Thin layer chromatography component analysis method and related equipment based on adaptive weight fusion

Abstract

The present invention discloses a thin layer chromatography component analysis method based on adaptive weight fusion and related equipment. The thin layer chromatography component analysis method based on adaptive weight fusion provided by the present invention performs feature vector extraction and spectral band extraction on the thin layer chromatography image of the sample to be analyzed and the thin layer chromatography image of the target sample with known components, uses feature vector matching and the relative position relationship of the chromatographic band as identification factors and inputs them into a neural network for component identification, integrates the feature point similarity and the spectral band position relationship similarity, and can improve the accuracy and efficiency of the thin layer chromatography component analysis results.

Description

Technical Field

[0001] The present invention relates to the technical field of chemical component analysis, and in particular to a thin layer chromatography component analysis method based on adaptive weight fusion and related equipment. Background Art

[0002] Thin layer chromatography component analysis refers to a technique in which a stationary phase is applied to an experimental plate and the corresponding solvent is used as the mobile phase to form a uniform thin layer to analyze the components of a sample. The chromatogram of the sample to be tested is compared with the chromatogram of the reference obtained by the same method. The chromatograms of the true sample and the reference will have the same spots at corresponding positions, and this is used to identify and analyze the components. It is widely used in the field of component analysis of medicinal materials.

[0003] In the existing TLC component analysis, manual comparison is often used. However, since the imaging of the chromatographic band of the TLC is greatly affected by external factors, the chromatographic band often has blurred spots, irregularities, and large differences in strength, which makes manual identification difficult and inefficient. In addition, due to the distortion of the actual chromatogram, the influence of various factors such as light and environmental humidity, manual judgment is also subjective. Depending on the knowledge reserves and identification experience of different personnel, the same chromatogram may produce multiple results, which also makes it difficult to effectively guarantee the accuracy and stability of manual comparison.

[0004] Therefore, the prior art still needs to be improved and enhanced. Summary of the invention

[0005] In view of the above-mentioned defects of the prior art, the present invention provides a thin layer chromatography component analysis method based on adaptive weight fusion, aiming to solve the problem of low efficiency of manual thin layer chromatography component analysis in the prior art.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is as follows:

[0007] A first aspect of the present invention provides a thin layer chromatography component analysis method based on adaptive weight fusion, the method comprising:

[0008] Acquire an image to be analyzed, wherein the image to be analyzed is a thin layer chromatography image of the sample to be analyzed, input the image to be analyzed into a preset filter, and extract each first feature vector in the image to be analyzed through the preset filter;

[0009] Matching the first feature vector with each template feature vector of the target sample, and obtaining a first similarity score between the sample to be analyzed and the target sample based on the matching result, wherein the template feature vector is a feature vector extracted from a thin layer chromatography image of the target sample;

[0010] Inputting the image to be analyzed into a trained first neural network, and obtaining each first spectral band in the image to be analyzed output by the first neural network;

[0011] Obtaining a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the respective first spectral bands and the relative positions between the respective template spectral bands of the target sample, wherein the template spectral band of the target sample is a spectral band extracted from the thin layer chromatography image of the target sample;

[0012] The first similarity score and the second similarity score of the sample to be analyzed and each of the target samples are input into a trained second neural network to obtain a component analysis result output by the second neural network.

[0013] The thin layer chromatography component analysis method based on adaptive weight fusion, wherein, before matching the first feature vector with each template feature vector of the target sample, the method further comprises:

[0014] Inputting the thin layer chromatography image of the target sample into the preset filter, and extracting each second feature vector in the thin layer chromatography image of the target sample through the preset filter;

[0015] At least one of the second feature vectors is selected according to the modulus of each of the second feature vectors as the template feature vector corresponding to the target sample.

[0016] The thin layer chromatography component analysis method based on adaptive weight fusion, wherein the matching of the first feature vector with each template feature vector of the target sample comprises:

[0017] Calculating the similarity between the first feature vector and each of the template feature vectors;

[0018] When the similarity between the first feature vector and the template feature vector is greater than a preset threshold, it is determined that the first feature vector matches the template feature vector.

[0019] The thin layer chromatography component analysis method based on adaptive weight fusion, wherein the step of obtaining a first similarity score between the sample to be analyzed and the target sample based on the matching result comprises:

[0020] The first similarity score is obtained according to the number of the first feature vectors that match the template feature vector of the target sample.

[0021] The thin layer chromatography component analysis method based on adaptive weight fusion, wherein the step of obtaining the second similarity score between the sample to be analyzed and the target sample based on the relative positions between the first spectral bands and the relative positions between the template spectral bands of the target sample comprises:

[0022] Calculating the difference in position coordinates between adjacent first spectral bands to obtain first relative position data between each of the first spectral bands;

[0023] Calculating the difference in position coordinates between adjacent template spectral bands to obtain second relative position data between the respective template spectral bands;

[0024] The second similarity score is acquired according to the first relative position data and the second relative position data.

[0025] The thin layer chromatography component analysis method based on adaptive weight fusion, wherein, before obtaining the second similarity score between the sample to be analyzed and the target sample based on the relative positions between the first spectral bands and the relative positions between the template spectral bands, the method further comprises:

[0026] The thin layer chromatography image of the target sample is input into the first trained neural network, and each of the template spectrum bands of the target sample output by the first neural network is obtained.

[0027] The thin layer chromatography component analysis method based on adaptive weight fusion, wherein the parameters of the preset filter are trained together with the parameters of the first neural network, and the training data includes multiple groups of sample data, each group of sample data includes a sample image to be analyzed, and the components of each target sample and the sample image corresponding to the sample image to be analyzed.

[0028] A second aspect of the present invention provides a thin layer chromatography component analysis device based on adaptive weight fusion, comprising:

[0029] A feature extraction module, wherein the feature extraction module is used to obtain an image to be analyzed, wherein the image to be analyzed is a thin layer chromatography image of a sample to be analyzed, input the image to be analyzed into a preset filter, and extract each first feature vector in the image to be analyzed through the preset filter;

[0030] a first similarity module, the first similarity module being used to match the first feature vector with each template feature vector of the target sample, and to obtain a first similarity score between the sample to be analyzed and the target sample based on the matching result, wherein the template feature vector is a feature vector extracted from a thin layer chromatography image of the target sample;

[0031] A spectral band extraction module, the spectral band extraction module is used to input the image to be analyzed into a trained first neural network, and obtain each first spectral band in the image to be analyzed output by the first neural network;

[0032] a second similarity module, the second similarity module being used to obtain a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the respective first spectral bands and the relative positions between the respective template spectral bands of the target sample, wherein the template spectral band of the target sample is a spectral band extracted from the thin layer chromatography image of the target sample;

[0033] A fusion module is used to input the first similarity score and the second similarity score of the sample to be analyzed and each of the target samples into a trained second neural network to obtain a component analysis result output by the second neural network.

[0034] The third aspect of the present invention provides a terminal, which includes a processor and a computer-readable storage medium communicatively connected to the processor, the computer-readable storage medium is suitable for storing multiple instructions, and the processor is suitable for calling the instructions in the computer-readable storage medium to execute the steps of implementing any of the above-mentioned thin layer chromatography component analysis methods based on adaptive weight fusion.

[0035] A fourth aspect of the present invention provides a computer-readable storage medium storing one or more programs, which can be executed by one or more processors to implement the steps of the thin layer chromatography component analysis method based on adaptive weight fusion as described in any of the above items.

[0036] Compared with the prior art, the present invention provides a thin layer chromatography component analysis method based on adaptive weight fusion and related equipment. In the thin layer chromatography component analysis method based on adaptive weight fusion provided by the present invention, feature vector extraction and spectral band extraction are performed on the thin layer chromatography image of the sample to be analyzed and the thin layer chromatography image of the target sample with known components. Feature vector matching and the relative position relationship of the chromatographic bands are used as identification factors and input into a neural network for component identification. The feature point similarity and the spectral band position relationship similarity are integrated, which can improve the accuracy and efficiency of the thin layer chromatography component analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flow chart of an embodiment of a thin layer chromatography component analysis method based on adaptive weight fusion provided by the present invention;

[0038] Figure 2A schematic diagram of feature vector matching in an embodiment of a thin layer chromatography component analysis method based on adaptive weight fusion provided by the present invention;

[0039] Figure 3 A schematic diagram of a second neural network in an embodiment of a thin layer chromatography component analysis method based on adaptive weight fusion provided by the present invention;

[0040] Figure 4 A structural principle diagram of an embodiment of a thin layer chromatography component analysis device based on adaptive weight fusion provided by the present invention;

[0041] Figure 5 A schematic diagram of the principles of an embodiment of a terminal provided by the present invention. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0043] The thin layer chromatography component analysis method based on adaptive weight fusion provided by the present invention can be applied to a terminal with computing capability. The terminal can execute the thin layer chromatography component analysis method based on adaptive weight fusion provided by the present invention to generate a component analysis result of a sample to be analyzed. The terminal can be, but is not limited to, various computers, mobile terminals, smart home appliances, wearable devices, etc.

[0044] Embodiment 1

[0045] like Figure 1 As shown, in one embodiment of the thin layer chromatography component analysis method based on adaptive weight fusion, the steps include:

[0046] S100 , obtaining an image to be analyzed, inputting the image to be analyzed into a preset filter, and extracting each first feature vector in the image to be analyzed through the preset filter.

[0047] The image to be analyzed is a thin layer chromatography image of the sample to be analyzed. In order to analyze the components of the sample to be analyzed, a thin layer chromatography image is firstly prepared according to the sample to be analyzed to obtain the image to be analyzed.

[0048] In this embodiment, feature points of the image to be analyzed are first extracted to obtain multiple feature vectors. Specifically, the image to be analyzed is input into a preset filter, and each feature point in the image to be analyzed is extracted through the preset filter. The feature value of each feature point is a feature vector. The preset filter can adopt an existing filter. In order to enable the preset filter to extract a more accurate feature vector, in this embodiment, the parameters of the preset filter are obtained through training. The preset filter can be trained separately. When trained separately, the training data used to train the preset filter can include sample images and annotation features corresponding to the sample images. The preset filter can also be trained together with other neural networks adopted in this method, which will be explained in detail later.

[0049] S200: Match the first feature vector with each template feature vector of the target sample, and obtain a first similarity score between the sample to be analyzed and the target sample based on the matching result.

[0050] The template feature vector is a feature vector extracted from the thin layer chromatography image of the target sample, the target sample is a sample of known components, there may be multiple target samples, and each target sample corresponds to multiple template feature vectors. Specifically, the thin layer chromatography image of the target sample is also input into the preset filter, and the template feature vector corresponding to the thin layer chromatography image of the target sample is extracted through the preset filter, that is, before matching the first feature vector with the template feature vector, the thin layer chromatography component analysis method based on adaptive weight fusion provided in this embodiment also includes the steps of:

[0051] Inputting the thin layer chromatography image of the target sample into the preset filter, and extracting each second feature vector in the thin layer chromatography image of the target sample through the preset filter;

[0052] At least one of the second feature vectors is selected according to the modulus of each of the second feature vectors as the template feature vector corresponding to the target sample.

[0053] That is to say, in this embodiment, a twin parallel filter is used to simultaneously extract features of the thin layer chromatography images of the sample to be analyzed and the target sample, which is highly efficient, and the extracted features can better extract the discrimination information of the sample and the template, thereby improving the matching accuracy.

[0054] Since the chromatographic image may have problems such as tilt, distortion, and bright area misalignment, in this embodiment, before the image to be analyzed and the thin layer chromatography image of the target sample are input into the preset filter, the image to be analyzed and the thin layer chromatography image of the target sample are also preprocessed. Specific preprocessing operations include but are not limited to operations such as full correction and length cutting, and the problems of tilt, distortion, and bright area misalignment of the chromatogram are solved as much as possible through preprocessing. After the image is preprocessed, the image is also scaled to a preset size, that is, the size of the image to be analyzed and the thin layer chromatography image of the target sample input into the preset filter is the same.

[0055] After obtaining each of the second feature vectors corresponding to the target sample, each of the second feature vectors can be directly used as a template feature vector. However, through the preset filter, multiple feature points and corresponding feature vectors can be extracted from an image. In order to improve efficiency and accuracy, in this embodiment, after obtaining each of the second feature vectors corresponding to the target sample, the second feature vectors are further screened to obtain the template feature vector. Specifically, the modulus of each of the second feature vectors corresponding to the target sample is obtained, and the K second feature vectors with the largest modulus are selected as the target feature vector corresponding to the target sample, K≥1. The significant point measurement method used in this embodiment pays more attention to the important areas in the thin layer chromatogram, which can more accurately reflect the similarity between the sample and the template and improve the recall rate.

[0056] like Figure 2 As shown, after obtaining each of the first feature vectors and the template feature vector of the target sample, the two are matched. Specifically, matching the first feature vector with each of the template feature vectors of the target sample includes:

[0057] Calculating the similarity between the first feature vector and each of the template feature vectors;

[0058] When the similarity between the first feature vector and the target feature vector is greater than a preset threshold, it is determined that the first feature vector matches the target feature vector.

[0059] Calculating the similarity between the first feature vector and each of the template feature vectors may be achieved by using a cosine similarity calculation method or other similarity calculation methods. For each of the first feature vectors, the similarity is calculated with each of the template feature vectors of the target sample. If the similarity between a first feature vector and a template feature vector is greater than a preset threshold, the two are determined to match. Furthermore, if the similarity between a first feature vector and multiple template feature vectors is greater than the preset threshold, the template feature vector with the greater similarity is selected as the template feature vector that matches the first feature vector, that is, it is determined that the first feature vector matches the template feature vector with the greater similarity and does not match the remaining template feature vectors.

[0060] Acquiring the first similarity score between the sample to be analyzed and the target sample according to the matching structure between the first feature vector and each target feature vector of the target sample specifically includes:

[0061] The first similarity score is obtained according to the number of the first feature vectors that match the template feature vector of the target sample.

[0062] That is, the more the first feature vectors that can be successfully matched with the template feature vector of the target sample, the higher the first similarity score.

[0063] Please refer again Figure 1 The thin layer chromatography component analysis method based on adaptive weight fusion provided in this embodiment further includes the steps of:

[0064] S300: input the image to be analyzed into a trained first neural network, and obtain each first spectral band in the image to be analyzed output by the first neural network.

[0065] In this embodiment, in addition to determining the similarity between the sample to be analyzed and the target sample by means of feature point matching, the similarity between the sample to be analyzed and the target sample is also determined in combination with the distribution characteristics of the spectral bands in the chromatogram. Specifically, in this embodiment, the first neural network is used to extract the key spectral bands in the chromatogram. Specifically, the first neural network is pre-trained using multiple groups of first training data, each group of the first training data includes a sample chromatogram and annotated spectral bands corresponding to the sample chromatogram, and the first training data can be obtained by manually annotating the key spectral bands in the sample chromatogram in the sample chromatogram. The structure of the first neural network can adopt the structure of an existing neural network, for example, the first neural network adopts a yolov3 detection network. The image to be analyzed is input into the trained first neural network, and the first neural network can extract the key spectral bands in the image to be analyzed, and the key spectral bands in the image to be analyzed extracted by the first neural network are called first spectral bands.

[0066] Please refer again Figure 1 , the method provided in this embodiment further includes the steps of:

[0067] S400 , obtaining a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the first spectrum bands and the relative positions between the template spectrum bands of the target sample.

[0068] The template spectral band of the target sample is a spectral band extracted from the thin layer chromatography image of the target sample. The thin layer chromatography image of the target sample is input into the first neural network, and the first neural network can extract the key spectral band in the thin layer chromatography image of the target sample. The key spectral band in the thin layer chromatography image of the target sample extracted by the first neural network is called the template spectral band of the target sample. That is, before obtaining the second similarity score between the sample to be analyzed and the target sample based on the relative positions between the first spectral bands and the relative positions between the template spectral bands, the method further includes:

[0069] The thin layer chromatography image of the target sample is input into the first trained neural network, and each of the template spectrum bands of the target sample output by the first neural network is obtained.

[0070] The step of obtaining a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the first spectral bands and the relative positions between the template spectral bands of the target sample comprises:

[0071] Calculating the difference in position coordinates between adjacent first spectral bands to obtain first relative position data between each of the first spectral bands;

[0072] Calculating the difference in position coordinates between adjacent template spectral bands to obtain second relative position data between the respective template spectral bands;

[0073] The second similarity score is acquired according to the first relative position data and the second relative position data.

[0074] After obtaining the key spectral bands in the image to be analyzed and the key spectral bands in the thin layer chromatography image of the target sample, the difference between the position coordinates of the adjacent first spectral bands in the image to be analyzed is calculated, that is, the position coordinates of the adjacent first spectral bands are subtracted, and a vector representing the relative position relationship between each of the first spectral bands can be obtained, and the vector is used as the first relative position data. Further, the vector can be normalized and used as the first relative position data. The second relative position data can be obtained in the same way. The similarity between the first relative position data and the second relative position data is calculated to obtain the second similarity score, and the second similarity score can be calculated by the cosine similarity calculation method or by other similarity calculation methods.

[0075] Using the relative position relationship of chromatographic bands as the key identification factor can alleviate the impact of chromatographic band deformation on identification accuracy in actual sampling.

[0076] Through the above steps, the first similarity score and the second similarity score of the sample to be analyzed and one of the target samples can be calculated, and the first similarity score and the second similarity score of the sample to be analyzed and each of the target samples can be calculated using the same method. The method provided in this embodiment also includes the steps of:

[0077] S400: Input the first similarity score and the second similarity score of the sample to be analyzed and each of the target samples into a trained second neural network, and obtain a component analysis result output by the second neural network.

[0078] The first similarity score and the second similarity score of the sample to be analyzed and each of the target samples are input into the second neural network, and a component analysis result of the sample to be analyzed is obtained through the second neural network.

[0079] The parameters of the preset filter are trained together with the parameters of the first neural network. The training data includes multiple groups of sample data, each group of sample data includes a sample image to be analyzed, each of the target samples and components of samples corresponding to the sample image to be analyzed.

[0080] Specifically, the second neural network is obtained by training multiple groups of sample data, each group of sample data includes sample images to be analyzed, each target sample and the components of the sample corresponding to the sample image to be analyzed. Specifically, in the training process of the second neural network, the sample image to be analyzed is processed by the processing method of steps S100-S300, and the first similarity score and the second similarity score corresponding to each target sample are obtained, and the MLP network is input with the category label, and the predicted classification category is obtained by forward propagation, and the loss function size of the predicted value and the true value is calculated by using the softmax loss function, and back propagation is performed to update the network parameters. The parameters of the preset filter can be updated together with the parameters of the second neural network. When a certain number of iterations are performed, the network parameters and the loss function size tend to be stable, and the training is stopped. In manual identification, according to the degree of emphasis on the relative position relationship between the characteristic points of the chromatographic bands and the chromatographic bands, the identification strategies can be divided into three categories, which are biased towards the use of chromatographic band characteristic points, biased towards the use of chromatographic band relative position relationships, and equally valued for the use of chromatographic band characteristic points and chromatographic band relative position relationships for thin layer chromatography identification. During training, we will collect the training data of the similarity scores of the relative position relationship of the spectral bands and the similarity scores of the characteristic points obtained by different strategies, as well as the types of chemical components. In this way, the MLP network finally trained can forward propagate adaptive weight fusion based on the similarity vector of the input chromatographic band characteristic points and the relative position relationship of the chromatographic bands, and obtain the identification result that combines the advantages of the two identification factors.

[0081] like Figure 3 As shown, the second neural network can adopt an MLP network structure. The second neural network can fuse the feature point similarity and the spectral band position relationship similarity, which can effectively improve the accuracy of the component analysis result.

[0082] In summary, this embodiment provides a thin layer chromatography component analysis method based on adaptive weight fusion, which performs feature vector extraction and spectral band extraction on the thin layer chromatography images of the sample to be analyzed and the thin layer chromatography images of the target sample with known components, uses feature vector matching and the relative position relationship of the chromatographic bands as identification factors and inputs them into the neural network for component identification, and integrates the feature point similarity and the spectral band position relationship similarity, which can improve the accuracy and efficiency of the thin layer chromatography component analysis results.

[0083] It should be understood that, although the various steps in the flowcharts given in the accompanying drawings of the present specification are displayed in sequence according to the indications of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless there is a clear description in this article, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0084] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0085] Embodiment 2

[0086] Based on the above embodiments, the present invention also provides a thin layer chromatography component analysis device based on adaptive weight fusion, such as Figure 4 As shown, the thin layer chromatography component analysis device based on adaptive weight fusion includes:

[0087] A feature extraction module, wherein the feature extraction module is used to obtain an image to be analyzed, wherein the image to be analyzed is a thin layer chromatography image of a sample to be analyzed, and the image to be analyzed is input into a preset filter, and each first feature vector in the image to be analyzed is extracted through the preset filter, as described in the first embodiment;

[0088] a first similarity module, the first similarity module being used to match the first feature vector with each template feature vector of the target sample, and to obtain a first similarity score between the sample to be analyzed and the target sample based on the matching result, wherein the template feature vector is a feature vector extracted from the thin layer chromatography image of the target sample, as specifically described in the first embodiment;

[0089] A spectral band extraction module, the spectral band extraction module is used to input the image to be analyzed into a trained first neural network, and obtain each first spectral band in the image to be analyzed output by the first neural network, as specifically described in the first embodiment;

[0090] a second similarity module, the second similarity module being used to obtain a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the respective first spectral bands and the relative positions between the respective template spectral bands of the target sample, wherein the template spectral band of the target sample is a spectral band extracted from the thin layer chromatography image of the target sample, as specifically described in the first embodiment;

[0091] A fusion module, wherein the fusion module is used to input the first similarity score and the second similarity score of the sample to be analyzed and each of the target samples into a trained second neural network, and obtain a component analysis result output by the second neural network, as specifically described in Example 1.

[0092] Embodiment 3

[0093] Based on the above embodiments, the present invention also provides a terminal, such as Figure 5 As shown, the terminal includes a processor 10 and a memory 20. Figure 5 Only some components of the terminal are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0094] The memory 20 may be an internal storage unit of the terminal in some embodiments, such as a hard disk or memory of the terminal. The memory 20 may also be an external storage device of the terminal in other embodiments, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the terminal. Further, the memory 20 may also include both an internal storage unit of the terminal and an external storage device. The memory 20 is used to store application software and various types of data installed in the terminal. The memory 20 may also be used to temporarily store data that has been output or is to be output. In one embodiment, a thin layer chromatography component analysis program 30 based on adaptive weight fusion is stored on the memory 20, and the thin layer chromatography component analysis program 30 based on adaptive weight fusion can be executed by the processor 10, thereby realizing the thin layer chromatography component analysis method based on adaptive weight fusion in the present application.

[0095] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor or other chip, used to run the program code or process data stored in the memory 20, such as executing the target-based multimodal named entity recognition method.

[0096] In one embodiment, when the processor 10 executes the thin layer chromatography component analysis program 30 based on adaptive weight fusion in the memory 20, the following steps are implemented:

[0097] Acquire an image to be analyzed, wherein the image to be analyzed is a thin layer chromatography image of the sample to be analyzed, input the image to be analyzed into a preset filter, and extract each first feature vector in the image to be analyzed through the preset filter;

[0098] Matching the first feature vector with each template feature vector of the target sample, and obtaining a first similarity score between the sample to be analyzed and the target sample based on the matching result, wherein the template feature vector is a feature vector extracted from a thin layer chromatography image of the target sample;

[0099] Inputting the image to be analyzed into a trained first neural network, and obtaining each first spectral band in the image to be analyzed output by the first neural network;

[0100] Obtaining a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the respective first spectral bands and the relative positions between the respective template spectral bands of the target sample, wherein the template spectral band of the target sample is a spectral band extracted from the thin layer chromatography image of the target sample;

[0101] The first similarity score and the second similarity score of the sample to be analyzed and each of the target samples are input into a trained second neural network to obtain a component analysis result output by the second neural network.

[0102] Before matching the first feature vector with each template feature vector of the target sample, the method further includes:

[0103] Inputting the thin layer chromatography image of the target sample into the preset filter, and extracting each second feature vector in the thin layer chromatography image of the target sample through the preset filter;

[0104] At least one of the second feature vectors is selected according to the modulus of each of the second feature vectors as the template feature vector corresponding to the target sample.

[0105] The matching of the first feature vector with each template feature vector of the target sample includes:

[0106] Calculating the similarity between the first feature vector and each of the template feature vectors;

[0107] When the similarity between the first feature vector and the template feature vector is greater than a preset threshold, it is determined that the first feature vector matches the template feature vector.

[0108] Wherein, obtaining a first similarity score between the sample to be analyzed and the target sample based on the matching result includes:

[0109] The first similarity score is obtained according to the number of the first feature vectors that match the template feature vector of the target sample.

[0110] Wherein, the obtaining of a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the respective first spectral bands and the relative positions between the respective template spectral bands of the target sample comprises:

[0111] Calculating the difference in position coordinates between adjacent first spectral bands to obtain first relative position data between each of the first spectral bands;

[0112] Calculating the difference in position coordinates between adjacent template spectral bands to obtain second relative position data between the respective template spectral bands;

[0113] The second similarity score is acquired according to the first relative position data and the second relative position data.

[0114] Before obtaining the second similarity score between the sample to be analyzed and the target sample based on the relative positions between the first spectral bands and the relative positions between the template spectral bands, the method further includes:

[0115] The thin layer chromatography image of the target sample is input into the first trained neural network, and each of the template spectrum bands of the target sample output by the first neural network is obtained.

[0116] The parameters of the preset filter are trained together with the parameters of the first neural network, and the training data includes multiple groups of sample data, each group of sample data includes a sample image to be analyzed, each of the target samples and components of samples corresponding to the sample image to be analyzed.

[0117] Embodiment 4

[0118] The present invention also provides a computer-readable storage medium, which stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the thin layer chromatography component analysis method based on adaptive weight fusion as described above.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A thin layer chromatography component analysis method based on adaptive weight fusion, characterized in that: The method comprises: Acquire an image to be analyzed, wherein the image to be analyzed is a thin layer chromatography image of the sample to be analyzed, input the image to be analyzed into a preset filter, and extract each first feature vector in the image to be analyzed through the preset filter; Matching the first feature vector with each template feature vector of the target sample, and obtaining a first similarity score between the sample to be analyzed and the target sample based on the matching result, wherein the template feature vector is a feature vector extracted from a thin layer chromatography image of the target sample; Inputting the image to be analyzed into a trained first neural network, and obtaining each first spectral band in the image to be analyzed output by the first neural network; Obtaining a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the respective first spectral bands and the relative positions between the respective template spectral bands of the target sample, wherein the template spectral band of the target sample is a spectral band extracted from the thin layer chromatography image of the target sample; Inputting the first similarity score and the second similarity score of the sample to be analyzed and each of the target samples into a trained second neural network, and obtaining a component analysis result output by the second neural network; The second neural network adopts an MLP structure and performs adaptive weight fusion according to the forward propagation of the first similarity score and the second similarity score.

2. The thin layer chromatography component analysis method based on adaptive weight fusion according to claim 1, characterized in that: Before matching the first feature vector with each template feature vector of the target sample, the method further includes: Inputting the thin layer chromatography image of the target sample into the preset filter, and extracting each second feature vector in the thin layer chromatography image of the target sample through the preset filter; At least one of the second feature vectors is selected according to the modulus of each of the second feature vectors as the template feature vector corresponding to the target sample.

3. The thin layer chromatography component analysis method based on adaptive weight fusion according to claim 1, characterized in that: The matching of the first feature vector with each template feature vector of the target sample includes: Calculating the similarity between the first feature vector and each of the template feature vectors; When the similarity between the first feature vector and the template feature vector is greater than a preset threshold, it is determined that the first feature vector matches the template feature vector.

4. The thin layer chromatography component analysis method based on adaptive weight fusion according to claim 1, characterized in that: The obtaining a first similarity score between the sample to be analyzed and the target sample based on the matching result includes: The first similarity score is obtained according to the number of the first feature vectors that match the template feature vector of the target sample.

5. The thin layer chromatography component analysis method based on adaptive weight fusion according to claim 1, characterized in that: The step of obtaining a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the first spectral bands and the relative positions between the template spectral bands of the target sample comprises: Calculating the difference in position coordinates between adjacent first spectral bands to obtain first relative position data between each of the first spectral bands; Calculating the difference in position coordinates between adjacent template spectral bands to obtain second relative position data between the respective template spectral bands; The second similarity score is acquired according to the first relative position data and the second relative position data.

6. The thin layer chromatography component analysis method based on adaptive weight fusion according to claim 1, characterized in that: Before obtaining the second similarity score between the sample to be analyzed and the target sample based on the relative positions between the first spectral bands and the relative positions between the template spectral bands, the method further includes: The thin layer chromatography image of the target sample is input into the first trained neural network, and each of the template spectrum bands of the target sample output by the first neural network is obtained.

7. The thin layer chromatography component analysis method based on adaptive weight fusion according to any one of claims 1 to 6, characterized in that: The parameters of the preset filter are trained together with the parameters of the first neural network. The training data includes multiple groups of sample data, each group of sample data includes a sample image to be analyzed, each of the target samples and components of samples corresponding to the sample image to be analyzed.

8. A thin layer chromatography component analysis device based on adaptive weight fusion, characterized in that: include: A feature extraction module, wherein the feature extraction module is used to obtain an image to be analyzed, wherein the image to be analyzed is a thin layer chromatography image of a sample to be analyzed, input the image to be analyzed into a preset filter, and extract each first feature vector in the image to be analyzed through the preset filter; a first similarity module, the first similarity module being used to match the first feature vector with each template feature vector of the target sample, and to obtain a first similarity score between the sample to be analyzed and the target sample based on the matching result, wherein the template feature vector is a feature vector extracted from a thin layer chromatography image of the target sample; A spectral band extraction module, the spectral band extraction module is used to input the image to be analyzed into a trained first neural network, and obtain each first spectral band in the image to be analyzed output by the first neural network; a second similarity module, the second similarity module being used to obtain a second similarity score between the sample to be analyzed and the target sample based on the relative positions between the respective first spectral bands and the relative positions between the respective template spectral bands of the target sample, wherein the template spectral band of the target sample is a spectral band extracted from the thin layer chromatography image of the target sample; A fusion module, wherein the fusion module is used to input the first similarity score and the second similarity score of the sample to be analyzed and each of the target samples into a trained second neural network to obtain a component analysis result output by the second neural network; The second neural network adopts an MLP structure and performs adaptive weight fusion according to the forward propagation of the first similarity score and the second similarity score.

9. A terminal, characterized in that: The terminal includes: a processor, a computer-readable storage medium communicatively connected to the processor, the computer-readable storage medium being suitable for storing a plurality of instructions, and the processor being suitable for calling the instructions in the computer-readable storage medium to execute the steps of implementing the thin layer chromatography component analysis method based on adaptive weight fusion as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the thin layer chromatography component analysis method based on adaptive weight fusion as described in any one of claims 1-7.

Images