RU2638910C1 - Method of object express control - Google Patents

Abstract

FIELD: measuring equipment.

SUBSTANCE: identified object and calibration template are lighten up by spectral narrow-band radiation flows non-overlapping spectral channels, fitted electrical signals and signals of identified object are formed, calibration of received signals is implemented, electrical signals are converted into spatiospectral graphic signals, and spectral and spatiospectral images of the identified object are formed. While measurement is performed in the reflected or transmitted, polarized and non-polarized light in the ultraviolet, visible and infrared ranges, multiple spectral-texture matrices of video signals are generated from the template in the spectral wavelengths of ultraviolet, visible and infrared spectrum in non-polarized and polarized reflected light for solid environments or transmitted light for liquid environments. Then, the obtained matrices are statistically processed separately from the spectral and textural features, and the results are compared with the results obtained in the measurement of the standard.

EFFECT: increase of reliability and simplification of the control method.

2 cl, 6 dwg

Description

The invention relates to technical physics and can be used for express control and identification of various objects, both organic and inorganic, by their spectral and structural features using optical and automation means.

State of the art

At present, spectral methods are widely used to determine the quality of various objects (solid and liquid samples of different materials, metals and alloys, biological objects, drugs, cosmetics, food, etc.).

Classical methods for spectral analysis of samples are expensive and non-operational, and sample preparation and spectral analysis itself are laborious and time consuming. In addition, laboratories equipped with complex and expensive instruments, as well as qualified specialists, are required to perform analyzes.

At the same time, due to the presence of a large number of counterfeit products on the market, there is a need for inexpensive methods and devices suitable for widespread use and use by ordinary unskilled users to control goods purchased and received.

A known method of identifying an object based on the use of the surface structure of the object, according to which analyze the data obtained on the relief of the surface area, which are recorded and used for comparison with the data of the same surface area in the control process (application UK No. 2097979, CL G09K 9/00 publ. 1982).

The disadvantage of this method is the need for visual perception of the relief, the subjectivity of the analysis and its inoperability.

There is a method of identifying an object, according to which a selected surface area of the object is illuminated with a reference electromagnetic radiation stream uniformly distributed in space, photo-detection is used to transform the image of the selected surface area of the object into electrical signals, converting them into a set of digital parameters that are used to compare with similar sets characterizing a reference image of an object and a spatially uniform calibration template placed in place of an ident fitsiruemogo object (see., e.g., specification of a patent RF №2117989, Cl. G09K 9/58, publ. 1998).

However, this method has the disadvantage that the images formed depend on both the lighting parameters and the orientation, as well as the scale of the object. To compare the characteristics of the obtained images with the characteristics of the reference images, it is necessary to manually adjust these parameters, which significantly reduces the efficiency and reliability of identification.

A device known as “Photo-electric sorting apparatus” is known. (ABSTRACT GB929104. Patent GB929104 (A) - Improvements relating to the sorting of translucent objects. April 13, 1962).

In this device for sorting translucent objects, such as rice grains, which change the polarization of polarized light passing through them, it contains a illuminator illuminating the grains with polarized light, a photosensitive device receiving light passing through the grains, a polarization analyzer, and a sorter controlled by the photosensitive device , and means for reducing or eliminating changes in the output of the device due to changes in the size of the object.

This device uses an optical shutter formed by crossed polarizers and analyzers. It has a narrow area of specialized application and cannot be used for express quality control of objects, in contrast to the method proposed in the application, which examines the polarization of the light flux reflected from an identifiable object, as well as the spectral-texture identification features of this object.

A device for detecting foreign bodies in meat products is known, which uses the optical method (Patent GB2446822 (A) - Quality control of meat products using optical imaging. ABSTRACT GB2446822 (2008-08-27)).

The device uses red and green light sources made on LEDs, an image forming device that creates images from red and green sources at the same time, crossed polarizing filters installed in front of the light sources and the image forming device, and an image processing processor that identifies dark areas in two images.

The device under consideration (and the implemented method) is based on the use of an optical shutter formed by crossed polarizers and an analyzer. It has a narrow area of specialized application and cannot be used for express quality control of objects, in contrast to the method proposed in the application, which analyzes the polarization of the light flux reflected from the identified object, as well as its spectral-textural identification features.

A device is known as "LCD panel test apparatus)) (Patent US5734158 (A), ABSTRACT US5734158).

This device is designed to automatically detect defects in LCD panels. Basically, the device includes an optical shutter of crossed polarizers, between which the tested LCD panel is placed, a controller for controlling the LCD panel, a camera for observing light signals transmitted through the optical shutter, an analog-to-digital converter, and a processor for processing the received images.

The device in question is based on the use of an optical shutter formed by crossed polarizers and analyzers. It has a narrow area of specialized application and cannot be used for express quality control of objects, in contrast to the method proposed in the application, which analyzes the polarization of the light flux reflected from the identified object, as well as its spectral-textural identification features.

The closest to the invention in technical essence and the achieved result is a method for identifying an object, which consists in the fact that the calibration template and the selected surface area of the identifiable object is illuminated with spectrally-narrow-band reference electromagnetic radiation fluxes of non-overlapping spectral channels of visible and near infrared ranges, in reflected electromagnetic radiation fluxes form calibrating electrical signals and electrical signals of identifiable volume They calibrate the indicated signals by changing the power of the reference flows and photodetection sensitivities in each of the spectral channels until the maximum values of the calibrating signals reach the specified reference value, convert the electrical signals into two final spatially spectral shaped signals, form the spectral and spatial -spectral images of the identified object as two-dimensional frequency diagrams of the initial and final spatial spectral spectra ase signals converting them into a set of digital parameters, which are used for comparison with the previously stored similar sets which characterize a reference object image and spatially uniform calibration pattern placed at the location identified object (patent RU №2178562 Cl. G01N 33/02, publ. 01/20/2002).

However, the disadvantage of this method is the dependence of the identification result on the scale of the surface being identified, the impossibility of identifying surfaces containing active atomic and molecular structures that polarize light, the complexity of processing the measured values, namely spectral shaped signals, and also representing the processing result in the form of a three-dimensional surface, which complicates its simple and clear interpretation.

The objective of the invention is to remedy these disadvantages with increasing the reliability of express control of samples, taking into account the scale and polarizing properties of objects.

The technical result consists in the possibility of increasing the reliability and at the same time simplifying the method of express control of various samples by comparing over a wider measurement base with presenting the result of the comparison with the standard in a form convenient for interpretation by an ordinary (unskilled) user.

This problem is solved, and the technical result is achieved due to the fact that the method of monitoring objects consists in the fact that the selected surface area of the identified object and the calibration template are illuminated by spectrally narrow-band flows of electromagnetic radiation from non-overlapping spectral channels of the visible and near infrared ranges, in reflected electromagnetic radiation streams form calibrating electrical signals and electrical signals of the identified object, carry out calibres these signals by changing the electromagnetic radiation power or photodetection sensitivity in each of the spectral channels until the maximum values of the calibrating signals reach a specified reference value, convert the electrical signals into final spatially spectral shaped signals, form the spectral and spatial spectral images of the identifiable object as two-dimensional diagrams of the initial and final spatially spectral shaped signals with transform putting them into a set of digital parameters, which is used for comparison with previously remembered similar sets characterizing the reference image of an object and a spatially uniform calibration template placed in place of an identifiable object, while additionally sequentially performing observations in different spectral ranges of the spectrum in reflected and transmitted, polarized and non-polarized light emitted by luminescent LEDs in the ultraviolet, visible and infrared ranges form multiple spectral-texture matrices of video signals from a sample in different spectral wavelength ranges of the ultraviolet, visible and infrared spectra of electromagnetic radiation in unpolarized and polarized reflected light for solid media or in transmitted light for liquid media, then the matrices are processed separately by spectral and texture signs based on methods of mathematical statistics, then compare the results with the results obtained by measuring e coupon, or with data from an electronic database, and the processing results are presented in graphic or analytical form.

Preferably, multiple sample evaluations are performed using video recording with optical zoom of the surface of the object to a reference value.

In the presented method, in contrast to the closest analogue, spectrozonal images are formed by multi-frame shooting of the test sample and standard when illuminated with spectrally narrow-band electromagnetic radiation fluxes in the spectral channels of the ultraviolet, visible and near infrared ranges, both in non-polarized and linearly polarized in the light, carrying out before this an optical zooming of the surface of the identifiable object to a reference value, which allows the totality of measurements to determine the measure of similarity of the inspected object with the sample.

SUMMARY OF THE INVENTION

Using a black-and-white television CCD camera by video recording of the studied objects, multiple images are obtained, which form the spectral-texture two-dimensional matrixes of the levels of video signals.

The levels of video signals depend on the spectral reflection (transmission) coefficients of the light flux emitted by the fluorescent LEDs in the ultraviolet, visible and infrared ranges, from the objects being studied. The obtained information is used to assess the similarity of the object with the standard.

Brief Description of the Drawings

In FIG. 1. shows the result of data processing by the method adopted for the prototype;

in FIG. 2. shows a diagram of the implementation of the method.

in FIG. 3. The petal diagram of the comparison of the object and the standard by the spectral feature in unpolarized light is shown;

in FIG. 4 shows a histogram comparing an object and a standard by spectral feature in unpolarized light;

in FIG. Figure 5 shows a comparison of the textures of an object and a reference in the green spectral range in unpolarized light.

in FIG. Figure 6 shows a comparison of the textures of an object and a reference in the blue spectral range in linearly polarized light.

DETAILED DESCRIPTION OF THE INVENTION

The proposed method is implemented as follows.

Calibrate UV, B, G, R, and IR video channels using a calibration template, achieving the same video signal span in the channels by adjusting the exposure of the CCD camera (by changing the illumination in each channel or exposure time).

Receive multi-frame images of the sample in UV, B, G, R and IR-video channels, gaining N frames for statistical processing. In this case, the exposure is not regulated.

As a result, N two-dimensional matrices of measured values of the sample and reference signals are obtained, which have the form:

,

,

where Ψ is the spectral range,

x _ij is the measured value of the j-th element of the i-th row of the matrix M.

Comparison of the spectral images of the sample and the standard.

For each matrix obtained in a series of N measurements in a given spectral range Ψ, the averaging operation is performed:

,

,

- среднее значение матрицы чисел, полученныхWhere

is the average value of the matrix of numbers obtained

when shooting in this spectral range,

n is the number of elements in the TV frame,

m is the number of lines in the TV frame.

The average value of the numbers of each of the N matrices in this spectral range is calculated:

and find the average of the average values of these matrices, measured in this spectral range:

.

.

раз.The resulting number characterizes the spectral image of the object in the given spectral range K _Ψ . As is known, when performing such averaging, the measurement accuracy increases in

time.

), дисперсию (σ) и стандартное отклонение среднего (

) в каждом спектральном диапазоне для образца и эталона по известным формулам математической статистики.Find the average (

), variance (σ) and standard deviation of the mean (

) in each spectral range for the sample and standard according to well-known formulas of mathematical statistics.

) из измерений для образца и эталона и нормируют по нему все остальные значения среднего. Полученные нормированные значения являются относительными спектральными коэффициентами диффузного отражения исследуемого образца и эталона k₀ и k_э.Perform data comparison. To do this, determine the maximum value of the average (

) from measurements for the sample and standard and normalize all other values of the average according to it. The obtained normalized values are the relative spectral coefficients of diffuse reflection of the test sample and the standard k ₀ and k _e .

The difference between the measurement data of the sample and the standard is determined. To do this, find the difference between the relative spectral coefficients of diffuse reflection of the standard and the sample k ₀ and k _e in each spectral range.

The total error of the standardized measurement values of the standard and the test sample is determined, and then the significance of the differences between the measurement results of the standard and the test sample in each spectral range is found.

Present the results of comparing the data of the sample and the standard in graphical form. For this, lobe charts or histograms are constructed (Fig. 3 and Fig. 4), which clearly show the difference between the sample and the standard.

Table 1 shows the difference in the area of the petal diagrams shown in FIG. 3.

Table 2 shows the result of an analytical comparison of the spectral reflection coefficients of two soybean flour varieties, the total measurement error is calculated, and the significance of the differences is determined based on the confidence interval at a confidence coefficient of 95.4%.

A measure of similarity between the sample and the standard of soybean flour was also determined by the formula:

- нормированные значения спектральных коэффициентов отражения эталона (α) и образца (β),Where:

- normalized values of the spectral reflection coefficients of the standard (α) and sample (β),

Ψ is the spectral range.

The measure of similarity in this particular case was M = 0.83.

To quantify the differences between the object and the standard, the areas of the petal diagrams are determined by the known planimetric formulas and the difference between these areas is calculated.

where S _E - the area of the diagram of the standard,

S _O is the area of the sample diagram.

Determine the measure of similarity (M) of the sample and the standard by the formula:

,

,

where k and _E k _O - relative spectral reflectances and the reference sample in the selected spectral range.

UV, B, G, R, IR - spectral ranges.

Additional information can be obtained by processing the matrices obtained by shooting the standard and the sample in polarized light.

Comparison of texture images of the sample and standard.

For texture comparison, the values of matrix elements of the sample (or standard) obtained during measurements are summarized, then these values are divided by the number of matrices N and matrixes of structural images of the Y sample (standard) are obtained:

,

,

where Ψ is the spectral range,

- измеренное значение j-го элемента i-ой строки матрицы в данном спектральном диапазоне,

- the measured value of the j-th element of the i-th row of the matrix in this spectral range,

- усредненное значение j-го элемента i-ой строки матрицы в данном спектральном диапазоне.

- the average value of the j-th element of the i-th row of the matrix in this spectral range.

Using the obtained values, normalized graphs of the distribution of matrix elements are constructed

,

,

max - максимальное значение элемента матрицы,Where

max - the maximum value of the matrix element,

n _EL is the number of identical matrix elements,

n _EL.MAX. is the total number of matrix elements.

Then the mathematical expectation and variance of these distributions are determined and their comparison is performed. The comparison results can be presented in the form of graphs, histograms or tables.

Claims (2)

1. The method of monitoring objects, which consists in the fact that the selected surface area of the identifiable object and the calibration template are illuminated with spectrally narrowband electromagnetic radiation fluxes of non-overlapping spectral channels of the visible and near infrared ranges, calibrating electrical signals and electrical signals of the identifiable object are formed in the reflected electromagnetic fluxes, calibrate these signals by changing the power in each of the spectral channels methods of electromagnetic radiation or photodetection sensitivity until the maximum values of the calibrating signals reach a predetermined reference value, convert electrical signals into finite spatially spectral shaped signals, form the spectral and spatial spectral images of the identified object as two-dimensional diagrams of the initial and final spatial spectral figurative signals with their conversion into a set of digital parameters, which is used for comparison with previously numerous analogous sets characterizing the reference image of an object and a spatially uniform calibration template placed in place of an identifiable object, characterized in that it additionally sequentially performs observations in different spectral ranges of the spectrum in reflected or transmitted, polarized and non-polarized light emitted by ultraviolet fluorescent LEDs, visible and infrared ranges, form multiple spectral-texture matrices of video signals from in different spectral wavelength ranges of the ultraviolet, visible and infrared spectra of electromagnetic radiation in unpolarized and polarized reflected light for solid media or in transmitted light for liquid media, then the matrices are statistically processed separately by spectral and texture features based on mathematical statistics methods , then compare the results with the results obtained from measurements of the standard, or with data from an electronic database, and the results the processing is presented in graphic or analytical form. 2. The method according to p. 1, characterized in that they perform multiple assessments of samples using video recording with optical zoom of the image of the surface of the object to a reference value.

Images