according to any one of claims 1 to 3.

12. A hyperspectral digital imaging device according to claim 11, characterized in that an analyzer (26) is present in front of the photosensitive sensors (10) of the optical detection device, the analyzer is capable of polarizing the light reflected by the food (14) in at least four different directions, and at least one photosensitive sensor is illuminated by light polarized by the analyzer at a single angle, and in that the storage module (20) comprises code instructions for the execution of a method for rectifying the brightness of a digital image according to any one of claims 4 to 10.

13. Hyperspectral digital imaging device according to claim 11 or 12, characterized in that the telemetry measuring device (12) comprises a lidar-type optoelectronic system.

Images