Clothing material identification system and clothing material identification method

Abstract

The invention discloses a clothing material identification system and a clothing material identification method, the clothing material identification system comprises an application layer, the application layer comprises a reflective placement part, a light emitting part, an identification device and ironing equipment, the reflective placement part is used for placing clothing, the light emitting part projects light to the clothing arranged on a reflective part, and the identification device is used for identifying the clothing on the reflective part; the recognition device is configured to collect spectral information of reflected light from the clothes and / or the reflective placement piece and recognize the material of the clothes, and the ironing equipment is used for adjusting the working mode according to the recognition result of the recognition device; and the edge computing layer comprises an equipment management module, a data preprocessing module, a model development module and a data storage module. According to the clothing material identification system, the clothing material can be accurately identified, and then the ironing equipment is adjusted to the working mode corresponding to the material to work, so that the ironing efficiency and quality are improved, and the clothing can be protected during ironing.

Description

Technical Field

[0001] The present invention relates to the field of optical technology, and in particular to a clothing material recognition system and a clothing material recognition method. Background Art

[0002] In daily life, more and more people use ironing machines and other clothing ironing equipment to iron out wrinkles on clothes. However, clothes are made of a variety of materials, including natural cotton, linen, silk, wool, cashmere, etc., as well as artificial synthetic materials such as denim, chemical fiber, polyester, blended fabrics, leather, woolen fabrics, etc. Moreover, with the advancement of technology, new materials continue to emerge, and the performance of each material is different. When ironing clothes, you need to choose different ironing gears or ironing modes according to different clothing materials.

[0003] In the prior art, the ironing machine requires the user to bend over and manually operate the control button several times to adjust the desired gear. Especially when ironing a lot of clothes, it is necessary to repeatedly operate between the gears according to different clothing materials, which makes the operation time-consuming and laborious, and the user is prone to make mistakes when he does not understand the material. Therefore, there is a need for improvement in the ironing machine and the ironing method in the prior art. Summary of the invention

[0004] The present invention proposes a clothing material recognition system and a clothing material recognition method. By establishing a clothing material recognition model, when the ironing device is working, the spectral information in the reflected light of the clothing material is processed to judge the clothing material, and then the ironing device is automatically adjusted to a working mode suitable for the material according to the clothing material, which can overcome the above-mentioned technical problems.

[0005] According to the first aspect of the present invention, the clothing material identification system includes: an application layer, the application layer includes a reflective mounting member, a light-emitting member, an identification device and an ironing device, the reflective mounting member is used to place clothing, the light-emitting member projects light to the clothing arranged on the reflective portion, the identification device is configured to collect spectral information of reflected light from the clothing and / or the reflective mounting member, and identify the clothing material, the ironing device is used to adjust the working mode according to the identification result of the identification device; an edge computing layer, the edge computing layer includes a device management module, a data preprocessing module, a model development module and a data storage module, the device management module is used to control the operation of the application layer, the data preprocessing module is used to preprocess the spectral information, the model development module is used to establish a clothing material identification model according to the preprocessed data, and the data storage module is used to store data of each layer.

[0006] According to the clothing material identification system of the present invention, by establishing a clothing material identification model, when the ironing device is working, the spectral information in the reflected light of the clothing material is processed to determine the clothing material, and then the ironing device is automatically adjusted to a working mode suitable for the material according to the clothing material, which can improve the ironing efficiency and quality, and also help to better protect the clothing, and is suitable for a variety of scenarios including ironing clothes in large quantities. In addition, it can avoid the problem of inefficiency and easy errors when the operator adjusts the working mode based on experience.

[0007] In some embodiments, the application layer and the edge computing layer are integrated into the ironing device.

[0008] In some embodiments, the reflective mounting member is made of a material with high reflectivity; and / or, the reflective mounting member includes a support member and a high reflective film covering the support member.

[0009] In some embodiments, the clothing material identification system further includes: a cloud service layer, the cloud service layer is communicatively connected to the edge computing layer, and the cloud service layer is used for data management, storage and optimization.

[0010] In some embodiments, the identification device is a spectral device, which includes a spectral chip, and the spectral chip includes a stacked filter structure, an image sensor and a data processor, wherein the filter structure is arranged on the photosensitive path of the image sensor to modulate the reflected light; the image sensor is used to receive the modulated spectrum and provide a differential response to the spectrum; the data processor is used to recover and calculate the differential response.

[0011] In some embodiments, the identification device further includes: a light homogenizer, which is disposed on the light incident side of the spectral device and located on the optical axis of the spectral chip to homogenize the reflected light.

[0012] In some embodiments, the model development module is used to establish a clothing material recognition model library based on the preprocessed data, specifically comprising the following steps:

[0013] Respectively obtain the reference spectrum data I1 of the reflective mounting member and the material spectrum data In of the clothing, 1≤n, where n is the total number of clothing samples;

[0014] Calculate the absorbance An of the clothing, Among them, the dark current signal I dark is a signal obtained by the recognition device in the absence of incident light;

[0015] Perform spectral preprocessing based on the absorbance An to obtain spectral data Bn of clothing classification of different materials;

[0016] A clothing material recognition model library is established based on the classified spectral data Bn obtained by preprocessing.

[0017] In some embodiments, the method for establishing the clothing material recognition model library includes: the edge computing layer first uses PCA to reduce the dimension of the pre-processed spectral data, and then calculates Hotelling T2 and SPE thresholds. The algorithm principle is as follows:

[0018] Hotelling 2 It is used to describe the distance of the sample from the PCA model, that is, the distance from the midpoint on the principal component projection hyperplane:

[0019]

[0020] where t i is the score vector of sample i, λ is the diagonal matrix of eigenvalues ​​(λ1 to λ k ), T 2 and the Mahalanobis distance are equivalent here, and then according to the residuals Q and T of all training set samples in the PCA model 2 The distribution of the statistic is estimated, and the corresponding confidence limits are calculated. Here, T 2 The calculation formula for the statistical confidence limit is as follows:

[0021]

[0022] The residual Q is the residual between the sample and the projection on the k principal components in the PCA model, that is, the part of the sample that is not explained by the model, which is used to evaluate the degree of fit between the sample and the given model.

[0023] e=x-tp T = x[I n -pp T ]

[0024] and

[0025]

[0026] Among them, e is the residual, t is the score after the principal component dimension reduction, p is the load after the principal component dimension reduction, In is the unit vector, and the threshold value of Q is calculated as follows

[0027]

[0028] in,

[0029] for i=1,2,3

[0030] and

[0031]

[0032] When the model predicts, the class closest to the sample is defined as,

[0033]

[0034] in

[0035]

[0036]

[0037] , d ij is the judgment threshold.

[0038] The second aspect of the present invention provides a method for identifying clothing materials, the method comprising the following steps: obtaining material spectrum data Im of the clothing to be identified, 1≤m, where m is the number of clothing to be identified; calculating the absorbance Am of the clothing, Among them, the dark current signal I dark is the signal obtained by the recognition device in the absence of incident light, I1 is the reference spectrum data of the reflective mounting member; based on the absorbance Am, the material of the clothing to be identified is determined.

[0039] In some embodiments, the material of the clothing to be identified based on the absorbance Am specifically includes the following steps: performing spectral preprocessing based on the absorbance Am to obtain preprocessed spectral data Bm of the clothing to be identified; comparing the preprocessed spectral data Bm with the clothing classification spectral data Bn of the clothing material identification model library to determine the material of the clothing to be identified.

[0040] In some embodiments, the comparing the preprocessed spectral data Bm with the clothing classification spectral data Bn of the clothing material recognition model library to determine the material of the clothing to be recognized specifically includes the following steps: if the preprocessed spectral data Bm falls into the clothing classification spectral data Bn of the clothing material recognition model library, then the clothing to be recognized is determined to be of the material corresponding to Bn; if the preprocessed spectral data Bm falls into at least two clothing classification spectral data Bn of the clothing material recognition model library, then the closest material of the clothing to be recognized is judged according to the correlation coefficient, and the judgment result is output, and the user confirms the judgment result; if the preprocessed spectral data Bm does not fall into the clothing material recognition model library, the user is prompted to manually output the corresponding material name, and the clothing material recognition model library is updated.

[0041] In some embodiments, the method of determining the closest material to the clothing to be identified based on the correlation coefficient and outputting the determination result, which is confirmed by the user, specifically includes the following steps: if the user confirms that the determination result is correct, the clothing material identification model library is updated; if the user confirms that the determination result is incorrect, the user is prompted to manually output the corresponding material name, and the clothing material identification model library is updated.

[0042] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a flow chart of establishing a clothing material recognition model library according to an embodiment of the present invention;

[0044] Figure 2 is a schematic flow chart of a method for identifying clothing material according to an embodiment of the present invention;

[0045] Figure 3 is a schematic diagram of the step of determining the material of the clothing to be identified based on the absorbance Am in the clothing material identification method according to an embodiment of the present invention;

[0046] Figure 4 is a flowchart of clothing material identification according to a specific embodiment of the present invention;

[0047] Figure 5 is a schematic diagram of a clothing material identification system according to the present invention;

[0048] Figure 6 is a schematic diagram of an application layer of a clothing material recognition system according to the present invention;

[0049] Figure 7 is a schematic diagram of an edge computing layer of a clothing material recognition system according to the present invention;

[0050] Figure 8 is a schematic structural diagram of a spectrum chip of a spectrum device according to an embodiment of the present invention;

[0051] Fig. 9 Schematic diagram of the structure of the filter structure of the spectral chip according to an embodiment of the present invention.

[0052] Reference numerals:

[0053] Clothing material recognition system 1000:

[0054] Application layer 100,

[0055] Identification device 10,

[0056] Filter structure 1, modulation unit 11, modulation micro-nano structure 12, image sensor 2, data processor 3,

[0057] Reflective mounting member 20, luminous member 30, ironing device 40,

[0058] Edge computing layer 200,

[0059] Equipment management module 201, data preprocessing module 202, model development module 203, data storage module 204,

[0060] Cloud service layer 300. DETAILED DESCRIPTION

[0061] Embodiments of the present invention are described in detail below, 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 should not be construed as limiting the present invention.

[0062] In daily life, more and more people use ironing machines and other clothing ironing equipment to iron out wrinkles on clothes. However, clothes are made of a variety of materials, including natural cotton, linen, silk, wool, cashmere, etc., as well as artificial synthetic materials such as denim, chemical fiber, polyester, blended fabrics, leather, woolen fabrics, etc. Moreover, with the advancement of technology, new materials continue to emerge, and the performance of each material is different. When ironing clothes, you need to choose different ironing gears or ironing modes according to different clothing materials.

[0063] In the prior art, the ironing machine requires the user to bend over and manually operate the control button several times to adjust the desired gear. Especially when ironing a lot of clothes, it is necessary to repeatedly operate between the gears according to different clothing materials, which makes the operation time-consuming and laborious, and the user is prone to make mistakes when he does not understand the material. Therefore, there is a need for improvement in the ironing machine and the ironing method in the prior art.

[0064] In view of this, the present invention provides a clothing material recognition system and clothing material recognition method, which establishes a clothing material recognition model, processes the spectrum information in the reflected light of the clothing material when the ironing device is working, and judges the clothing material, and then automatically adjusts the ironing device to a working mode suitable for the material according to the clothing material, which can improve the ironing efficiency and quality, and also help to better protect the clothes, and is suitable for a variety of scenarios including ironing clothes in large quantities. In addition, it can avoid the problem of inefficiency and easy errors when the operator adjusts the working mode based on experience.

[0065] Reference below Figure 1-Figure 9A clothing material identification system 1000 according to an embodiment of the first aspect of the present invention is described.

[0066] The clothing material recognition system 1000 of the embodiment of the present invention can be used for clothing material recognition. Specifically, the clothing material recognition system 1000 includes: an application layer 100 and an edge computing layer 200 .

[0067] Among them, reference Figure 6 , the application layer 100 may include a reflective placement member 20, a light emitting member 30, an identification device 10 and an ironing device 40. The reflective placement member 20 may be used to place clothes. For example, the reflective placement member 20 may be constructed as an ironing workbench of an ironing machine, or may be constructed as other platform members suitable for flattening clothes. The light emitting member 30 is a light source. The light emitting member 30 may be constructed to generate light of a specific wavelength band, or the light emitting member 30 may be constructed to generate light of a single wavelength. For example, the light emitting member 30 may be an LED lamp, a halogen lamp, etc. Preferably, the light emitting member 30 is a broad spectrum light source. The light emitting member 30 may generate infrared light, ultraviolet light, X-ray, visible light, etc., as well as a combination of light of different wavelength bands and different types. Of course, the present invention is not limited to this, and the type of the light emitting member 30 may be reasonably selected according to actual needs. Further, the application layer 100 may also include a discrimination module, which is used to output the material category of the clothes so that the user can obtain the material category.

[0068] The light emitting element 30 emits light to illuminate the clothes set on the reflective mounting element 20. After the light is irradiated to the clothes, it is diffusely reflected and generates reflected light. The identification device 10 is configured to collect spectral information of the reflected light from the clothes and / or the reflective mounting element 20 to identify the material of the clothes.

[0069] It should be noted that, when ironing clothes, the clothes are usually placed on a table. Since some clothes are easy to transmit light due to their material or thickness, the light emitted by the light-emitting element 30 to the clothes is partially reflected on the surface of the clothes, and another part may pass through the clothes to reflect on the table, or may be absorbed by the table. If the reflection is generated on the table, the spectral information of the table material will be mixed into the spectral data of the clothes, resulting in the inability to accurately judge the material of the clothes based on the spectral data. If the light is absorbed by the table, the corresponding energy loss will be lost, which is also not conducive to identification.

[0070] Therefore, in this embodiment, the clothing is placed on the reflective mounting member 20, and the reflective mounting member 20 has a high reflectivity, that is, a reflectivity greater than or equal to 90%, preferably greater than or equal to 95%, for example, it can be ceramic, whiteboard, etc.; the light generated by the light source can be partially reflected on the surface of the clothing, and partially pass through the clothing and reflect on the surface of the reflective mounting member 20 and then pass through the clothing again, and the identification device 10 is configured to receive the reflected light from the clothing and / or the reflective mounting member 20, and collect the spectral information of the reflected light. In other words, the identification device 10 can receive the reflected light from the clothing, and can also receive the reflected light from the reflective mounting member 20, and collect the spectral information of the reflected light, so as to identify the material of the clothing. The use of a reflective mounting member 20 with a high reflectivity can reduce the interference of the spectral information of the reflective mounting member 20 itself on the spectral information of the clothing.

[0071] The ironing device 40 may be an iron, a hanging iron, etc. The ironing device 40 may adjust the working mode according to the recognition result of the recognition device 10, for example, selecting the corresponding working mode, steam injection rate, steam injection amount, steam temperature, etc. according to the recognized clothing material.

[0072] refer to Figure 7 , the edge computing layer 200 includes a device management module 201, a data preprocessing module 202, a model development module 203 and a data storage module 204. Here, edge computing refers to an open platform that integrates network, computing, storage and application core capabilities on the side close to the object or data source, providing the nearest service nearby. Optionally, any one of the device management module 201, data preprocessing module 202, model development module 203 and data storage module 204 of the edge computing layer 200 can be integrated on the ironing device 40, or can be set independently of the ironing device 40, and communicate through electrical signals, Bluetooth, local area network, etc.

[0073] Among them, the device management module 201 can monitor terminal smart devices, issue instructions, control operations and query historical data, etc., so as to control the operation of the application layer 100. For example, the device management module 201 can control the operation of the light-emitting component 30 and the recognition device 10 to obtain the spectral information of the reflected light of the clothing to be identified and the reflective mounting component 20, and can also control the working mode, steam injection rate, steam injection amount, steam temperature, etc. of the ironing device 40.

[0074] The data preprocessing module 202 is used to perform data preprocessing on the spectral information of the reflected light obtained by the recognition device 10, and the model development module 203 is used to establish a clothing material recognition model based on the preprocessed data, so as to compare with the relevant data in the model library in the actual clothing material recognition process, and then determine the clothing material. The data storage module 204 can be used to store the data of each layer, that is, the data storage module 204 can record, store, query, etc. the information flow generated in the clothing material recognition process.

[0075] According to the clothing material identification system 1000 of the embodiment of the present invention, by establishing a clothing material identification model, when the ironing device 40 is working, the spectral information in the reflected light of the clothing material is processed to determine the clothing material, and then the ironing device 40 is automatically adjusted to a working mode suitable for the material according to the clothing material, which can improve the ironing efficiency and quality, and also help to better protect the clothing, and is suitable for a variety of scenarios including ironing a large number of clothes. In addition, it can avoid the problem of low efficiency and easy errors when the operator adjusts the working mode based on experience.

[0076] In some embodiments, the light emitting element 30 and the identification device 10 are both disposed on the ironing device 40 , so that the clothing material identification system 1000 is more integrated and is convenient to use and move as a whole.

[0077] Alternatively, in other embodiments, the light emitting element 30 and the identification device 10 can be placed on an ironing table, such as a desktop. After placing the clothes on the ironing table, the reflective mounting element 20 can be pressed onto the clothes. At this time, the light emitting element 30 projects light onto the clothes, a portion of the light is directly reflected by the clothes, and another portion of the light passes through the clothes and reaches the reflective mounting element 20 and is reflected by the reflective mounting element 20. The reflected light can be received by the identification device 10 to collect the spectral information of the reflected light, and then identify the material of the clothes. After the identification is completed, the reflective mounting element 20 is removed from the clothes, and the ironing device 40 can be ironed in the corresponding working mode. In this way, the light emitting element 30 and the identification device 10 are kept away from the high temperature of the ironing device 40, which helps to extend the service life of the light emitting element 30 and the identification device 10 to a certain extent and improve the reliability of the system 1000.

[0078] In some embodiments, the application layer 100 and the edge computing layer 200 can be integrated into the ironing device 40, so that data processing and identification in the clothing material identification process can be performed on the ironing device 40, thereby realizing localized processing of clothing material identification, which is more efficient and has lower usage costs.

[0079] In some embodiments, the reflective mounting member 20 is a highly reflective material member, that is, the reflective mounting member 20 may be made of a highly reflective material, for example, the reflective mounting member 20 may be a metal member, or may be made of other highly reflective materials.

[0080] Alternatively, the reflective mounting member 20 may include a support member and a high reflective film covering the support member, for example, the high reflective film may be a metal reflective film, a full dielectric reflective film, or a metal dielectric reflective film formed by a combination of the two, etc. In this way, it is possible to ensure that the reflective mounting member 20 has a high reflectivity, which is conducive to improving the accuracy of clothing material recognition.

[0081] In some embodiments, reference Figure 5 The clothing material recognition system 1000 also includes: a cloud service layer 300, which is communicatively connected to the edge computing layer 200, and the cloud service layer 300 includes a cloud server. The cloud service layer 300 is used for data management, storage, model optimization, and model distribution. For example, the cloud service layer 300 can store a clothing material recognition model library, that is, the model development module 203 of the edge computing layer 200 can upload the clothing material recognition model library to the cloud server, and the cloud service layer 300 can also optimize the model according to the new material recognized by the system 1000 or the new material manually input by the user, and store and manage the data in the recognition process, and can also distribute the model library to terminal devices such as ironing equipment 40 for real-time recognition.

[0082] In some embodiments, the identification device 10 can be a spectroscopic device, for example, the spectroscopic device can be a dispersion spectrometer, a Fourier spectrometer, a narrow-band spectrometer (for example, a spectrometer constructed using an FP cavity), or it can be a computational spectrometer in the field of spectral calculation. The identification device 10 can be reasonably selected according to actual needs.

[0083] refer to Figure 8 Taking a computational spectroscopy device as an example, the spectroscopy device may include a spectroscopy chip, which includes a stacked filter structure 1, an image sensor 2, and a data processor 3, wherein the filter structure 1 is a broadband filter structure 1 in the frequency domain or wavelength domain. The pass spectra of different wavelengths of the filter structures 1 in each place are not exactly the same. The filter structure 1 may be a structure or material with light scattering characteristics such as a metasurface, a photonic crystal, a nanocolumn, a multilayer film, a dye, a quantum dot, a MEMS (micro-electromechanical system), an FP etalon, a cavity layer, a waveguide layer, a diffraction element, a grating element, etc. The filter structure 1 is arranged on the light-sensitive path of the image sensor 2, that is, the filter structure 1 is arranged on the light-incident side of the image sensor 2 to modulate the reflected light.

[0084] The image sensor 2 may be a CMOS image sensor 2 (CIS), a CCD, an array photodetector or other sensors. The image sensor 2 is used to receive the spectrum modulated by the filter structure 1 and provide a differential response to the spectrum.

[0085] The data processing unit may be at least one of MCU, CPU, GPU, FPGA, NPU, ASIC, etc., which may export the data generated by the image sensor 2 to the outside for processing, thereby realizing the recovery calculation of the differential response to obtain the original spectrum.

[0086] In some embodiments, the filter structure 1 and the image sensor 2 are integrally formed. For example, the filter structure 1 can be formed in the process of manufacturing the image sensor 2, without having to form the filter structure 1 on the basis of the image sensor 2 after obtaining the image sensor 2. In this way, the manufacturing process of the spectral chip can be simplified.

[0087] In some embodiments, the identification device 10 may further include: a light homogenizer. Specifically, the light homogenizer may be implemented as a light homogenizer, which is arranged on the light incident side of the filter structure 1 and on the optical axis of the spectral chip to homogenize the reflected light, thereby improving the uniformity of the light incident on the spectral chip, and facilitating the accuracy of the reflected light processing result.

[0088] In some embodiments, reference Figure 8 and Fig. 9 The filter structure 1 is a light modulation layer, which includes a plurality of modulation units 11. Each modulation unit 11 includes a plurality of modulation micro-nano structures 12. The modulation micro-nano structures 12 are at least one of a hole structure, a column structure or a line structure.

[0089] For example Fig. 9 As shown, the modulating micro-nano structure 12 is formed as a modulation hole, and the specific cross-sectional shape of the modulation hole in each modulation unit 11 can be one of a circle, an ellipse, a cross, a regular polygon, a star or a rectangle, or any combination of the above shapes. The modulation holes in the same modulation unit 11 on the optical modulation layer are arranged into a two-dimensional graphic structure with a specific arrangement rule. For example, all the modulation holes are arranged in an array in a gradual order of the size of the structural parameters to form a two-dimensional graphic structure. The structural parameters of the modulation hole include inner diameter, major axis length, minor axis length, rotation angle, number of angles or side length, etc. Different two-dimensional graphic structures are used to achieve the modulation effect on light of different wavelengths. The difference in the two-dimensional graphic structure can also be used to improve the difference in spectral response between different regions, thereby improving the accuracy of spectral analysis.

[0090] In some embodiments, the data processor 3 is used to perform recovery calculation on the differential response, specifically including:

[0091] The intensity signal of the reflected light of different wavelengths λ is denoted as x(λ), the number of modulation units 11 on the light modulation layer is denoted as m, the number of effective transmission spectra of the light modulation layer is denoted as Ti(λ) (i=1,2,3,…,m), the differential response of the image sensor 2 is denoted as R(λ), and the transmission spectrum curve of the light modulation layer is denoted as T(λ).

[0092] The transmission spectrum of each modulation unit 11 is different from each other. There is a corresponding physical pixel under each modulation unit 11 to detect the light intensity bi modulated by the light modulation layer. In some possible embodiments, one physical pixel may correspond to one modulation unit 11. Of course, the present invention is not limited to this. In other embodiments, multiple physical pixels may be a group, and the group corresponds to one modulation unit 11. In other words, multiple physical pixels correspond to one modulation unit 11.

[0093] In the spectral device of the present embodiment, at least two modulation units 11 constitute a spectral pixel (that is, at least two modulation units 11 and the corresponding image sensor 2 together constitute a spectral pixel). It should be emphasized that the number of effective transmission spectra Ti(λ) of the light modulation layer may be inconsistent with the number of modulation units 11. The effective transmission spectrum here refers to the transmission spectrum used for spectral recovery. The transmission spectrum of the modulation unit 11 can be artificially set, tested, or calculated according to certain rules according to the needs of identification or recovery. The transmission spectrum of each modulation unit 11 obtained through testing is the effective transmission spectrum. Therefore, the number of effective transmission spectra of the light modulation layer can be the same as the number of modulation units 11, or different, and can be less than the number of modulation units 11, or more than the number of modulation units 11.

[0094] In some possible embodiments, the transmission spectrum curve may be determined by one or more modulation units 11. Furthermore, in this embodiment, at least one spectrum pixel may be used to restore the spectrum.

[0095] Then, the relationship between the spectral distribution of the incident light and the measurement value of the image sensor 2 can be expressed by the following formula:

[0096] bi=∫x(λ)*Ti(λ)*R(λ)dλ,

[0097] Discretize to get: bi = Σ(x(λ)*Ti(λ)*R(λ)),

[0098] Among them, note: Ai(λ)=Ti(λ)*R(λ),

[0099] The above formula can be expanded into matrix form:

[0100]

[0101] Among them, bi (i = 1, 2, 3, ..., m) is the response of the image sensor 2 after the reflected light to be measured passes through the light modulation layer, which corresponds to the light intensity measurement value of the image sensor 2 corresponding to the m modulation units 11. When one physical pixel corresponds to one modulation unit 11, it can be understood as the light intensity measurement value corresponding to the m physical pixels, which is a vector with a length of m. Matrix A (which can be understood as a transmission spectrum curve, or also called a transmission spectrum matrix) is the system's response to light of different wavelengths, which is determined by two factors: the transmittance of the light modulation layer and the quantum efficiency of the image sensor 2. A is a matrix, and each row vector corresponds to the response of a modulation unit 11 to incident light of different wavelengths. Here, the incident light is discretely and uniformly sampled, and there are n sampling points in total. The number of columns of A is the same as the number of sampling points of the incident light. Here, x (λ) is the light intensity of incident light of different wavelengths λ, and is also the incident light spectrum to be measured. On the basis of the above implementation method, the spectral pixels are arrayed, and a snapshot spectral imaging device can be realized.

[0102] In some embodiments, reference Figure 1 The model development module 203 is used to establish a clothing material recognition model library based on the preprocessed data, which specifically includes the following steps:

[0103] S11, respectively obtaining reference spectrum data I1 of the reflective mounting member 20 and material spectrum data In of the clothing, 1≤n, where n is the total number of clothing samples;

[0104] S12, calculate the absorbance Am of the clothing, Among them, the dark current signal I dark is a signal obtained by the recognition device 10 in the absence of incident light;

[0105] S13, performing spectral preprocessing based on the absorbance An to obtain spectral data Bn of clothing classification of different materials;

[0106] S14, establishing a clothing material recognition model library based on the classified spectral data Bn obtained through preprocessing.

[0107] Specifically, first, the light-emitting element 30 is turned on, and the light-emitting element 30 emits light to illuminate the reflective mounting element 20, generating a first reflected light, and the recognition device 10 receives the first reflected light and obtains the spectrum data of the first reflected light, that is, the reference spectrum data I1 of the reflective mounting element 20; then the clothes are placed on the reflective mounting element 20, and the light-emitting element 30 emits light to illuminate the clothes again, generating a second reflected light, and the recognition device 10 receives the second reflected light and obtains the spectrum data of the second reflected light, that is, the material spectrum data In of the clothes. According to the obtained signal, the corresponding absorbance An is calculated, where, Among them, the dark current signal I darkIt is the signal obtained by the identification device 10 when there is no incident light. If there is no dark current, Idark is a 0 vector of the same length as In. The data preprocessing module 202 can preprocess the absorbance An to obtain the classification spectrum data Bn of clothing of different materials. Then, the model development module 203 establishes a clothing material recognition model library for clothing of different materials according to the classification spectrum data Bn obtained by the preprocessing. The data storage module 204 can store and manage the data generated by the identification device, the data preprocessing module 202, the model development module 203, etc. during the establishment of the clothing material recognition model library and the clothing material recognition process.

[0108] Furthermore, when classifying clothing materials, they can be divided into B1-Bn, n major categories, and then each major category can be subdivided into i small categories. Finally, each material can be classified and recorded as Bni.

[0109] In some embodiments, the method for establishing a clothing material recognition model library includes: the edge computing layer 200 or the cloud service layer first uses PCA to reduce the dimension of the pre-processed spectral data, and then calculates Hotelling T2 and SPE (Squared Prediction Error) thresholds as follows:

[0110] Hotelling 2 It is used to describe the distance of the sample from the PCA model, that is, the distance from the midpoint on the principal component projection hyperplane:

[0111]

[0112] where t i is the score vector of sample i, λ is the diagonal matrix of eigenvalues ​​(λ1 to λ k ), T 2 and the Mahalanobis distance are preferably equivalent, and then according to the residuals Q and T of all training set samples in the PCA model 2 The distribution of the statistic is estimated, and the corresponding confidence limits are calculated. Here, T 2 The calculation formula for the statistical confidence limit is as follows:

[0113]

[0114] The residual Q is the residual between the sample and the projection on the k principal components in the PCA model, that is, the part of the sample that is not explained by the model, which is used to evaluate the degree of fit between the sample and the given model. m is the number of samples in each category, and α is the confidence level, where the confidence level is preferably above 95%.

[0115] e=x-tp T = x[I n -ppT ]

[0116] and

[0117]

[0118] Where e is the residual, t is the score after the principal component dimension reduction, p is the load after the principal component dimension reduction, In is the unit vector, and the threshold value of Q is calculated as follows

[0119]

[0120] in,

[0121] for i=1,2,3

[0122] and

[0123]

[0124] When the model predicts, the class closest to the sample is defined as,

[0125]

[0126] in

[0127]

[0128]

[0129] , d ij is the judgment threshold.

[0130] By d ij Whether the corresponding sample falls into the big category is determined by whether it exceeds the numerical range of the big category.

[0131] Reference below Figure 2-Figure 4 A method for identifying clothing material according to an embodiment of the second aspect of the present invention is described.

[0132] refer to Figure 2 The clothing material recognition method of this embodiment can be applied to the clothing material recognition system 1000 in the above embodiment. Specifically, the clothing material recognition method includes the following steps:

[0133] S101, obtaining material spectrum data Im of the clothing to be identified, 1≤m, where m is the number of clothing to be identified;

[0134] S102, calculate the absorbance Am of the clothing, Among them, the dark current signal I dark is the signal obtained by the recognition device 10 in the absence of incident light, and I1 is the reference spectrum data of the reflective mounting member 20;

[0135] S103: Identify the material of the clothing to be identified based on the absorbance Am.

[0136] Specifically, the clothing to be identified is placed on the reflective mounting member 20, and the light emitting member 30 emits light again to illuminate the clothing to be identified, generating a second reflected light. The identification device 10 receives the second reflected light and obtains the spectral data of the second reflected light, i.e., the material spectral data Im of the clothing to be identified. According to the obtained signal, the corresponding absorbance Am is calculated, where: Among them, the dark current signal I dark is the signal obtained by the recognition device 10 when there is no incident light. If there is no dark current, then I dark is a 0 vector of the same length as Im. In addition, since the reference spectral data I1 of the reflective mounting member 20 is obtained by processing the first reflected light of the reflective mounting member 20 during the establishment of the clothing material recognition model library, the reference spectral data I1 of the reflective mounting member 20 can be directly obtained from the server or the data storage module 204 of the edge computing layer 200; or, before the first clothing material recognition, the light-emitting member 30 can be turned on first, and the light-emitting member 30 emits light to illuminate the reflective mounting member 20, and the reflective mounting member 20 generates the first reflected light, and then the recognition device 10 collects the spectral data of the first reflected light, and the subsequent clothing material recognition process uses the reference spectral data I1 of the reflective mounting member 20 obtained for the first time; or, the reference spectral data I1 of the reflective mounting member 20 can be obtained separately each time the clothing material is recognized.

[0137] After calculating the absorbance Am of the clothing to be identified according to the above formula, the material of the clothing can be determined based on the absorbance Am. For example, the calculated absorbance Am can be compared with the absorbance An in the clothing material recognition model library to determine the clothing material. Alternatively, spectral preprocessing can be performed based on the absorbance Am to obtain preprocessed spectral data Bm of the clothing to be identified, and then the preprocessed spectral data Bm can be compared with the clothing classification spectral data Bn in the clothing material recognition model library to determine the material of the clothing to be identified.

[0138] According to the clothing material identification method of the clothing material identification system 1000 of the embodiment of the present invention, based on the clothing material identification model library, when the ironing device 40 is working, the spectral information in the reflected light of the clothing material is processed to judge the clothing material, and then the ironing device 40 is automatically adjusted to a working mode suitable for the material according to the clothing material, which can improve the ironing efficiency and quality, and also help to better protect the clothing, and is suitable for a variety of scenarios including ironing clothes in large quantities. In addition, it can avoid the problem of inefficiency and easy errors when the operator adjusts the working mode based on experience.

[0139] Considering that some new or complex materials are different from the classified materials, the clothing material recognition system 1000 may not be able to accurately recognize them, and it may not be able to control the ironing device 40 to select the corresponding working mode. Therefore, the clothing material recognition system 1000 of this embodiment also has a prompting function. For example, the device layer may also be provided with a prompting device such as a pronunciation device, a light-emitting device, etc., and the service layer has a prompting module for issuing prompts. For example, the prompt may be in the form of voice, text, light or other information carriers, prompting the user to manually select the working mode. After the user manually selects the working mode, the recognition system 1000 can classify the material into the category corresponding to the working mode, so that the material can be automatically recognized and the corresponding mode can be selected the next time it is encountered.

[0140] refer to Figure 3 In some embodiments, based on the absorbance Am, the material of the clothing to be identified is identified, that is, step S103, which specifically includes the following steps:

[0141] S1031, performing spectral preprocessing based on the absorbance Am to obtain preprocessed spectral data Bm of the clothing to be identified;

[0142] S1032, comparing the pre-processed spectral data Bm with the clothing classification spectral data Bn in the clothing material recognition model library to determine the material of the clothing to be recognized.

[0143] Specifically, the absorbance Am is preprocessed to obtain preprocessed spectral data Bm of the clothing to be identified, the preprocessed spectral data Bm is compared with the clothing classification spectral data Bn of the clothing material identification model library, the material of the clothing to be identified is determined, and the ironing device 40 is controlled to select the corresponding working mode.

[0144] In some embodiments, the preprocessed spectral data Bm is compared with the clothing classification spectral data Bn of the clothing material recognition model library to determine the material of the clothing to be identified, which specifically includes the following steps: if the preprocessed spectral data Bm falls into the clothing classification spectral data Bn of the clothing material recognition model library, it is determined that the clothing to be identified is of the material corresponding to Bn; if the preprocessed spectral data Bm falls into at least two clothing classification spectral data Bn of the clothing material recognition model library, the closest material of the clothing to be identified is judged according to the correlation coefficient, and the judgment result is output, and the user confirms the judgment result; if the preprocessed spectral data Bm does not fall into the clothing material recognition model library, the user is prompted to manually output the corresponding material name to update the clothing material recognition model library.

[0145] In this way, when the clothing material recognition model library has archived the material corresponding to the clothing to be recognized, the material of the clothing to be recognized can be accurately recognized; when there are multiple related materials in the clothing material recognition model library, the user confirms the judgment result of the system 1000, which is conducive to improving the accuracy, thereby selecting the best working mode for the ironing device 40. In the case that the clothing material recognition model library does not record the material corresponding to the clothing to be recognized, the user can first determine the material and input it into the recognition system 1000, and the recognition system 1000 updates and improves the model library, thereby further improving the recognition accuracy of the clothing material recognition system 1000. In summary, the clothing material recognition system 1000 of this embodiment can actively optimize the clothing material recognition model library, which is conducive to improving the accuracy of the clothing material recognition result.

[0146] In some embodiments, the closest material to the clothing to be identified is determined based on the correlation coefficient, and the judgment result is output, and the user confirms the judgment result, which specifically includes the following steps: if the user confirms that the judgment result is correct, the clothing material recognition model library is updated; if the user confirms that the judgment result is incorrect, the user is prompted to manually input the corresponding material name, and the clothing material recognition model library is updated.

[0147] In this way, when there are multiple related materials in the model library, the user can judge the material and update and improve the model library to ensure the accuracy of subsequent recognition.

[0148] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0149] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0150] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, an electrical connection, or a communication; it can be a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0151] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0152] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A clothing material recognition system, characterized in that: include: an application layer, the application layer comprising a reflective mounting member, a light emitting member, an identification device and an ironing device, the reflective mounting member being used to mount clothing, the light emitting member projecting light to clothing disposed on the reflective portion, the identification device being configured to collect spectral information of reflected light from the clothing and / or the reflective mounting member, and identifying clothing material, and the ironing device being used to adjust a working mode according to an identification result of the identification device; The edge computing layer includes a device management module, a data preprocessing module, a model development module and a data storage module. The device management module is used to control the operation of the application layer. The data preprocessing module is used to perform data preprocessing on the spectral information. The model development module is used to establish a clothing material recognition model based on the preprocessed data. The data storage module is used to store data of each layer.

2. The clothing material identification system according to claim 1, characterized in that: The application layer and the edge computing layer are integrated into the ironing device.

3. The clothing material identification system according to claim 1, characterized in that: The reflective mounting member is made of a material with high reflectivity; or, the reflective mounting member includes a support member and a high reflective film covering the support member.

4. The clothing material identification system according to claim 1, characterized in that: Also includes: A cloud service layer, the cloud service layer is communicatively connected to the edge computing layer, and the cloud service layer is used for data management, storage and optimization.

5. The clothing material identification system according to claim 1, characterized in that: The identification device is a spectral device, which includes a spectral chip, and the spectral chip includes a stacked filter structure, an image sensor, and a data processor. Wherein, the light filtering structure is arranged on the light sensing path of the image sensor to modulate the reflected light; The image sensor is used to receive the modulated spectrum and provide a differential response to the spectrum; The data processor is used for performing recovery calculation on the differential response.

6. The clothing material identification system according to claim 5, characterized in that: The identification device further includes: a light homogenizer, which is arranged on the light incident side of the spectral device and located on the optical axis of the spectral chip to perform homogenization processing on the reflected light.

7. The clothing material identification system according to any one of claims 1 to 6, characterized in that: The model development module is used to establish a clothing material recognition model library based on the preprocessed data, and specifically includes the following steps: Respectively obtain the reference spectrum data I1 of the reflective mounting member and the material spectrum data In of the clothing, 1≤n, where n is the total number of clothing samples; Calculate the absorbance An of the clothing, Among them, the dark current signal I dark is a signal obtained by the recognition device in the absence of incident light; Perform spectral preprocessing based on the absorbance An to obtain spectral data Bn of clothing classification of different materials; A clothing material recognition model library is established based on the classified spectral data Bn obtained by preprocessing.

8. The clothing material identification system according to claim 7, characterized in that: The method for establishing the clothing material recognition model library includes: The edge computing layer or cloud service layer first uses PCA to reduce the dimension of the preprocessed spectral data, and then calculates the Hotelling T2 and SPE thresholds as follows: Hotelling 2 It is used to describe the distance of the sample from the PCA model, that is, the distance from the midpoint on the principal component projection hyperplane: where t i is the score vector of sample i, λ is the diagonal matrix of eigenvalues ​​(λ1 to λ k ), T 2 and the Mahalanobis distance are equivalent here, and then according to the residuals Q and T of all training set samples in the PCA model 2 The distribution of the statistic is estimated, and the corresponding confidence limits are calculated. Here, T 2 The calculation formula for the statistical confidence limit is as follows: The residual Q is the residual between the sample and the projection on the k principal components in the PCA model, which is used to evaluate the degree of fit between the sample and the given model. m is the number of samples in each category, and α is the confidence level. e=x-tp T =x[I n -pp T ] and Among them, e is the residual, t is the score after the principal component dimension reduction, p is the load after the principal component dimension reduction, In is the unit vector, and the threshold value of Q is calculated as follows in, and When the model predicts, the class closest to the sample is defined as, in , d ij is the judgment threshold.

9. A method for identifying clothing material, characterized in that: The following steps are involved: Obtain material spectrum data Im of the clothes to be identified, 1≤m, where m is the number of clothes to be identified; Calculate the absorbance Am of the clothing, Among them, the dark current signal I dark is the signal obtained by the recognition device in the absence of incident light, and I1 is the reference spectrum data of the reflective mounting member; Based on the absorbance Am, the material of the clothing to be identified is determined.

10. The clothing material identification method according to claim 9, characterized in that: The method of identifying the material of the clothing to be identified based on the absorbance Am specifically includes the following steps: Performing spectral preprocessing based on the absorbance Am to obtain preprocessed spectral data Bm of the clothing to be identified; The preprocessed spectral data Bm is compared with the clothing classification spectral data Bn of the clothing material recognition model library to determine the material of the clothing to be recognized.

11. The clothing material identification method according to claim 10, characterized in that: The step of comparing the preprocessed spectral data B with the clothing classification spectral data Bn of the clothing material recognition model library to determine the material of the clothing to be recognized specifically includes the following steps: If the preprocessed spectrum data Bm falls into the clothing classification spectrum data Bn of the clothing material recognition model library, it is determined that the clothing to be recognized is of the material corresponding to Bn; If the preprocessed spectrum data Bm falls into at least two clothing classification spectrum data Bn of the clothing material recognition model library, the closest material of the clothing to be recognized is determined according to the correlation coefficient, and the determination result is output, and the user confirms the determination result; If the pre-processed spectral data Bm does not fall into the clothing material recognition model library, the user is prompted to manually input the corresponding material name to update the clothing material recognition model library.

12. The clothing material identification method according to claim 11, characterized in that: The method of determining the closest material of the clothing to be identified according to the correlation coefficient and outputting the determination result, and the user confirming the determination result, specifically includes the following steps: If the user confirms that the judgment result is correct, the clothing material recognition model library is updated; If the user confirms that the judgment result is incorrect, the user is prompted to manually input the corresponding material name, and the clothing material recognition model library is updated.

Images