Method and device for determining multispectral combination, object recognition method and device

Abstract

The present invention discloses a method and device for determining a multispectral combination, and a method and device for object recognition. The method for determining the multispectral combination includes: obtaining a characteristic wavelength of the object under a light source based on the spectral reflectance of the object and the spectral distribution of the light source; and obtaining a multispectral combination for identifying the object based on the characteristic wavelength. The determined multispectral combination can respond to changes in the external environment, such as the light source or hardware module, to effectively distinguish objects and similar objects in different external environments.

Description

Technical Field

[0001] The present invention relates to the technical field of multispectral imaging, and in particular to a method and device for determining a multispectral combination, and a method and device for object recognition. Background Art

[0002] Multispectral imaging is an image captured through specific wavelengths. This can be achieved by designing filters to separate wavelengths or by using instruments tuned to specific wavelengths, such as visible light, infrared light, or ultraviolet light. Due to its wide range of wavelength channels, multispectral imaging can capture phenomena that are invisible to the naked eye.

[0003] Currently, the methods for using multispectral detection to detect objects and similar objects (such as skin color and similar skin color, green plants and similar green plants, etc.) are mostly observational methods. That is, by observing the difference in spectral reflectance between the object and similar objects at the corresponding wavelengths, the difference in spectral reflectance between the two is found, and then the conclusion that the two objects can be detected or distinguished as different objects at these specific wavelengths with large differences is made. Alternatively, multispectral imaging is used to calculate the proportion of specific wavelengths by differentiating them from each other, and the object and similar objects are distinguished based on the different ranges of the area on the scale diagram. Although these two methods can distinguish specific objects at specific wavelengths, they lack a holistic assessment of the entire visible light band and its adjacent bands (ultraviolet light and infrared light). As a result, when the external environment, such as the light source or hardware module, changes, it is easy to fail to distinguish or make a wrong judgment. Summary of the Invention

[0004] The present invention aims to address, at least to some extent, one of the technical problems in the related art. To this end, a first object of the present invention is to provide a method for determining a multispectral combination. The multispectral combination determined by this method can effectively distinguish objects and similar objects in different external environments in response to changes in the external environment, such as light sources or hardware modules.

[0005] The second object of the present invention is to provide a device for determining a multi-spectral combination.

[0006] The third object of the present invention is to provide an object recognition method.

[0007] A fourth objective of the present invention is to provide an object recognition device.

[0008] A fifth object of the present invention is to provide an electronic device.

[0009] A sixth object of the present invention is to provide a computer-readable storage medium.

[0010] To achieve the above-mentioned objectives, an embodiment of the first aspect of the present invention proposes a method for acquiring a multi-spectral combination, comprising: acquiring the characteristic wavelength of the object under the light source based on the spectral reflectance of the object and the spectral distribution of the light source; and acquiring a multi-spectral combination for identifying the object based on the characteristic wavelength.

[0011] According to the method for obtaining a multi-spectral combination in an embodiment of the present invention, the characteristic wavelength of the object under the light source can be obtained through the spectral reflectance of the object and the spectral distribution of the light source; and the multi-spectral combination used to identify the object is obtained based on the characteristic wavelength. The determined multi-spectral combination can respond to changes in the external environment such as the light source or hardware module, and effectively distinguish between objects and control objects in different external environments.

[0012] According to one embodiment of the present invention, the spectral reflectivity is the spectral reflectivity of the object within the full wavelength band or a partial wavelength band.

[0013] According to one embodiment of the present invention, the characteristic wavelength of the object under the light source is obtained based on the spectral reflectivity of the object and the spectral distribution of the light source, including: obtaining the spectral reflection intensity of the object based on the spectral reflectivity of the object and the spectral distribution of the light source; obtaining the characteristic wavelength of the object under the light source based on the spectral reflection intensity.

[0014] According to an embodiment of the present invention, obtaining the spectral reflection intensity of the object according to the spectral reflectivity of the object and the spectral distribution of the light source includes: performing convolution processing on the spectral reflectivity and the spectral distribution to obtain the spectral reflection intensity.

[0015] According to one embodiment of the present invention, the characteristic wavelength of an object under a light source is obtained based on the spectral reflection intensity, including: performing non-negative matrix decomposition on the spectral reflection intensity to obtain multiple basis functions of the spectral reflection intensity; obtaining the maximum energy of each basis function and the wavelength corresponding to the energy maximum; respectively obtaining the ratio of the maximum energy of each basis function to the maximum energy of all basis functions; and obtaining the characteristic wavelength from the wavelength corresponding to the energy maximum based on the ratio.

[0016] According to one embodiment of the present invention, a characteristic wavelength is obtained from the wavelength corresponding to the maximum energy value based on the ratio, including: sorting the comparison values ​​from large to small, obtaining the wavelength corresponding to the ratio of the first number in the sorting order to obtain the characteristic wavelength; or obtaining the wavelength corresponding to the ratio greater than or equal to the first ratio to obtain the characteristic wavelength.

[0017] According to one embodiment of the present invention, obtaining the characteristic wavelength of an object under a light source based on the spectral reflection intensity also includes: obtaining the wavelength corresponding to the energy in each basis function where the energy is discontinuous and the difference between the energy and the corresponding energy in other basis functions exceeds a first threshold, and using the wavelength as the characteristic wavelength.

[0018] According to one embodiment of the present invention, a multispectral combination for identifying an object is obtained based on characteristic wavelengths, including: performing curve fitting on each characteristic wavelength to obtain a spectral segment corresponding to each characteristic wavelength; obtaining, from the spectral segment, a wavelength that maximizes the color difference between the object and the reference object based on the spectral reflectance of the object, the spectral reflectance of the reference object, and the spectral distribution of the light source; and combining the wavelengths that maximize the color difference to obtain a multispectral combination.

[0019] According to one embodiment of the present invention, curve fitting is performed on each characteristic wavelength to obtain a spectrum segment corresponding to each characteristic wavelength, including: performing normal distribution curve fitting or square wave curve fitting on each characteristic wavelength to obtain a spectrum segment corresponding to each characteristic wavelength.

[0020] According to one embodiment of the present invention, based on the spectral reflectivity of the object, the spectral reflectivity of the reference object, and the spectral distribution of the light source, obtaining from the spectrum the wavelength that maximizes the color difference between the object and the reference object includes: based on the spectral reflectivity of the object, the spectral reflectivity of the reference object, and the spectral distribution of the light source, using one of a recursive method, an exhaustive method, and a brute force method to obtain from the spectrum the wavelength that maximizes the color difference between the object and the reference object.

[0021] To achieve the above-mentioned purpose, the second embodiment of the present invention proposes a device for determining a multi-spectral combination, including: a wavelength calculation unit, used to obtain the characteristic wavelength of an object under a light source based on the spectral reflectance of the object and the spectral distribution of the light source; a multi-spectral calculation unit, used to obtain a multi-spectral combination for identifying an object based on the characteristic wavelength.

[0022] According to the apparatus for determining a multispectral combination according to an embodiment of the present invention, the multispectral combination determined by the method for determining a multispectral combination described above can effectively distinguish between objects and reference objects in different external environments in response to changes in external environments such as light sources or hardware modules.

[0023] To achieve the above-mentioned purpose, the third aspect of the present invention proposes an object recognition method, including: obtaining a multi-spectral combination corresponding to the target object, the multi-spectral combination being obtained based on the characteristic wavelength of the target object under a preset spectral distribution; and identifying the target object based on the multi-spectral combination.

[0024] According to the object recognition method of an embodiment of the present invention, the multispectral combination determined by the above-mentioned method for determining the multispectral combination can effectively distinguish the target object and the reference object under different external environments in response to changes in the external environment such as the light source or the hardware module.

[0025] According to one embodiment of the present invention, the characteristic wavelength is obtained based on the energy of each basis function and the wavelength corresponding to the energy in multiple basis functions obtained by performing non-negative matrix decomposition on the spectral reflection intensity of the object under a preset spectral distribution. The characteristic wavelength includes the wavelength corresponding to the first ratio of the ratio of the maximum energy of a single basis function to the maximum energy of all basis functions, the wavelength corresponding to the ratio of the first number whose contrast values ​​are sorted from large to small, and the wavelength corresponding to the energy in each basis function is discontinuous and the difference between the energy and the corresponding energy in other basis functions exceeds a first threshold.

[0026] According to an embodiment of the present invention, the spectroscopic reflection intensity is obtained based on the spectroscopic reflectivity of the target object and a preset spectral distribution.

[0027] According to one embodiment of the present invention, the multi-spectral combination is obtained based on the spectral reflectance of the target object, the spectral reflectance of the reference object and the preset light source distribution, and the wavelength that maximizes the color difference between the target object and the reference object is obtained from the corresponding spectral bands obtained by curve fitting the characteristic wavelengths.

[0028] To achieve the above-mentioned objectives, an embodiment of the fourth aspect of the present invention proposes an object recognition device, including: an acquisition unit, used to obtain a multi-spectral combination corresponding to a target object; and an identification unit, used to identify the target object based on the multi-spectral combination, wherein the multi-spectral combination is obtained based on the characteristic wavelength of the target object under a preset spectral distribution.

[0029] According to the object recognition device of the embodiment of the present invention, the multispectral combination determined by the above-mentioned method for determining the multispectral combination can effectively distinguish the target object and the reference object under different external environments in response to changes in the external environment such as the light source or the hardware module.

[0030] To achieve the above-mentioned objectives, the fifth embodiment of the present invention proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the steps of the method for determining the multi-spectral combination as in the first embodiment above, or implements the steps of the object recognition method as in the third embodiment.

[0031] According to the electronic device of an embodiment of the present invention, the multispectral combination determined by the above-mentioned method for determining a multispectral combination or object recognition method can respond to changes in the external environment, such as light sources or hardware modules, and effectively distinguish between objects and reference objects in different external environments.

[0032] To achieve the above-mentioned objectives, the sixth aspect of the present invention proposes a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for determining the multi-spectral combination in the first embodiment are implemented, or the steps of the object recognition method in the third embodiment are implemented.

[0033] According to the computer-readable storage medium of an embodiment of the present invention, the multispectral combination determined by the above-mentioned method for determining a multispectral combination or object recognition method can respond to changes in the external environment, such as light sources or hardware modules, to effectively distinguish between objects and control objects in different external environments.

[0034] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A flowchart of a method for determining a multi-spectral combination according to an embodiment of the present invention;

[0036] Figure 2 A flowchart of obtaining a characteristic wavelength of an object under a light source according to one embodiment of the present invention;

[0037] Figure 3 A flowchart of obtaining a characteristic wavelength of an object under a light source based on the spectral reflection intensity according to one embodiment of the present invention;

[0038] Figure 4 is a basis function representing the spectral reflectance of skin color under various light sources according to an embodiment of the present invention;

[0039] Figure 5 A schematic diagram of obtaining a multi-spectral combination for identifying an object according to one embodiment of the present invention;

[0040] Figure 6 A multi-spectral combination that is most capable of distinguishing skin color and skin-like skin color according to one embodiment of the present invention;

[0041] Figure 7 A schematic structural diagram of an apparatus for determining a multi-spectral combination according to an embodiment of the present invention;

[0042] Figure 8 is a flowchart of an object recognition method according to one embodiment of the present invention;

[0043] Figure 9 FIG. 4 is a schematic structural diagram of an object recognition device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0045] It should be noted that the color C usually perceived through various media can be expressed using the following formula (1):

[0046] C=∑R(λ)×S(λ)×M n (λ) (1)

[0047] Here, R is the object's spectral reflectance, S is the spectral distribution of the light source, M is the spectral response curve of a medium, and λ is the corresponding wavelength. The medium here can be a device such as the human eye, a camera, or a multispectral sensor that generates its own response spectrum based on a specific wavelength. n is the number of spectral response curves, which varies depending on the type of device. For example, the human eye has three spectral response curves (red, green, and blue), a camera can have 3-4 spectral response curves depending on sensor characteristics, and a multispectral sensor can have 6-10 spectral response curves, or more depending on demand.

[0048] At present, the method of using multispectral detection of objects and similar objects is mostly observation method, specifically using observation method to analyze under spectral reflectance, and the so-called perceived color is generally calculated according to the above formula (1), which is related to the characteristics and number of the light source and the spectral response curve. It is easy for the same color to produce different perception results under different light sources or metamerism (metamerism means that under a given light source, two different spectral reflectances are indistinguishable to a given observer, that is, the same). Therefore, the traditional observation method cannot respond to some sudden changes, such as changes in light sources or changes in hardware modules (such as camera sensors, multispectral devices, etc.), resulting in situations where it cannot be distinguished or misjudged. Based on this, the present application provides a method for determining a multispectral combination. The multispectral combination determined by the method can respond to changes in external environments such as light sources or hardware modules, and effectively distinguish objects and control objects under different external environments. The following is a detailed description with reference to the accompanying drawings.

[0049] Figure 1 FIG. 1 is a flow chart of a method for determining a multi-spectral combination according to an embodiment of the present invention. Figure 1 As shown, the method for determining a multi-spectral combination includes the following steps:

[0050] Step S101 : acquiring a characteristic wavelength of the object under the light source according to the spectral reflectivity of the object and the spectral distribution of the light source.

[0051] Specifically, after obtaining the spectral reflectance of the object and the spectral distribution of the light source, the characteristic wavelength of the object under the light source can be obtained according to a certain algorithm.

[0052] In some embodiments, the spectral reflectivity is the spectral reflectivity of the object over the entire wavelength band or a portion of the wavelength band. That is, the spectral reflectivity of the object over the entire wavelength band or a portion of the wavelength band can be obtained as needed, where the entire wavelength band includes ultraviolet light (wavelength less than 380nm), visible light (wavelength 380nm-780nm), and infrared light (wavelength greater than 780nm). In specific implementations, the spectral reflectivity of the object over the entire wavelength band or a portion of the wavelength band can be obtained by measurement and / or from a spectral reflectivity database, such as obtaining the spectral reflectivity of the object under ultraviolet light, visible light, infrared light, or a combination of the three.

[0053] It should be noted that the light source includes both outdoor and indoor light sources, and the spectral distribution of the light source can be pre-stored in the device. Prior to this, data collection is required on the spectral distribution of the light source so that the obtained multi-spectral combination can be adapted to different light sources. For example, when the object and similar objects are only in outdoor scenes, the light source is an outdoor light source, and the spectral distribution of the outdoor light source is obtained; when the object and similar objects are only in indoor scenes, the light source is an indoor light source, and the spectral distribution of the indoor light source is obtained; when the object and similar objects are in both outdoor and indoor scenes, the light source is both outdoor and indoor light sources, and the spectral distribution set of the outdoor and indoor light sources is obtained.

[0054] For outdoor light sources, since the outdoor light source is mainly daylight, its spectral distribution can be obtained by calculating the principal components of the actual daylight spectrum and the weight corresponding to each principal component. Specifically, it can be calculated using the following formula (2):

[0055] S(λ)=S0(λ)+M1S1(λ)+M2S2(λ) (2)

[0056] Wherein, S is the spectral distribution of the outdoor light source, S0, S1, and S2 are the principal components of the actual daylight spectrum, respectively. Specifically, they can be the first three eigenvalues ​​obtained by analyzing the actual daylight spectrum using principal component analysis (PCA). λ is the corresponding wavelength. M1 and M2 can be regarded as the weights of the principal components S1 and S2, or as the weights of the corresponding wavelengths. Specifically, they can be derived from the x and y coordinates corresponding to the Planck curve on the CIE (x, y) chromaticity diagram.

[0057] For indoor light sources, the spectral distribution can be obtained by measuring in light boxes or in frequently used indoor scenes such as shopping malls and offices. Specifically, it can be obtained by using a spectroradiometer or illuminance meter.

[0058] Step S102: acquiring a multi-spectral combination for identifying an object according to the characteristic wavelength.

[0059] It should be noted that the characteristic wavelength is the specific wavelength that can most effectively distinguish between objects and control objects under various light source conditions. The most common way to obtain it is to select it according to the wavelength corresponding to the maximum energy of each basis function in the spectral distribution of the light source, and perform curve fitting (normal distribution curve fitting or square wave curve fitting) on ​​all the selected characteristic wavelengths that can most effectively distinguish between objects and control objects in a certain way, so as to obtain a multi-spectral combination for identifying objects.

[0060] According to the method for obtaining a multi-spectral combination in an embodiment of the present invention, the characteristic wavelength of the object under the light source can be obtained through the spectral reflectance of the object and the spectral distribution of the light source; and the multi-spectral combination used to identify the object is obtained based on the characteristic wavelength. The determined multi-spectral combination can respond to changes in the external environment such as the light source or hardware module, and effectively distinguish between objects and control objects in different external environments.

[0061] As mentioned above, the characteristic wavelength of the object under the light source can be obtained based on the spectral reflectivity of the object and the spectral distribution of the light source. Figure 2 As shown, according to the spectral reflectivity of the object and the spectral distribution of the light source, the characteristic wavelength of the object under the light source is obtained, including:

[0062] Step S201 : obtaining the spectral reflection intensity of the object according to the spectral reflectivity of the object and the spectral distribution of the light source.

[0063] In some embodiments, obtaining the spectral reflection intensity of the object according to the spectral reflectivity of the object and the spectral distribution of the light source includes: performing convolution processing on the spectral reflectivity and the spectral distribution to obtain the spectral reflection intensity.

[0064] Specifically, after obtaining the spectral reflectance of the object and the spectral distribution of the light source, the spectral reflectance and the spectral distribution can be convolved. The resulting value is called the spectral reflectance intensity I of the object, as shown in the following formula (3):

[0065] I=∑R(λ)×S(λ) (3)

[0066] Among them, R is the spectral reflectance of the object, S is the spectral distribution of the light source, and λ is the corresponding wavelength.

[0067] Step S202: Acquire the characteristic wavelength of the object under the light source according to the spectral reflection intensity.

[0068] Specifically, after obtaining the spectroscopic reflection intensity of the object, a certain algorithm can be used to calculate the characteristic wavelength of the object under the light source.

[0069] In some embodiments, as Figure 3 As shown, according to the spectral reflection intensity, the characteristic wavelength of the object under the light source is obtained, including:

[0070] Step S301 : performing non-negative matrix decomposition on the spectral reflection intensity to obtain a plurality of basis functions of the spectral reflection intensity.

[0071] Specifically, after obtaining the spectroscopic reflectance intensity, the NMF (Non-negative Matrix Factorization) method can be used to obtain the characteristic wavelength of the object under various light sources. NMF, also known as non-negative matrix factorization, is a matrix decomposition method under the constraint that all elements in the matrix are non-negative. Its basic idea is that given a non-negative matrix V, NMF can find a non-negative matrix W and a non-negative matrix H such that the product of matrices W and H is approximately equal to the values ​​in matrix V. W can also be called the base image matrix, which is equivalent to the features extracted from the original matrix V, and H can be called the coefficient matrix or weight matrix.

[0072] Furthermore, the spectral reflection intensity I is subjected to non-negative matrix decomposition according to the following formula (4) according to the NMF method:

[0073]

[0074] Where I is the object's spectral reflection intensity, W is the basis function that makes up the spectral reflection intensity I, and H is the weight of the corresponding basis function. The number of basis functions W can be defined according to needs. By defining the number of basis functions W and the combination of weights H, the spectral reflection intensity I in various situations can be accurately reproduced. For example, Figure 4 is the basis function representing the spectral reflectance of skin color under various light sources. Each curve represents a basis function, with the horizontal axis representing wavelength and the vertical axis representing energy. Figure 4 Basis function 1 is the basis function representing the spectral reflectance of skin color under the first light source, and basis function 2 is the basis function representing the spectral reflectance of skin color under the second light source. Therefore, the spectral reflectance of skin color can be represented by basis functions under various light sources.

[0075] It should be noted that principal component analysis can also be used instead of the NMF method. However, compared with the principal component analysis method, the basis functions obtained by the NMF method are all positive, thus eliminating the need to calculate the proportion of explained variation.

[0076] Step S302 : Obtain the maximum energy value of each basis function and the wavelength corresponding to the maximum energy value.

[0077] Specifically, after obtaining the basis functions, the maximum energy of each basis function and the wavelength corresponding to the maximum energy are obtained. For example, Figure 4 As shown, each curve represents a basis function, the horizontal axis represents the wavelength, and the vertical axis represents the energy. It can be clearly seen from the figure that the maximum energy of each basis function (in the specific implementation, the maximum energy can be automatically calculated to obtain the maximum energy). For example, the maximum energy of basis function 1 is 41.2, corresponding to a wavelength of 610, and the maximum energy of basis function 2 is 43.5, corresponding to a wavelength of 770.

[0078] Step S303 , respectively obtaining the ratio of the maximum energy of each basis function to the maximum energy of all basis functions.

[0079] Specifically, after obtaining the maximum energy value of each basis function, these energy maximum values ​​can be compared and judged to determine the maximum value of these energy maximum values, and use it as the maximum energy value of all basis functions. Then, the ratio of the maximum energy value of each basis function to the maximum energy value of all basis functions is calculated, which can be specifically expressed by the following formula (5):

[0080]

[0081] Where E is the maximum energy of a certain basis function, n is the wavelength corresponding to its maximum energy, P is the importance of the wavelength corresponding to its maximum energy, and max(E) is the maximum energy among all basis functions. For example, refer to Figure 4 As shown, the maximum energy value max(E) among all basis functions is the maximum energy value of basis function 2, that is, 43.5. Then, the corresponding value of the maximum energy value of basis function 2 is used as the denominator, and the maximum energy value of basis function 1 is ratioed to the maximum energy value of basis function 2. Similarly, the ratio of the maximum energy value of each basis function to the maximum energy value of basis function 2 can be calculated.

[0082] Step S304: Obtain a characteristic wavelength from the wavelength corresponding to the maximum energy value according to the ratio.

[0083] In some embodiments, obtaining a characteristic wavelength from the wavelengths corresponding to the energy maximum values ​​based on the ratios includes: sorting the ratios from largest to smallest and obtaining the wavelength corresponding to the ratio of the first number in the sorted order to obtain the characteristic wavelength; or obtaining the wavelength corresponding to the ratio greater than or equal to the first ratio to obtain the characteristic wavelength. That is, after obtaining the ratio of the energy maximum of each basis function to the energy maximum of all basis functions, as an example, the obtained ratios can be sorted, and then the wavelengths of the first numbers in the sorted order (e.g., 5) can be used as the characteristic wavelengths of the object under each light source; as another example, the obtained wavelength ratios can be compared with the first ratio (e.g., 60%), and then the wavelengths greater than or equal to the first ratio can be used as the characteristic wavelengths of the object under each light source.

[0084] It should be noted that in some embodiments, obtaining the characteristic wavelength of an object under a light source based on the spectral reflectance intensity further includes obtaining the wavelength corresponding to an energy discontinuity in each basis function where the energy differs from the corresponding energy in the other basis functions by more than a first threshold, and using this wavelength as the characteristic wavelength. In other words, if there is a discontinuity in the energy distribution in the basis function and the energy differs significantly from the energy at the corresponding position in the other basis functions, it indicates that the spectral reflectance of the object is unique at that wavelength, which is also recorded and used as the characteristic wavelength of the object under the light source.

[0085] As previously mentioned, a multispectral combination for object recognition can be obtained based on characteristic wavelengths. Specifically, after obtaining the characteristic wavelengths of an object under various light sources, a multispectral evaluation can be performed on these characteristic wavelengths, and the final multispectral combination for object recognition can be obtained based on the evaluation results.

[0086] In some embodiments, as Figure 4 As shown, according to the characteristic wavelength, a multi-spectral combination for identifying an object is obtained, including:

[0087] Step S401 : performing curve fitting on each characteristic wavelength to obtain a spectrum segment corresponding to each characteristic wavelength.

[0088] In some embodiments, curve fitting is performed on each characteristic wavelength to obtain a spectrum segment corresponding to each characteristic wavelength, including: performing normal distribution curve fitting or square wave curve fitting on each characteristic wavelength to obtain a spectrum segment corresponding to each characteristic wavelength. In other words, all spectrum segments of the multispectral spectrum are considered to be ideal Gaussian distributions or square wave distributions, wherein when all spectrum segments of the multispectral spectrum are considered to be ideal Gaussian distributions, the corresponding fitting formula is shown in formula (6):

[0089]

[0090] Where μ is the characteristic wavelength, σ is the half-wavelength width of the corresponding multispectral band, and x is the random variable of wavelength.

[0091] Step S402 : obtaining a wavelength that maximizes the color difference between the object and the reference object from the spectrum according to the spectral reflectance of the object, the spectral reflectance of the reference object, and the spectral distribution of the light source.

[0092] In some embodiments, based on the spectral reflectivity of the object, the spectral reflectivity of the reference object, and the spectral distribution of the light source, obtaining from the spectrum the wavelength that maximizes the color difference between the object and the reference object includes: based on the spectral reflectivity of the object, the spectral reflectivity of the reference object, and the spectral distribution of the light source, using one of a recursive method, an exhaustive method, and a brute force method to obtain from the spectrum the wavelength that maximizes the color difference between the object and the reference object.

[0093] Specifically, after obtaining the spectral band corresponding to each characteristic wavelength, the object and the reference object are distinguished based on these spectral bands. For example, a recursive method can be used to estimate the most effective multi-spectral combination that can distinguish the object and the reference object based on the known spectral band (i.e., wavelength distribution) through the spectral reflectance of the object and the reference object and the above formula (1). The overall expression is shown in the following formula (7):

[0094]

[0095] Where R is the spectral reflectance of the object, R' is the spectral reflectance of the reference object, S is the spectral distribution of the light source, μ is the characteristic wavelength, σ is the half-wavelength width of the corresponding multispectral band, m is the wavelength fine-tuning factor, and k is the half-wavelength width fine-tuning factor. This formula can be used to find the wavelength where the color difference between the object and the reference object is the largest.

[0096] It should be noted that the scheme of obtaining the wavelength that maximizes the color difference between the object and the reference object in the spectrum using other methods such as exhaustive method and brute force method will not be described in detail here.

[0097] Step S403: combining wavelengths with the maximum color difference to obtain a multi-spectral combination.

[0098] That is to say, the wavelengths that can most effectively distinguish the object from the reference object are combined to form a multispectral combination. Specifically, after recursive calculation using the above formula (7), the fine-tuning term m of each characteristic wavelength and the fine-tuning term k of the half-wave width of the corresponding multispectral band can be obtained. Then, based on the fine-tuning term m, the wavelength of the final multispectral combination can be determined, specifically, the characteristic wavelength μ minus the fine-tuning term m. At the same time, based on the fine-tuning term k, the half-wave width of the final multispectral band can be determined, specifically, the half-wave width σ plus the fine-tuning term k. Finally, Figure 6 The multispectral combination shown. Figure 6 is the normal distribution curve obtained after normal distribution fitting of each characteristic wavelength in the obtained multi-spectral combination, where the final characteristic wavelength μ is the characteristic wavelength minus the fine-tuning term m, and the half-wavewidth σ is the half-wavewidth plus the fine-tuning term k.

[0099] In order to enable those skilled in the art to understand the present application more clearly, a specific example is provided below for explanation.

[0100] Assuming that we need to identify skin color and skin-like skin color indoors and outdoors, we can perform convolution processing on the spectral reflectance of skin color and skin-like skin color and the spectral distribution of the light source to obtain the spectral reflectance intensity of skin color and skin-like skin color indoors and outdoors. Then, we perform non-negative matrix decomposition on the spectral reflectance intensity to obtain the basis function of the spectral reflectance intensity, such as Figure 4 As shown, the energy maximum of each basis function and the wavelength corresponding to the energy maximum are obtained. After obtaining the energy maximum of each basis function, these energy maximum values ​​can be compared and judged to determine the maximum value of these energy maximum values, and used as the energy maximum value of all basis functions. Then, the ratio of the energy maximum value of each basis function to the energy maximum value of all basis functions is calculated. After obtaining the ratio of the energy maximum value of each basis function to the energy maximum value of all basis functions, the obtained ratios can be sorted, and the characteristic wavelengths of skin color and skin-like colors under various light sources are selected from the wavelengths corresponding to the energy maximum values ​​according to the ratio sorting. Next, for each characteristic wavelength, the corresponding half-wavewidth is determined, and the recursive method is used to determine the fine-tuning items of each characteristic wavelength and the corresponding half-wavewidth. Finally, the final wavelength is determined based on the fine-tuning items of each characteristic wavelength, and the final half-wavewidth is determined based on the fine-tuning items of the half-wavewidth. The final multi-spectral combination can be determined based on the final wavelength and half-wavewidth, as shown in FIG. Figure 6 This multi-spectral combination can effectively distinguish skin color and skin-like colors.

[0101] In summary, according to the method for obtaining a multi-spectral combination in an embodiment of the present invention, the characteristic wavelength of an object under a light source can be obtained through the spectral reflectance of the object and the spectral distribution of the light source; and the multi-spectral combination for identifying the object is obtained based on the characteristic wavelength. The determined multi-spectral combination can respond to changes in the external environment such as the light source or hardware module, and effectively distinguish between objects and control objects in different external environments.

[0102] Figure 7 FIG. 1 is a schematic diagram of a device for determining a multi-spectral combination according to an embodiment of the present invention. Figure 7As shown, the multi-spectral combination acquisition device 100 includes: a wavelength calculation unit 110 and a multi-spectral calculation unit 120, wherein the wavelength calculation unit 110 is specifically used to obtain the characteristic wavelength of the object under the light source according to the spectral reflectance of the object and the spectral distribution of the light source; the multi-spectral calculation unit 120 is used to obtain the multi-spectral combination for identifying the object according to the characteristic wavelength.

[0103] In some embodiments, the spectral reflectivity is the spectral reflectivity of the object within the entire wavelength band or a partial wavelength band.

[0104] In some embodiments, the wavelength calculation unit 110 is specifically configured to: obtain the spectral reflection intensity of the object according to the spectral reflectivity of the object and the spectral distribution of the light source; and obtain the characteristic wavelength of the object under the light source according to the spectral reflection intensity.

[0105] In some embodiments, the wavelength calculation unit 110 is specifically configured to perform convolution processing on the spectral reflectance and the spectral distribution to obtain the spectral reflection intensity.

[0106] In some embodiments, the wavelength calculation unit 110 is specifically used to: perform non-negative matrix decomposition on the spectral reflection intensity to obtain multiple basis functions of the spectral reflection intensity; obtain the maximum energy of each basis function and the wavelength corresponding to the energy maximum; obtain the ratio of the maximum energy of each basis function to the maximum energy of all basis functions; and obtain the characteristic wavelength from the wavelength corresponding to the energy maximum based on the ratio.

[0107] In some embodiments, the wavelength calculation unit 110 is specifically used to: sort the comparison values ​​from large to small, obtain the wavelength corresponding to the ratio of the first number in the sorting order to obtain the characteristic wavelength; or obtain the wavelength corresponding to the ratio greater than or equal to the first ratio to obtain the characteristic wavelength.

[0108] In some embodiments, the wavelength calculation unit 110 is specifically configured to obtain a wavelength corresponding to an energy in each basis function where the energy is discontinuous and the difference between the energy and the corresponding energy in other basis functions exceeds a first threshold, and use the wavelength as a characteristic wavelength.

[0109] In some embodiments, the multi-spectral calculation unit 120 is specifically used to: perform curve fitting on each characteristic wavelength to obtain a spectral segment corresponding to each characteristic wavelength; obtain the wavelength that maximizes the color difference between the object and the reference object from the spectral segment based on the spectral reflectance of the object, the spectral reflectance of the reference object, and the spectral distribution of the light source; and combine the wavelengths that maximize the color difference to obtain a multi-spectral combination.

[0110] In some embodiments, the multi-spectral calculation unit 120 is specifically configured to perform normal distribution curve fitting or square wave curve fitting on each characteristic wavelength to obtain a spectrum segment corresponding to each characteristic wavelength.

[0111] In some embodiments, the multispectral calculation unit 120 is specifically used to: based on the spectral reflectance of the object, the spectral reflectance of the reference object and the spectral distribution of the light source, use one of the recursive method, the exhaustive method and the brute force method to obtain the wavelength that maximizes the color difference between the object and the reference object from the spectrum segment.

[0112] It should be noted that for the description of the device for determining the multi-spectral combination in this application, please refer to the description of the method for determining the multi-spectral combination in this application, and the details will not be repeated here.

[0113] According to an embodiment of the present invention, a device for determining a multispectral combination obtains a characteristic wavelength of an object under a light source based on the spectral reflectance of the object and the spectral distribution of the light source through a wavelength calculation unit; and obtains a multispectral combination for identifying the object based on the characteristic wavelength through the multispectral calculation unit. The determined multispectral combination can respond to changes in the external environment, such as the light source or hardware module, and effectively distinguish between objects and control objects in different external environments.

[0114] Figure 8 FIG. 1 is a flow chart of an object recognition method according to an embodiment of the present invention. Figure 8 As shown, the object recognition method includes the following steps:

[0115] Step S501 : obtaining a multi-spectral combination corresponding to a target object, where the multi-spectral combination is obtained based on characteristic wavelengths of the target object under a preset spectral distribution.

[0116] In some embodiments, the characteristic wavelength is obtained based on the energy of each basis function and the wavelength corresponding to the energy in multiple basis functions obtained by performing non-negative matrix decomposition on the spectral reflection intensity of the object under a preset spectral distribution. The characteristic wavelength includes the wavelength corresponding to the ratio of the maximum energy of a single basis function to the maximum energy of all basis functions is greater than or equal to a first ratio, the wavelength corresponding to the ratio of the first number whose comparison values ​​are sorted from large to small, and the wavelength corresponding to the energy in each basis function is discontinuous and the difference between the energy and the corresponding energy in other basis functions exceeds a first threshold.

[0117] Specifically, after obtaining the spectral reflection intensity of the object under the preset spectral distribution, the NMF method can be used to obtain each basis function of the multiple basis functions obtained. After obtaining the basis functions, the energy maximum of each basis function and the wavelength corresponding to the energy maximum are obtained respectively. After obtaining the energy maximum of each basis function, these energy maxima can be compared and judged to determine the maximum value of these energy maxima, and used as the energy maximum value of all basis functions. Then, the ratio of the energy maximum value of each basis function to the energy maximum value of all basis functions is calculated. After obtaining the energy maximum value of each basis function and the energy maximum value of all basis functions, the ratio of the energy maximum value of each basis function to the energy maximum value of all basis functions is calculated. After the ratio of the largest values ​​is obtained, as an example, the obtained ratios can be sorted, and then the wavelength of the first number (such as 5) in the first order can be used as the characteristic wavelength of the object under each light source; as another example, the obtained wavelength ratio can be compared with the first ratio (such as 60%), and then the wavelength greater than or equal to the first ratio can be used as the characteristic wavelength of the object under each light source; as another example, if there is an energy distribution discontinuity in the basis function and the energy is too different from the energy at the corresponding position in other basis functions, it means that the spectral reflectance of this object is unique at this wavelength, and it is also recorded and the wavelength is used as the characteristic wavelength of the object under the light source.

[0118] It should be noted that the spectral reflection intensity is obtained based on the spectral reflectivity of the target object and the preset spectral distribution. In other words, by convolving the spectral reflectivity of the target object with the preset spectral distribution, the spectral reflection intensity of the target object under the preset spectral distribution can be obtained. The formula for obtaining this is shown in formula (3) above and will not be repeated here.

[0119] Furthermore, the multi-spectral combination is obtained based on the spectral reflectance of the target object, the spectral reflectance of the reference object and the preset light source distribution, and the wavelength that maximizes the color difference between the target object and the reference object is obtained from the corresponding spectral bands obtained by curve fitting the characteristic wavelengths.

[0120] Specifically, a normal distribution curve fitting or a square wave curve fitting is performed on each characteristic wavelength to obtain the spectral segment corresponding to each characteristic wavelength. After obtaining the spectral segment corresponding to each characteristic wavelength, the target object and the reference object are distinguished based on these spectral segments. A recursive method can be used to find the wavelength portion where the color difference between the target object and the reference object is the largest through the above formula (7). The wavelengths that can most effectively distinguish the target object from the reference object are combined to form a multi-spectral combination.

[0121] Step S502: Identify the target object based on the multi-spectral combination.

[0122] Specifically, when target object identification is required, the user inputs the spectral reflectance of the target object. At this time, the multispectral sensor obtains the corresponding multispectral combination under the preset spectral distribution and then identifies the target object.

[0123] According to the object recognition method of an embodiment of the present invention, the multispectral combination corresponding to the target object is obtained according to the above-mentioned method for determining the multispectral combination, and the target object is identified based on the multispectral combination. This can respond to changes in the external environment such as light sources or hardware modules, and effectively distinguish between target objects and control objects in different external environments.

[0124] Figure 9 FIG. 1 is a schematic diagram of the structure of an object recognition device according to an embodiment of the present invention. Figure 9 As shown, the object recognition device 200 includes: an acquisition unit 210 and an identification unit 220, wherein the acquisition unit 210 is used to obtain a multi-spectral combination corresponding to the target object; the identification unit 220 is used to identify the target object according to the multi-spectral combination, wherein the multi-spectral combination is obtained based on the characteristic wavelength of the target object under a preset spectral distribution.

[0125] In some embodiments, the characteristic wavelength is obtained based on the energy of each basis function and the wavelength corresponding to the energy in multiple basis functions obtained by performing non-negative matrix decomposition on the spectral reflection intensity of the object under a preset spectral distribution. The characteristic wavelength includes the wavelength corresponding to the ratio of the maximum energy of a single basis function to the maximum energy of all basis functions is greater than or equal to a first ratio, the wavelength corresponding to the ratio of the first number whose comparison values ​​are sorted from large to small, and the wavelength corresponding to the energy in each basis function is discontinuous and the difference between the energy and the corresponding energy in other basis functions exceeds a first threshold.

[0126] In some embodiments, the spectroscopic reflection intensity is obtained based on the spectroscopic reflectivity of the target object and a preset spectral distribution.

[0127] In some embodiments, the multi-spectral combination is obtained based on the spectral reflectance of the target object, the spectral reflectance of the reference object and the preset light source distribution, and the wavelength that maximizes the color difference between the target object and the reference object is obtained from the corresponding spectral segments obtained by curve fitting the characteristic wavelengths.

[0128] It should be noted that for the description of the object recognition device in this application, please refer to the description of the object recognition method in this application, and the details will not be repeated here.

[0129] According to an embodiment of the present invention, the object recognition device obtains the multi-spectral combination corresponding to the target object through a control unit and identifies the target object based on the multi-spectral combination. It can respond to changes in the external environment, such as light sources or hardware modules, and effectively distinguish between the target object and the reference object in different external environments.

[0130] An embodiment of the present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the steps of the method for determining the multi-spectral combination as described above, or implements the steps of the object recognition method as described above.

[0131] According to the electronic device of an embodiment of the present invention, the multispectral combination determined by the above-mentioned method for determining a multispectral combination or object recognition method can respond to changes in the external environment, such as light sources or hardware modules, and effectively distinguish between objects and reference objects in different external environments.

[0132] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the computer program implements the steps of the method for determining a multi-spectral combination as described above, or implements the steps of the method for object recognition as described above.

[0133] According to the computer-readable storage medium of an embodiment of the present invention, the multispectral combination determined by the above-mentioned method for determining a multispectral combination or object recognition method can respond to changes in the external environment, such as light sources or hardware modules, to effectively distinguish between objects and control objects in different external environments.

[0134] It should be noted that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic device), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing it in another suitable manner if necessary, and then storing it in a computer memory.

[0135] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0136] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0137] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0138] In the present invention, unless otherwise specified or limited, the terms "installed," "connected," "connect," "fixed," etc. should be understood in a broad sense. For example, they can refer to fixed connection, detachable connection, or integration; mechanical connection, electrical connection; direct connection, or indirect connection through an intermediate medium; internal communication between two components, or interaction between two components, unless otherwise specified. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0139] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for determining a multi-spectral combination, characterized in that: include: Obtaining the spectral reflection intensity of the object according to the spectral reflectivity of the object and the spectral distribution of the light source; Performing non-negative matrix decomposition on the spectral reflection intensity to obtain a plurality of basis functions of the spectral reflection intensity; Obtaining the maximum energy value of each basis function and the wavelength corresponding to the maximum energy value; Obtaining the ratio of the maximum energy of each basis function to the maximum energy of all basis functions respectively; According to the ratio, obtaining the characteristic wavelength from the wavelength corresponding to the energy maximum value; A multi-spectral combination for identifying the object is obtained according to the characteristic wavelength.

2. The method for determining a multi-spectral combination according to claim 1, wherein: The spectral reflectivity is the spectral reflectivity of the object within the full wavelength band or a partial wavelength band.

3. The method for determining a multi-spectral combination according to claim 1, wherein: The obtaining of the spectral reflection intensity of the object according to the spectral reflectivity of the object and the spectral distribution of the light source includes: The spectral reflectivity and the spectral distribution are convolved to obtain the spectral reflection intensity.

4. The method for determining a multi-spectral combination according to claim 1, wherein: The step of obtaining the characteristic wavelength from the wavelength corresponding to the energy maximum value according to the ratio includes: The ratios are sorted from large to small, and the wavelength corresponding to the ratio of the first number in the sorting order is obtained to obtain the characteristic wavelength; or the wavelength corresponding to the ratio greater than or equal to the first ratio is obtained to obtain the characteristic wavelength.

5. The method for determining a multi-spectral combination according to claim 1 or 4, characterized in that: The step of obtaining a characteristic wavelength of the object under the light source according to the spectral reflection intensity further includes: A wavelength corresponding to an energy in which the energy in each basis function is discontinuous and the difference between the energy and the corresponding energy in other basis functions exceeds a first threshold is obtained, and the wavelength is used as the characteristic wavelength.

6. The method for determining a multi-spectral combination according to claim 1, wherein: The acquiring, based on the characteristic wavelength, a multi-spectral combination for identifying the object includes: Performing curve fitting on each characteristic wavelength to obtain a spectrum segment corresponding to each characteristic wavelength; According to the spectral reflectance of the object, the spectral reflectance of the reference object, and the spectral distribution of the light source, obtaining from the spectral band a wavelength that maximizes the color difference between the object and the reference object; The wavelengths at which the color difference is maximized are combined to obtain the multi-spectral combination.

7. The method for determining a multi-spectral combination according to claim 6, wherein: The performing curve fitting on each characteristic wavelength to obtain a spectrum segment corresponding to each characteristic wavelength includes: Perform normal distribution curve fitting or square wave curve fitting on each characteristic wavelength to obtain a spectrum segment corresponding to each characteristic wavelength.

8. The method for determining a multi-spectral combination according to claim 6, wherein: The step of obtaining, from the spectrum segment, a wavelength that maximizes the color difference between the object and the reference object based on the spectral reflectance of the object, the spectral reflectance of the reference object, and the spectral distribution of the light source comprises: According to the spectral reflectance of the object, the spectral reflectance of the reference object, and the spectral distribution of the light source, a wavelength that maximizes the color difference between the object and the reference object is obtained from the spectrum using one of a recursive method, an exhaustive method, and a brute force method.

9. A device for determining a multi-spectral combination, characterized in that: include: a wavelength calculation unit, configured to obtain the spectral reflection intensity of the object based on the spectral reflectivity of the object and the spectral distribution of the light source; Performing non-negative matrix decomposition on the spectral reflection intensity to obtain a plurality of basis functions of the spectral reflection intensity; Obtaining the maximum energy value of each basis function and the wavelength corresponding to the maximum energy value; Obtaining the ratio of the energy maximum of each basis function to the energy maximum of all basis functions respectively; obtaining the characteristic wavelength from the wavelength corresponding to the energy maximum according to the ratio; A multispectral calculation unit is used to obtain a multispectral combination for identifying the object according to the characteristic wavelength.

10. An object recognition method, characterized in that: include: Obtaining a multispectral combination corresponding to a target object based on the method for determining a multispectral combination according to any one of claims 1 to 8, wherein the multispectral combination is obtained based on characteristic wavelengths of the target object under a preset spectral distribution; The target object is identified according to the multi-spectral combination.

11. The object recognition method according to claim 10, wherein: The characteristic wavelength is obtained based on the energy of each basis function and the wavelength corresponding to the energy in multiple basis functions obtained by performing non-negative matrix decomposition on the spectral reflection intensity of the object under the preset spectral distribution. The characteristic wavelength includes the wavelength corresponding to the first ratio of the ratio of the maximum energy of a single basis function to the maximum energy of all basis functions, the wavelength corresponding to the ratio of the first number sorted from large to small, and the wavelength corresponding to the energy in each basis function that is discontinuous and the difference between the energy and the corresponding energy in other basis functions exceeds a first threshold.

12. The object recognition method according to claim 11, characterized in that: The spectral reflection intensity is obtained based on the spectral reflectivity of the target object and the preset spectral distribution.

13. The object recognition method according to any one of claims 10 to 12, characterized in that: The multi-spectral combination is obtained based on the spectral reflectance of the target object, the spectral reflectance of the reference object and the preset light source distribution, and is obtained from the corresponding spectral band obtained by curve fitting the characteristic wavelength to obtain the wavelength that maximizes the color difference between the target object and the reference object.

14. An object recognition device, characterized in that: include: An acquiring unit, configured to acquire a multispectral combination corresponding to a target object based on the method for determining a multispectral combination according to any one of claims 1 to 8; An identification unit is configured to identify the target object according to the multi-spectral combination, wherein the multi-spectral combination is obtained based on a characteristic wavelength of the target object under a preset spectral distribution.

15. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the processor implements the steps of the method for determining a multi-spectral combination according to any one of claims 1 to 8, or implements the steps of the object recognition method according to any one of claims 10 to 13.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program implements the steps of the method for determining a multi-spectral combination according to any one of claims 1 to 8, or implements the steps of the object recognition method according to any one of claims 10 to 13.

Images