Description

Technical Field

The present invention finds application in the technical field of measurement systems and has particularly as object a system and a process for the recognition, characterization and classification of foods and nutrients present in foodstuffs.

State of the art

As is known, the term "nutraceuticals" refers to particular substances contained in certain foods, normally derived from plants or from microbial sources, having medicinal properties or however beneficial on the health of the human or animal body. Examples of nutraceuticals are polyphenols, tocols, carotenoids, prebiotics, polyunsaturated fatty acids (omega-3, omega-6), vitamins, some enzymatic complexes.

Nutraceuticals can be taken either as a "naturally nutraceutical food" or "enriched food" of a specific active ingredient, or in the form of food supplements in liquid formulations, tablets or capsules.

Another category of food is instead that of the so-called "functional foods", ie foods that demonstrate satisfactorily to have positive effects on one or more specific functions of the organism, which go beyond the normal nutritional effects, so that it is relevant for improving health and well-being and / or reducing the risk of illness. Examples of functional foods are foods that contain certain minerals, vitamins, fatty acids or dietary fibers and those added with biologically active substances, such as active ingredients of plant origin or other antioxidants and probiotics that have live cultures with beneficial properties.

However, the industrial processes of production and packaging of food tends to cancel the presence of nutraceuticals, so the consumer can never be certain of the actual presence of these elements inside the food, even in the case where one would expect to find them given the nature of some foods.

In fact, the consumer cannot discriminate the presence of the nutraceutical or functional components present in the food at the time of purchase as European legislation prohibits the labeling of any food to prevent, treat or cure a human disease or to refer to these properties.

The consumer cannot even assess whether the food being consumed still contains the specific substances present at the time of production and marketing among those of a nutraceutical or functional nature for the treatment or prevention of food-related diseases.

Last but not least, the consumer is not able to assess whether all the components indicated by the manufacturer on the label and /or in the information disclosed to the market are really present in the consumed food or have been only indicated in an ambiguous, elusive and/or malicious way from producers and sellers.

Currently, systems that allow the analysis and classification of foods and their respective nutraceutical substances are exclusively systems bom for laboratory use and not consumer-oriented.

In particular, molecular sensors and analyzers are used, whose operating technology is based on near-infrared spectroscopy, both NIR (Near-Infra-Red), where they operate with waves from 700 to lOOOnm, and SWIR (Short-Wave -Infra-Red) wherein they operate with waves from l000-2500nm.

All existing products represent laboratory equipment, used to carry out qualitative and quantitative tests on the chemical composition of foods functional to types of applications such as quality control in the industrial processes of agri-food companies and for the detection of adulterations and counterfeit foods, increasing the quality of production thanks to a closer control, improvement of the efficiency of the laboratory and production, prevention of out- specification, verification of the final product in accordance with the guidelines.

These solutions therefore have obvious limitations as they limit the analysis only at the laboratory scale, do not allow to relate information from multiple sources and compare them in a single archive, they lack personalized support services that can guide users towards virtuous paths of nutrition, wellness and health.

Furthermore, such solutions are affected by poor selectivity and need to proceed to a complex modeling of the samples.

Moreover, with the reduction of the spectrometer’ s size, the levels of precision of the analysis are reduced, as it can also require additional or secondary methods to increase the sensitivity.

Scope of the invention

Object of the present invention is to overcome the aforementioned drawbacks by providing a system and a process for the recognition, characterization and classification of foods and nutrients present in foods which exhibit characteristics of high efficiency and relative cost-effectiveness which in particular can be easily used by the food consumer.

A particular object is to provide a system and a process for the recognition, characterization and classification of food and nutrients present in food that allows to recognize exactly the food at the time of consumption and, consequently, identify the presence of ingredients and substances with nutraceutical qualities present in the food being consumed as they are present in the raw materials, in the ingredients and in the final product of the transformation phase in food.

A particular object is to provide a system and a process for the recognition, characterization and classification of food and nutrients present in foodstuffs that can be implemented by means of compact and portable instrumentation and in particular by means of a spectrometer having such a size so that it may be integrated into a smartphone or other commonly used mobile device, without compromising the reliability of the analysis.

Another object is to provide a system and a process for the recognition, characterization and classification of food and nutrients present in foodstuffs that allows to create and constantly update a food database classified according to the actual content of nutraceutical substances thereof.

In this way, the consumer will be able to deepen his knowledge on the quality and the salutary value of the products present in the database established under this procedure and visible through the tool object of the patent among the daily products on our tables, such as wheat flour an illustrative but not exhaustive title, through the characterization of foods, ingredients and components by means of portable compact spectrometry of dimensions compatible with the inclusion in smartphones, combined with a database of foods, ingredients and components obtained with the spectrometry of laboratory and with classical chemical and biological laboratory analyzes.

Advantageous embodiments of the invention are obtained according to the dependent claims.

Brief disclosure of the drawings

Further features and advantages of the invention will become more apparent in the light of the detailed description of some preferred but not exclusive embodiments of the process and of the system according to the invention, illustrated as a non-limiting example with the aid of the attached drawing table wherein:

Fig. 1 is an example of average values of the spectra, in the wavelength range (1000- 2500) nm, obtained from the hypercubes associated with the ROIs of different samples of flours;

Fig. 2 is an example of pre-processed spectra, in the wavelength range (1000-2500) nm, in order to better highlight the spectral attributes associated with the ROIs of different flour samples;

Fig. 3 is an example of results related to the PCA analysis in order to evaluate the degree of discrimination of the samples in relation to their intrinsic features and to the detected hyper- spectral attributes;

Fig. 4 is an example of the loadings values for the PC1 and PC2 components referred to the Score Plot of Fig. 3;

Fig. 5 is an example of selection of wavelength intervals obtained by applying the forward-type iPLS-DA analysis, wherein the y-axis represents the RMSECV values and the x-axis represents the wavelengths;

Fig. 6 shows a flowchart of the model construction and classification procedure.

Best modes of carrying out the invention

With reference to the attached figures, some methods of execution of the method according to the invention and of the system for its implementation are illustrated.

The process will essentially comprise a starting step i) phase of chemical/biological food analysis to identify the presence of nutraceutical substances in them.

Subsequently a step a) will be carried out to obtain a spectral measurement with laboratory instrumentation characterized by another degree of precision, both in terms of spatial and spectral resolution, of the same foods previously analyzed.

In particular, the spectral measurements will be obtained by means of one or more precision NIR + SWIR spectrometers and will be carried out directly on food prepared according to the modalities with which they are placed on the market, as well as on raw or cooked, hot or cold, whole or seasoned variants in mix with third parties or with spices or condiments and on the basic ingredients and raw materials from which the same food is derived.

The spectral measurements thus obtained will then be processed (step b) by means of laboratory analyses in the range from 700nm to 2500nm with the PCA method (Principal Components Analysis) and Interval PLSDA analysis in order to identify the characteristic wavelengths of each food.

Finally, the consumer can proceed (step c) to a spectral measure by reflection through portable compact instrumentation, characterized by a lower spatial and spectral resolution, within the spectral areas defined in the range from 700nm to 2500nm following the aforementioned PCA and Interval PLSDA analysis, of the same foods previously analyzed.

The compact instrumentation will process the discrete spectral signatures obtained for these intervals in order to compare and identify the foods based on the spectral characteristics obtained with the reference database defined starting from the determinations carried out at the laboratory scale.

The starting step of chemical/biological analysis may be carried out according to different methods and does not represent a limitation of the scope of protection of the present invention.

In an exemplary way, for the extraction of antioxidant substances, the literature reports different methods that mainly use two types of extraction solutions: acetone/water alone (Pellegrini et al., 2003 and 2006, and Wang et al. , 1996) or in combination (George et al., 2005) or ethanol/water (Ismail et al 2004, Goupy et al., 1996), followed by spectrophotometric or chromatographic assays.

With regard to the extraction of soluble phenols from cereals, aqueous solutions of methanol, ethanol and acetone can be used (Serpen et al., 2008, Liu et al., 2007), while the insoluble component is obtained from the residue of the previous extraction by alkaline hydrolysis (with sodium hydroxide), pH adjustment with a weak acid and a series of extractions with ethyl acetate (Adorn et al., 2003 and 2005, Adorn and Liu 2008).

As far as carotenoids and tocoli are concerned, they are generally extracted with a single-phase method that involves the use of a miscible organic solvent, or in a double phase with an alcohol and an organic solvent which can be immiscible with water, or even by saponification (Lang et ah, 1992).

The extraction of biologically active substances may also be carried out using validated or to be validated techniques, not mentioned here but in use in the reference laboratory provided they are compatible with both HPLC (High Performance Liquid Chromatography) and FTIR (Infrared absorption Spectroscopy), according to the laboratory where the latter is in use.

Polyphenols can also be quantified with the colorimetric assay of Folin-Ciocalteu, used on different food matrices, but it seems to have limitations with regard to the lipophilic component (Huang et al., 2005).

The step a) of obtaining the spectral measurements of the food may provide already used and/or known techniques of laboratory spectrometric analysis.

In an exemplifying and non-limiting manner, the spectrometric analysis techniques of food in the laboratory may be of three types:

- measures with the signal reflectance method are applicable for solid and powder foods;

- measures with the signal transmission method are applicable for liquid and creamy foods;

- for opaque liquid food and "thin film" samples the signal transmittance method may be applied.

For example, for grains, laboratory spectrometer analysis techniques can determine moisture, proteins, oil, fiber, starch, density, while for cheeses it will be possible to determine moisture, fat and protein. For milk, fats, proteins, lactose and dry matter were evaluated, also in comparison with destructive laboratory analysis, typical of step i).

In general, the techniques that are limited only to phase a) of the laboratory allow the verification of a few samples and under standard conditions and therefore different from those of normal consumption and use.

The step b) of the innovative process involves the processing of the spectral measurements carried out in the previous steps by means of the PCA (Principal Components Analysis) method and the "Interval PLSDA" analysis. In particular, the spectral signatures are processed by applying the PCA analysis for a first exploratory verification of the significant wavelengths of the food.

Subsequently, through a first PLS-DA analysis and an iPLS-DA analysis, then, for each food, a reduced set of wavelength ranges that characterize the food is identified.

In summary, the samples will be acquired, through the laboratory spectrometer and then proceed to the selection of the regions of interest (ROI) within the hyperspectral images acquired and to the subsequent mosaicization of the selected images (ROI) for the generation of the sample of training, after pretreatment of the spectra.

At this point, the PCA analysis of the data and the subsequent construction of the iPLS- DA model will be performed.

Studies carried out by the inventors have already shown that results of good classification of some grains treated with this analysis, other than those that determine the flours and pastas subject of this procedure, are obtained by limiting the observations in the wavelength ranges 1209 -l230nm, l489-l5l0nm and 1601- l622nm.

The execution of the above phases will allow to create a database with the spectral measurements of the analyzed foods, so that they can be used for the subsequent comparison with the measurements obtained by the consumer through the spectral measurement by reflection through portable compact instrumentation of a single food. This further step will use a portable spectrometer operating in limited intervals of the NIR + SWIR spectrum.

The compact portable instrument, object of the present invention, will be suitably a reflecting instrument, which therefore allows to analyze the food from a distance of a few centimeters, for example about 10 cm, with the aid of an illuminator which records the response of the food in three or four limited ranges of spectral bands.

In this way, the instrument will be able to compare the response with the values in the same spectral bands of the database of laboratory spectral signatures and will be able to establish a correlation between the discrete spectrum analyzed at the time of consumption of the food with the spectral characteristics of the laboratory present in the database to determine the type of food and the presence of nutraceutical substances. In this step, the spectral measurement by reflection of the food is carried out with portable compact instrumentation, characterized by a lower spatial and spectral resolution, and by a measurement in specific spectral areas in the range from 700 to 2500 nm identified following the analysis PCA for analysis with Interval PLSDA.

The process allows to associate the spectral attributes of food acquired through the use of compact portable spectrometers with a level of precision lower than those of laboratory the corresponding spectrum among those acquired with a laboratory spectrometer, and present in the database available, and to evaluate the components of the corresponding food or ingredient acquired with traditional chemical, physical, and biological laboratory analysis techniques.

In this way, by comparison and subtraction and through the date of the spectra of the nutraceutical substances of interest analyzed by the portable and laboratory spectrometer in samples in which they are isolated or in composition to be considered almost pure, it is possible to associate at the moment of the consumption the authenticity of the product, the integrity of the conservation and the permanence of the typical substances of the same present at the moment of its production.

Therefore, the data directly collected by the users through the use of the compact tool will be overlapped and / or crossed with those that are part of the laboratory database. As a consequence, all the factors linked to the lower precision that characterizes the measurements made with the compact instrumentation supplied to the consumer will be compensated thanks to the procedure of crossing / integration with the database, containing precise and precise information obtained through the analyzes, and subsequent data processing and modeling processes, carried out with very precise laboratory tools.

The compact instrumentation will consist of a compact portable spectrometer operating in reflectance and equipped with an autonomous illuminator, operating in the near infrared both NIR (Near-Infra_Red) (700-1000 nm) and SWIR (Short-Wave- Infra-Red) (1000 -2500 nm).

The dimensions and the constructive solution of the spectrometer will be such as to be inserted inside a smartphone or other mobile device, for example a tablet, or directly applied on the same as a consumer-oriented series accessory component.

The compact spectrometer will be calibrated for use for food at the time of consumption, which are also raw or cooked, hot or cold, whole or seasoned in mix with third parties or with spices or seasonings. The compact spectrometer will operate on a discrete series of wavelengths in the NIR + SWIR interval as a function of the signal peaks deriving from the PCA analysis and the analysis with "Interval PLSDA.

The system will also include the implementation of an application to compare the discrete spectral response recorded by the compact spectrometer and the database of laboratory spectral signatures and to calculate the quantities of food consumed by recognizing the portions taken.

The application will allow comparing the value of the spectrum received by the compact spectrometer in the operating wavelength ranges of the reflected signal receptor with the values of the laboratory continuous spectra, generating an appropriate response.

In particular, the application will inform if the answer is comparable with those of the laboratory and therefore if the food is identifiable with those acquired in the laboratory, or if the answer is not comparable with those of the laboratory and therefore if the food is not identifiable with those acquired in the laboratory.

The application will also allow you to compare the code acquired in the label by reading the barcode and / or QR code with the classification codes of the laboratory spectra from the label of the consumed food recognized by the sensor, also indicating in this case if the code is present in the food database analyzed in the laboratory and therefore if the food is identifiable with those acquired in the laboratory, or if the code is not present in the food database analyzed in the laboratory and therefore if the food is not identifiable with those acquired in the laboratory.

Last but not least, the application can compare the image of the food that the compact spectrometer acquires during the consumption phase with the library of values of dimensions and weights contained in the typical containers of consumption, based on already available image treatment applications. , indicating if the container is present in the database of food containers in the library and if the level of filling of the container is comparable with one of the levels present in the library and with one among the quantities of food associated with container and level, or if the container is not present in the database of food containers in the library and / or if the filling level of the container is not comparable with one of the levels present in the library and / or with one of the quantities of food associated with container and level. The application, on the basis of the above comparisons, can provide indications regarding the recognition of the food at the time of consumption with one of those present in the food database analyzed in the laboratory both with spectrometry and with chemical-physical-biological analyzes -nutritional, determining that the food has arrived intact at the time of consumption and that it contains the nutraceutical components isolated from laboratory analysis.

Furthermore, it may indicate the quantity consumed by the user and therefore the nutritional and / or nutraceutical components acquired during consumption, in order to draw up a statistic of the user's consumption of the application.

In cases of failure to fully recognize the food with those in the laboratory database, even if the classification code obtained from the label is present, the system will indicate that the food has not arrived intact at the time of consumption and therefore may not contain the nutraceutical components isolated from laboratory analysis, even if the presence of the signal peaks associated with some of the ingredients / components present in the database indicate that with a certain margin of error the food contains the nutraceutical components isolated from the laboratory analysis.

Also in this case the quantity consumed by the user and therefore the nutritional and / or nutraceutical components acquired during consumption will be indicated.

In case of non-recognition of the food with those present in the laboratory database and the simultaneous absence of classification code obtained from the label, it will be possible to receive information that the food has not been analyzed in the laboratory, but in any case approaches other foods present in the database for the extent of the response in the intervals acquired by the compact spectrometer, indicating to the user the type of food that is nearer consuming and requiring confirmation to the user based on his knowledge and / or the label of the food.

In case of confirmation of the user of the suggestion proposed by the application, the application will indicate that the food has arrived intact at the time of consumption, but may not contain the nutraceutical components isolated from laboratory analysis. Any recognition of the signal peaks associated with some of the ingredients / components present in the database may indicate that, with a certain margin of error, the food contains the nutraceutical components isolated from the laboratory analyzes; A further case could be that in which there was a lack of recognition of the food with those present in the laboratory database and the absence of classification code obtained from the label, for which the food was not analyzed in the laboratory, and in which the food does not approach other foods in the database because of the magnitude of the response in the intervals acquired by the compact spectrometer.

Yet another possibility is the one wherein there is no recognition of the food with those present in the laboratory database and the absence of classification code obtained from the label, and also the absence of recognition of the signal peaks associated with some of the ingredients / components present in the database, for which the food does not contain any of the nutraceutical components isolated from laboratory analysis.

In this case, the application will allow you to insert one of the standard foods in the library and report to the system the presence of a food not present in the database, but present in the library, to activate the insertion of the same in the laboratory database. A further possibility is the one wherein it is not possible to insert one of the standard foods present in the application library. In this case, the system will be notified to activate the insertion of both the library and the laboratory database.

The compact instrument will also be able to detect the presence of a spectral signature of pollutants and/or elements of deterioration and/or contamination and/or food degeneration through each of the above mentioned cases by comparison with the spectral signatures present in the laboratory database for polluting foodstuffs and/or for elements of deterioration and/or for contaminants and/or for states of degeneration thereof.

In this way the system may inform the user if the food reported is a biological food or in any case free from the consumption of these negative components.

Below is an example of application of the procedure to the case of the grain, flour and pasta supply chain.

The procedure provides the acquisition of samples for the laboratory spectrometer. In particular, after the calibration, the acquisition is made, through the use of a hyper- spectral camera, adapted to acquire information in the range 1000-2500 nm with a spatial resolution of 150 pm/pixel, of an image field of 5x5 cm. The acquisition speed is equal to 18 mm/sec. The lighting system consists of No. 25 halogen lamps (12V- 10W-50 ⁰ - Paulman_G4). For each pixel a spectrum consisting of a discrete number of bands in the range 1000-2500 nm is then available, with a band gap of 6.7 nm.

This generates a hypercube of information that is representative of the surface characteristics of the sample (Fig. 1).

Hyperspectral analysis (HIS) allows the characterization, classification and quality control for all foods.

The HIS analysis is based on the use of integrated hardware and software able to digitally capture and process the spectra, such as image sequences in each of which each column represents the discrete value of the spectra of the corresponding unit element taken in the grid of the two-dimensional scanning of the sample, so as to create a three-dimensional hypercube characterized by the two spatial dimensions of movement of the sensor with respect to the sample and the vertical dimension of the spectrum.

Each unit of the sample generates a pixel and can therefore be analyzed in a non destructive way, determining an accurate and detailed extraction of the information generated by absorption of waves sent to the sample in the NIR + SWIR frequency and/or by reflection thereof.

Based on the different wavelengths of the source and the spectral sensitivity of the sample, different physical and chemical characteristics of the food can therefore be investigated. The calibration of the instrument is performed in the laboratory by recording two images, one for a white reference and one for a black reference, and with the application of a specific corrective formula for the original intensity of the hyperspectral image acquired for each pixel.

Starting from the image acquired, one of his region of interest (ROI) is sampled, in order to maximize the area useful for the survey (for example pixels belonging to the flour). It then operates a pre-processing of the information (i.e spectra constituting the hypercube associated with the ROI) through the application of the algorithm of Mean Centering (MC) (Fig. 2).

The PCA analysis is then applied to the evaluation of the Score Plots (PC1-PC2-PC3) for the evaluation of possible non-homogeneities of the sample, due for example to errors during the preparation phase, of possible errors during acquisition, due, for example, to an incorrect setting up of the lighting system or to the scanning speed

(Figs. 3 and 4).

This process is repeated for each of the samples object of the investigation aimed at their recognition, characterization and classification.

Through the use of a specific sequence of pre-treatment of the spectra and subsequent PC A analysis, the main components that will be taken as a reference for the construction of the next food recognition / classification model are identified and defined.

The samples analyzed within the acquired hyperspectral images are then combined to form a single data set: mosaic image (i.e. hyperspectral images of the ROIs and associated hypercubes) that are used for the generation of the classification model. The new data set generated in this way is subjected to a pre-processing sequence in order to better highlight the regions to which each belongs within the Score Plots (PC 1 - PC2-PC3) resulting from the application of the PCA.

The pre-processing used are usually: lst Derivative and Standard Normal Variate (SNV), both in sequence and individually. In the end, "data centering" is always performed through the Mean Centering procedure (MC). The training samples are selected following two paths, in relation to the characteristics of the sample in terms of hyperspectral attributes, ie through the definition of a new series of ROIs, each series being defined starting from an area of the mosaic image representative of a certain sample, or through the selection of domains/regions within the Score Plot, resulting from the PCA analysis of the mosaic image.

PCA analysis cannot be used to construct predictive models, such as to allow the classification of samples in a given category of food.

Then, from the sample defined using the pre-processing carried out according to the methods described above, we proceed to the construction of the PLS-DA model (Partial Least Square Discriminant Analysis), used to find a model able to perform optimal discrimination between the categories of samples and to predict the images taken by the compact sensor used by the consumer.

During the model generation, in order to evaluate the robustness of the hyperspectral information of the training sample, Cross-Validation procedures are applied, based on the application of the Contiguous Block algorithm. Since the spectral data resulting from the sample have a superabundance of information as they are continuous in the wavelength generated by the illuminator and picked up by the sensor (for example from 6700 to 2,500 nm for NIR + SWIR analysis), the iPLS analysis -DA performs an exclusion of wavelength intervals to facilitate and speed up the classification of the food, operating for successive interactions in "forward" mode, in which some intervals of wavelengths are added from time to time, or in "reverse" mode, in which some intervals are removed from time to time, in order to identify the intervals and / or the individual wavelengths significant to characterize the food group.

The iPLS-DA analysis works by dividing the total spectrum into equal width intervals and calculating the classification models for each of these spectral micro-regions. In this way the best wavelength range is selected as the one that determines the minimum value of the square root of the cross-validation error (RMSECV).

This range then becomes the reference range for the scope of operation of the compact spectrometer, thus eliminating the emission and recording of signals in the other wavelengths, in order to optimize the size of the hardware itself.

The defined PLS-DA model is then validated using a new sample to which the same pre-processing used in the training phase are applied.

The quality of the results is evaluated through the analysis of the confusion matrix reported in Table 1

Predicted as White Flour Percia Sacchi 26073 52 7106 131 Predicted as Flour Percia Sacchi 541 32392 181 1419 Predicted as Flour Tumminia 7559 1 27840 809 Predicted as Organic Wheat Flour \ 1107 988 27232 Table 1

The wavelengths relevant for the food are defined with the iPLS-DA technique as in Fig. 5 and in this way the type of compact sensor for the consumer is identified, which can be miniaturized until it is inserted or directly applied to smartphones.

Contrary to standard methods that have similar characteristics of the reflectance between similar foods (for example whole wheat flour made from different types of grains and cereals) both in the visible field and in the NIR + SWIR field, the present technique allows to characterize different flours and pastes both raw that cooked and / or seasoned using 3 or 4 wavelength intervals each consisting of 4 wavelengths, in order to reduce the size and response times for the sensors and the software, for a quick quality control and correlation of the spectrum taken from the compact instrument with respect to the laboratory spectra.

The technique is innovative because, starting from a data set to be classified, the mosaic technique is applied starting from which, using the same processing logic of the data used for the construction of the model, the recognition / classification of the food is performed.

At this point the samples are taken for the compact portable spectrometer supplied to the consumer.

In general, the iPLS-DA analysis of the laboratory spectrum for the food marketed identifies the significant spectral ranges for the recognition of the food through the compact portable spectrometer and determines what the wavelengths of the emission spectra will have to be. and of the sensor for receiving the reflection spectrum.

Through the application, if the food is identified from the barcode or QR code during consumption, the portable compact sensor operates on these intervals lower than the entire spectral range 700-2500 nm.

As an exhaustive but not exclusive example, for the flours indicated in Fig. 5, the acquisition is made through the use of a compact portable spectrometer, able to acquire information at least in the range 1000-1700 nm.

The lighting system consists of a tungsten filament source integrated within the spectrometer.

In general, the compact spectrometer could be even more specialized in a series of 3 or 4 intervals in the range 1200-1700 nm and the illuminator could be composed of LEDs with reflecting shell composed of alloys such as to generate the majority of lighting intensity in such intervals.

For each pixel a spectrum consisting of a discrete number of bands in the range 1000- 1700 nm is then available. This spectrum is taken as representative of the flour sample. It then proceeds to its recognition / classification using the model developed in the laboratory starting from the database produced through the use of bands made up of laboratory spectra in the range of 1000-2500 nm, the latter acquired with a bandwidth equal to 6, 7 nm. Model input is the spectrum acquired by the portable spectrometer. The application of the compact instrument is able to perform an iteration of the iPLS- DA method to the spectrum acquired in the bands identified in the iPLS-DA analysis of the laboratory spectrum, for example the green ones in Fig. 6, and in this way defines even better the differences in shape, intensity and color between the different foods similar to the naked eye (for example the sack flour and the tuminia flour). Through this new iteration we define the values of the parameters of the set that identifies the food of the laboratory database, we associate the consumed food analyzed by the compact portable sensor supplied to the consumer with the corresponding food among those analyzed leaving the producers in the laboratory, and the mix of components and / or nutraceutical ingredients determined by the chemical- physical-biological-nutritional analysis performed on the product food is correlated to the consumed food.

The procedure thus generates a recognition / assignment path, by the software of the application supplied with the compact portable spectrometer, of the information representative of the surface characteristics of the flour and with it a distinction is made, by comparison, of the flour analyzed by the consumer in order to recognize its characteristics, based on the results provided by the classification model, starting from the reference database, so as to be able to associate it with the nutraceutical components that it presents.

In the event of discrepancies consisting of significant deviations from those of reference and/or evident macroscopic differences (such as color or grain size), it is anyway reported to the consumer, always through the application of the portable spectrometer, the belonging of the flour and/or the ingredient/food analyzed to a certain class of product/s, but also that the flour and/or the ingredients/foods could contain added substances and/or show some nutritional properties, suggesting contacting the subject (s) from which the consumer purchased / received the product for clarification.

Fig. 6 shows an exemplary flowchart of the model construction and classification procedure.

This process will comprise the following steps:

1. Acquisition of raw data related to reference samples of different forms. Generation of the hypercube. 2. Construction of the reference data set, that is, of all the samples, individuals or families, each representative of a certain type of flour.

3. Pre-processing of the acquired information (i.e spectra constituting the hypercube) and generation of a new data set (i.e elaborated spectra) of reference.

4. Analysis in principal components (PC A) in order to verify the significance level of the spectral data referred to each sample, or family of samples, and of the effect of the pre-processing in order to operate the discrimination of the samples, or of the families of samples.

5. Evaluation of the significance of the data acquired for the purposes of discrimination of samples, or families of samples.

6. Construction of reference spectral libraries.

7. Generation, starting from the information acquired in (1) of a training data set for the generation of the classification model (9).

8. Generation, starting from the information acquired in (1) of a data set for the validation of the classification model (9).

9. Application of the PLS -DA procedure for the generation of the classification model.

10. Identification of significant wavelengths and their extrapolation.

11. Generation of a training data set using "only" significant wavelengths as resulting from the application of the procedure (10) to be used to generate the new PLS-DA classification model (14).

12. Generation of a verification data set using the "only" significant wavelengths as resulting from the application of the procedure (10) to be used to validate the new PLS- DA classification model (14).

13. Application of the PLS -DA procedure for the generation of the classification model operating on a reduced number of wavelength intervals.

According to a further aspect of the invention, the spectrometer and / or the camera supplied with the smartphone also detects the dimensions and the container on which the food is consumed and, through the correlation library between images and quantities connected with the optical recognition of the portion of food, the amount of food consumed is determined with adequate precision and, consequently, the quantity of the components and / or nutritional and / or nutraceutical nutrients consumed with it. The system and the process according to the invention are susceptible of numerous modifications and variations, all of which are within the inventive concept expressed in the appended claims. All the details may be replaced by other technically equivalent elements, and the materials may be different according to requirements, without departing from the scope of protection of the present invention.