Microstructure image prediction method and device for small-angle X-ray scattering

The Fourier transform and loss function iterate on SAXS images through convolutional neural network prediction model solves the problem of low data processing efficiency of SAXS images, and realizes the rapid acquisition of electron density distribution information of polymer materials, improving analysis efficiency.

CN118982455APending Publication Date: 2024-11-19UNIV OF SCI & TECH OF CHINA
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411011442.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-26
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

In the prior art, the data processing and analysis efficiency of SAXS images is low, making it difficult to quickly obtain microstructure information of polymer materials, and lacks a simple data processing method.

Method used

The convolutional neural network prediction model is used to input the original SAXS image into the encoder and decoder, and iterate through Fourier transform and loss function until the loss function curve converges, so as to realize the mapping of the SAXS image to the electron density distribution in real space.

Benefits of technology

It improves the processing and analysis efficiency of SAXS images, can quickly predict electron density distribution images, simplifies the acquisition process of polymer material structure information, and is suitable for batch data analysis of various types of materials.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118982455A_ABST
    Figure CN118982455A_ABST
Patent Text Reader

Abstract

The invention provides a small-angle X-ray scattering microstructure image prediction method and device, which can be applied to the field of material microstructures and the technical field of artificial intelligence. The method comprises the steps that an original scattering image is input into a prediction model, a predicted initial microstructure image is obtained, the original scattering image represents the scattering pattern of small-angle X-rays of a target object made of a high polymer material, and the initial microstructure image represents the electron density distribution condition of the target object; performing Fourier transform on the initial microstructure image to obtain a simulated scattering image; the prediction model is iterated based on the loss function value between the original scattering image and the simulated scattering image until a loss function curve reaches convergence, and the loss function curve is constructed based on the loss function value obtained in each round of multi-round iteration; and taking an output image of the prediction model obtained by iteration of the original scattering image as a target microstructure image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the fields of material microstructure and artificial intelligence technology, and more specifically, to a method and device for predicting microstructure images of small-angle X-ray scattering. Background Art

[0002] Synchrotron radiation X-ray scattering technology has become a powerful tool for studying the structural evolution of polymer materials on multiple length and time scales. In time-resolved in-situ small-angle X-ray scattering (SAXS) experiments, a large number of SAXS images will be generated. SAXS images are processed or analyzed to obtain the real microstructural evolution laws of polymer materials.

[0003] In the process of realizing the concept of the present disclosure, the inventors found that there are at least the following problems in the related art: the data processing and analysis efficiency is low in the process of obtaining and analyzing the structural information of polymer materials from SAXS images. Summary of the invention

[0004] In view of this, the present disclosure provides a method and device for predicting microstructure images of small-angle X-ray scattering.

[0005] One aspect of the present disclosure provides a method for predicting a microstructure image of small-angle X-ray scattering, comprising: inputting an original scattering image into a prediction model to obtain a predicted initial microstructure image, wherein the original scattering image represents a scattering pattern of small-angle X-rays of a target object made of a polymer material, and the initial microstructure image represents the electron density distribution of the target object; performing Fourier transform on the initial microstructure image to obtain a simulated scattering image; iterating the prediction model based on a loss function value between the original scattering image and the simulated scattering image until a loss function curve converges, wherein the loss function curve is constructed based on the loss function values ​​obtained in each round of multiple iterations; and using an output image of the prediction model iteratively obtained using the original scattering image as the target microstructure image.

[0006] According to an embodiment of the present disclosure, inputting the original scattering image into a prediction model to obtain a predicted initial microstructure image includes: inputting the original scattering image into an encoder of the prediction model to obtain a feature image; and inputting the feature image into a decoder of the prediction model to obtain an initial microstructure image.

[0007] According to an embodiment of the present disclosure, the encoder includes multiple convolutional network layers, and the convolutional network layers include convolutional layers, activation function layers and pooling layers.

[0008] According to an embodiment of the present disclosure, the decoder includes multiple deconvolutional network layers, and the deconvolutional network layers include deconvolution layers, activation function layers and inverse pooling layers.

[0009] According to an embodiment of the present disclosure, performing Fourier transform on the initial microstructure image to obtain a simulated scattering image includes: performing a two-dimensional Fourier transform on the initial microstructure image to obtain a Fourier image, wherein the Fourier image represents frequency domain information of the initial microstructure image;

[0010] According to the mapping relationship between the frequency domain information and the scattering intensity information of small-angle X-rays, the Fourier transform image is pixel-converted to obtain a simulated scattering image.

[0011] According to an embodiment of the present disclosure, the loss function of the prediction model is:

[0012]

[0013] NPCC represents the negative of the Pearson correlation coefficient between the original scattering image A and the simulated scattering image B, Ai represents the pixel value of the pixel point in the original scattering image A, represents the pixel average value of the original scattering image A, Bi represents the pixel value of the pixel point in the simulated scattering image B, represents the pixel average value of the simulated scattering image B.

[0014] According to an embodiment of the present disclosure, the parameters of the prediction model are updated by a gradient-based optimizer.

[0015] According to an embodiment of the present disclosure, the method further includes: graying the small-angle X-ray scattering image to obtain a grayscale scattering image; normalizing the grayscale scattering image to obtain a normalized scattering image; and adjusting the normalized scattering image based on a preset image size to obtain an original scattering image.

[0016] According to an embodiment of the present disclosure, the prediction model further includes a normalization layer, and the input image of the normalization layer is the output image of the activation function layer.

[0017] Another aspect of the present disclosure provides a microstructure image prediction device for small-angle X-ray scattering, including: an input module, used to input an original scattering image into a prediction model to obtain a predicted initial microstructure image, wherein the original scattering image represents the scattering pattern of small-angle X-rays of a target object made of a polymer material, and the initial microstructure image represents the electron density distribution of the target object; a Fourier transform module, used to perform Fourier transform on the initial microstructure image to obtain a simulated scattering image; an iteration module, used to iterate the prediction model based on the loss function value between the original scattering image and the simulated scattering image until the loss function curve converges, wherein the loss function curve is constructed based on the loss function values ​​obtained in each round of multiple iterations; and a determination module, used to use the output image of the prediction model obtained by iterating the original scattering image as the target microstructure image.

[0018] According to an embodiment of the present disclosure, by inputting the original scattering image into the prediction model, a predicted initial microstructure image is obtained, the original scattering image represents the scattering pattern of small-angle X-rays of the target object made of polymer materials, and the initial microstructure image represents the electron density distribution of the target object. The initial microstructure image is Fourier transformed to obtain a simulated scattering image, thereby obtaining information about the brightness in the initial microstructure image. The prediction model is iterated based on the loss function value between the original scattering image and the simulated scattering image until the loss function curve converges; thereby the output image of the prediction model obtained by iterating the original scattering image is used as the target microstructure image, and a complex mapping from the reciprocal space to the real space of the SAXS image is realized, that is, the SAXS image directly predicts the electron density distribution of the target object, which improves the processing and analysis efficiency of the SAXS image. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0020] Figure 1 An exemplary system architecture of a microstructure image prediction method using small-angle X-ray scattering according to an embodiment of the present disclosure is schematically shown;

[0021] Figure 2 A flowchart of a method for predicting a microstructure image of small-angle X-ray scattering according to an embodiment of the present disclosure is schematically shown;

[0022] Figure 3 Schematically shows an original scattering image according to an embodiment of the present disclosure;

[0023] Figure 4 The following schematically shows an iterative process diagram of a prediction model according to an embodiment of the present disclosure;

[0024] Figure 5A Schematically shows a schematic diagram of the convergence of the loss function curve according to an embodiment of the present disclosure;

[0025] Figure 5B Schematically shows an original scattering image, a characteristic image, a Fourier image and a target microstructure image according to an embodiment of the present disclosure;

[0026] Figure 6 Schematically showing the original scattering images of different stretching temperature regions according to an embodiment of the present disclosure;

[0027] Figure 7 Schematically showing target microstructure images of different stretching temperature regions according to an embodiment of the present disclosure;

[0028] Figure 8 A block diagram schematically shows a microstructure image prediction device of small-angle X-ray scattering according to an embodiment of the present disclosure; and

[0029] Fig. 9 A block diagram of an electronic device suitable for implementing a microstructure image prediction method of small-angle X-ray scattering according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0030] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0031] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0032] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0033] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0034] In the embodiments of the present disclosure, the collection, updating, analysis, processing, use, transmission, provision, disclosure, storage, etc. of the data involved (for example, including but not limited to user personal information) are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information and network security.

[0035] In the embodiments of the present disclosure, the user's authorization or consent is obtained before obtaining or collecting the user's personal information.

[0036] An embodiment of the present disclosure provides a method for predicting a microstructure image of small-angle X-ray scattering, comprising: inputting an original scattering image into a prediction model to obtain a predicted initial microstructure image, wherein the original scattering image represents the scattering pattern of small-angle X-rays of a target object made of a polymer material, and the initial microstructure image represents the electron density distribution of the target object; performing Fourier transform on the initial microstructure image to obtain a simulated scattering image; iterating the prediction model based on a loss function value between the original scattering image and the simulated scattering image until a loss function curve converges, wherein the loss function curve is constructed based on the loss function values ​​obtained in each round of multiple iterations; and using the output image of the prediction model obtained by iterating the original scattering image as the target microstructure image.

[0037] Figure 1 The exemplary system architecture of the microstructure image prediction method using small angle X-ray scattering according to the embodiment of the present disclosure is schematically shown. It should be noted that: Figure 1 What is shown is merely an example of a system architecture to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0038] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0039] The user may use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software, etc. (only for example).

[0040] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0041] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0042] It should be noted that the microstructure image prediction method of small-angle X-ray scattering provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the microstructure image prediction device of small-angle X-ray scattering provided in the embodiment of the present disclosure can generally be set in the server 105. The microstructure image prediction method of small-angle X-ray scattering provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the microstructure image prediction device of small-angle X-ray scattering provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Alternatively, the method for predicting a microstructure image of small-angle X-ray scattering provided in the embodiment of the present disclosure may also be performed by the first terminal device 101, the second terminal device 102, or the third terminal device 103, or may also be performed by other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103. Accordingly, the microstructure image prediction device of small-angle X-ray scattering provided in the embodiment of the present disclosure may also be provided in the first terminal device 101, the second terminal device 102, or the third terminal device 103, or may be provided in other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103.

[0043] For example, the initial microstructure image may be originally stored in any one of the first terminal device 101, the second terminal device 102, or the third terminal device 103 (for example, the first terminal device 101, but not limited thereto), or stored in an external storage device and may be imported into the first terminal device 101. Then, the first terminal device 101 may locally execute the microstructure image prediction method for small-angle X-ray scattering provided in the embodiment of the present disclosure, or send the initial microstructure image to other terminal devices, servers, or server clusters, and the other terminal devices, servers, or server clusters that receive the image to be processed may execute the microstructure image prediction method for small-angle X-ray scattering provided in the embodiment of the present disclosure.

[0044] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0045] Small-angle X-ray scattering (SAXS) technology has become an indispensable means of analyzing the microstructure of polymer materials due to its unique statistical averaging and non-destructive detection of microstructures. However, data processing and analysis of SAXS images have always been a difficult problem. Usually, the X-ray scattering intensity we measure represents the modulus of the complex Fourier transform (FT) of the object, but the phase of the wave will be lost. Therefore, the image of the object cannot be recovered from a simple inverse Fourier transform. The current development trend of data analysis is mainly focused on how to select a simplified model and derive an autocorrelation function with a characteristic shape to indirectly describe the electron cloud density distribution of the system, so as to calculate some specific structural parameters based on the scattering intensity distribution. In view of this, if the complex scattering principle is avoided and the electron density distribution image is directly predicted from the SAXS image, the data processing and analysis efficiency in the process of obtaining and analyzing the structural information of polymer materials from the SAXS image can be greatly improved. Complex mapping from reciprocal space to real space can be achieved for the SAXS scattering signal of any type of material, which can be used to quickly reveal the real microstructural evolution law of samples in large SAXS data sets. The development of data processing software determines whether small-angle X-ray scattering technology can achieve a big leap in application. Although some data batch processing software has appeared at this stage, it is still in the niche stage and is not easy to operate. This data processing method has great potential to be applied to more intelligent one-click data analysis software in the future.

[0046] The general idea is that a set of mature SAXS images and a data set from each SAXS data to the real-space microstructure of the material are needed, which is very difficult to achieve. Since we want to achieve a processing method that can be used in two-dimensional SAXS image processing, the requirements for the data set are very high. It is necessary to cover SAXS data of various types of materials and the corresponding real-space structure images. Such a data set does not exist at present. Taking a step back, even using simulation methods to build a data set from scratch is difficult to achieve, because there is no accurate means to know the microstructure of a certain sample. When using SAXS to analyze the microstructure of materials, scanning electron microscopes (Scanning Electron Microscope, SEM), transmission electron microscopes (Transmission Electron Microscope, TEM), differential scanning calorimeters (Differential Scanning Calorimeter, DSC) and other experiments are often required to assist in verifying the analysis results.

[0047] The disclosed embodiment uses an unsupervised learning method to establish a self-iterative microstructure image prediction method, skipping the step of labeling the data, and realizing rapid iterative search for real-space electron distribution images that may correspond to small-angle X-ray scattering images, and realizing rapid batch prediction of electron density distribution images corresponding to several scattering images, thereby serving as an auxiliary tool for analyzing the massive SAXS data generated in synchrotron radiation in-situ experiments.

[0048] Figure 2 The flowchart of the microstructure image prediction method of small-angle X-ray scattering according to an embodiment of the present disclosure is schematically shown.

[0049] like Figure 2 As shown, the method includes operations S210 to S240.

[0050] In operation S210 , the original scattering image is input into a prediction model to obtain a predicted initial microstructure image.

[0051] According to an embodiment of the present disclosure, the original scattering image represents the scattering pattern of small-angle X-rays of a target object made of a polymer material.

[0052] According to an embodiment of the present disclosure, the polymer material may be polyethylene, polypropylene, polyvinyl chloride, polytetrafluoroethylene, etc.

[0053] According to an embodiment of the present disclosure, the target object may be a plastic made of polyethylene, a pipe made of polyvinyl chloride, a non-stick pan made of polytetrafluoroethylene, and the like.

[0054] Figure 3The figure schematically shows an original scattering image according to an embodiment of the present disclosure.

[0055] like Figure 3 As shown, the polymer material may be polypropylene, and the target object may be a hard elastic film. The original scattering image represents the scattering pattern of small-angle X-rays of the hard elastic film made of polypropylene.

[0056] According to an embodiment of the present disclosure, the prediction model may be a neural network model, such as a deep learning model, a convolutional neural network model, etc.

[0057] According to an embodiment of the present disclosure, the initial microstructure image represents the electron density distribution of the target object.

[0058] In operation S220, Fourier transform is performed on the initial microstructure image to obtain a simulated scattering image.

[0059] According to an embodiment of the present disclosure, since the brightness in the original scattering image is related to the scattering intensity of small-angle X-rays, the initial microstructure image output by the prediction model can be Fourier transformed to obtain information about the brightness in the initial microstructure image.

[0060] In operation S230 , the prediction model is iterated based on the loss function value between the original scatter image and the simulated scatter image until the loss function curve reaches convergence.

[0061] According to an embodiment of the present disclosure, the loss function curve is constructed based on the loss function values ​​obtained in each round of multiple iterations. For example, in the coordinate system of the loss function curve, the horizontal axis can be the iteration round, and the vertical axis is the loss function value corresponding to the iteration round.

[0062] According to an embodiment of the present disclosure, the loss function curve reaches convergence when, after reaching a certain number of rounds, the change in the loss function value is small, and no significant fluctuation or oscillation occurs.

[0063] According to an embodiment of the present disclosure, the loss function for calculating the loss function value between the original scattering image and the simulated scattering image may be a mean square error function, a cross entropy loss function, a logarithmic loss function, or the like.

[0064] In operation S240, an output image of the prediction model obtained by iteratively using the original scattering image is used as a target microstructure image.

[0065] According to the embodiments of the present disclosure, since the above-mentioned iterative process does not take a long time, the output image of the prediction model obtained by iteration is directly used as the target microstructure image, which can improve the output efficiency of the prediction model.

[0066] According to the embodiments of the present disclosure, the original scattering image is input into the prediction model to obtain the predicted initial microstructure image; the initial microstructure image is Fourier transformed to obtain the simulated scattering image, thereby obtaining information about the brightness in the initial microstructure image. The prediction model is iterated based on the loss function value between the original scattering image and the simulated scattering image until the loss function curve converges; thereby the output image of the prediction model obtained by iterating the original scattering image is used as the target microstructure image, and the complex mapping of the SAXS image from the reciprocal space to the real space is realized, that is, the SAXS image directly predicts the electron density distribution of the target object, thereby improving the processing and analysis efficiency of the SAXS image.

[0067] According to an embodiment of the present disclosure, inputting the original scattering image into a prediction model to obtain a predicted initial microstructure image includes: inputting the original scattering image into an encoder of the prediction model to obtain a feature image; and inputting the feature image into a decoder of the prediction model to obtain an initial microstructure image.

[0068] According to an embodiment of the present disclosure, the encoder includes multiple convolutional network layers, and the convolutional network layers include convolutional layers, activation function layers and pooling layers.

[0069] According to an embodiment of the present disclosure, the convolutional network layer can realize feature extraction of SAXS information in the original scattering image. The convolutional network layer can be stacked in the order of a convolutional layer, an activation function layer, and a pooling layer. The original scattering image can be encoded into a size of 2×2 through three cyclic stacked convolutional network layers.

[0070] According to an embodiment of the present disclosure, the decoder includes multiple deconvolutional network layers, and the deconvolutional network layers include deconvolution layers, activation function layers and inverse pooling layers.

[0071] According to an embodiment of the present disclosure, the prediction model further includes a normalization layer, and the input image of the normalization layer is the output image of the activation function layer.

[0072] According to the embodiments of the present disclosure, since different parameter settings will significantly affect the characteristics and performance of the generated image, the number of layers and related parameter settings of the decoder do not need to be completely consistent with the encoder. It is better to set four layers of deconvolution network layers that are the same as the encoder, and the number of convolution kernels is 16, 32, 64, 64 respectively; the step sizes are 2, 2, 2, 1 respectively. An activation function layer and an inverse pooling layer are set between each layer. A normalization layer (BN layer) needs to be added after each activation function layer of the entire network, which can speed up the convergence speed during model training, make the model training process more stable, avoid gradient explosion or gradient disappearance, and play a certain regularization role.

[0073] According to an embodiment of the present disclosure, Fourier transform is performed on the initial microstructure image to obtain a simulated scattering image, including: performing a two-dimensional Fourier transform on the initial microstructure image to obtain a Fourier image, wherein the Fourier image represents frequency domain information of the initial microstructure image; and performing pixel conversion on the Fourier transform image according to a mapping relationship between the frequency domain information and the scattering intensity information of small-angle X-rays to obtain a simulated scattering image.

[0074] According to the embodiments of the present disclosure, the two-dimensional Fourier transform can convert the initial microstructure image into the frequency domain. Since there is a mapping relationship or correlation relationship between the frequency intensity and the scattering intensity of the small-angle X-ray, the Fourier transform image is pixel-converted to achieve mapping the pixels of the Fourier image to the pixel range of the original scattering image, which is convenient for the subsequent calculation of the Pearson correlation coefficient between the original scattering image and the simulated scattering image, and then the loss function value is obtained, thereby improving the efficiency of data processing.

[0075] According to an embodiment of the present disclosure, the loss function of the prediction model is:

[0076] (1)

[0077] NPCC represents the negative of the Pearson correlation coefficient between the original scattering image A and the simulated scattering image B, Ai represents the pixel value of the pixel point in the original scattering image A, represents the pixel average value of the original scattering image A, Bi represents the pixel value of the pixel point in the simulated scattering image B, represents the pixel average value of the simulated scattering image B.

[0078] According to an embodiment of the present disclosure, the parameters of the prediction model are updated by a gradient-based optimizer.

[0079] According to an embodiment of the present disclosure, the gradient-based optimizer may be an Adam optimizer (Adaptive Moment Estimation). The Adam optimizer enables a faster convergence speed of the loss function curve, is more robust to default parameter settings, and has the ability to adaptively adjust the learning rate. The Adam optimizer may also tune hyperparameters.

[0080] According to an embodiment of the present disclosure, the method further includes: graying the small-angle X-ray scattering image to obtain a grayscale scattering image; normalizing the grayscale scattering image to obtain a normalized scattering image; and adjusting the normalized scattering image based on a preset image size to obtain an original scattering image.

[0081] According to an embodiment of the present disclosure, the magnitude of the grayscale value in the grayscale scattering image may describe the magnitude of the small-angle X-ray scattering intensity.

[0082] According to an embodiment of the present disclosure, the grayscale scattering image is normalized or standardized to achieve that each pixel value of the grayscale scattering image is converted from 0 to 255 to 0 to 1.

[0083] According to an embodiment of the present disclosure, the size of the normalized scattering image needs to be adjusted to 128×128 in the preprocessing stage. For example, the preset image size may be 128×128.

[0084] According to the embodiments of the present disclosure, the resolution of the input SAXS image is one of the important parameters. Different resolutions will affect the subsequent network structure, convergence speed, learning rate and other parameters. 128×128 can be used as the resolution standard of the input image.

[0085] For example, the SAXS scattering signal originates from the mean square fluctuation of the electron density at the nanoscale inside the sample, which itself cannot provide intuitive information about the microstructure of the film. According to the basic principle of scattering, if the system is composed of N identical particles, its scattering intensity I(Q) is:

[0086] (2)

[0087] Q is the modulus of the scattering vector, F(Q) is the Thomson scattering factor of the particle, and S(Q) is the structure factor introduced to consider the interaction between particles. If there is a correlation between the position and orientation of the particle (this is the case in oriented polymer materials, such as polypropylene hard elastic film), S(Q) will become very complicated. In the process of restoring the scattering signal in the reciprocal space to the microstructure of the material in the real space, a large number of mathematical operations and the fitting of the corresponding theoretical model are involved. If you are not careful, it is very likely to get wrong results. Therefore, the processing and analysis of SAXS data has always been a difficult problem.

[0088] In order to solve the above technical problems, the encoder-decoder containing the convolutional neural network can extract the structural information implicit in the SAXS pattern and reversely expand it into a real-space electron density distribution image. The last layer of the decoder is Fourier transformed and the loss function is calculated with the original SAXS image, thereby achieving self-iterative convergence of the prediction model. This algorithm converges very quickly, taking only a few minutes for each image, and can automatically realize batch conversion of the entire set of in-situ SAXS data. A verification experiment was carried out using a set of SAXS data of IPP (hard elastic polypropylene) membranes, and it was found that the real-space image predicted by the algorithm is very consistent with the structural evolution characteristics of this sample in the continuous temperature-strain space.

[0089] First, in the data preprocessing stage, the original scattering image (original SAXS image) input to the prediction model needs to be converted into a grayscale image, and the grayscale value is used to describe the scattering intensity. A simple data set is established to represent a certain original SAXS image and then used as an input item in the algorithm for iterative convergence. Before entering the prediction model, it needs to be normalized or standardized once, and each pixel value in the matrix used to describe the grayscale value is converted from 0~255 to 0~1. The resolution of the original SAXS image input is one of the important parameters. Different resolutions will affect the subsequent network structure, convergence speed, learning rate and other parameters. In the example, 128×128 will be used as the resolution standard of the original SAXS image. In the preprocessing stage, the original SAXS image needs to be resized to 128×128.

[0090] Figure 4 A schematic diagram of the iterative process of the prediction model according to an embodiment of the present disclosure is schematically shown.

[0091] Build the encoder and decoder of the prediction model. First, import all relevant environment packages, including pandas, cv2, numpy, keras, etc. Figure 4 As shown in the figure, the first part of the prediction model is the encoder, which aims to extract features from the original image through a convolutional neural network and finally encode it into a 2×2 hidden feature layer. The size of the hidden layer is also one of the important parameters that affect the stability of the image reconstruction result. After testing, the reconstruction effect is best when the size is reduced to 2×2. The convolutional network layer is stacked in the order of a convolution layer, an activation function layer, a pooling layer, and a normalization layer. After three cycles, the original image is finally encoded into a size of 2×2.

[0092] The reverse decoder is constructed. The number of layers and related parameter settings of the decoder do not need to be completely consistent with the encoder, and different parameter settings will significantly affect the characteristics and performance of the generated image. After testing, it is best to set a total of four deconvolution layers that are the same as the encoder. The number of convolution kernels is 16, 32, 64, 64, and the step size is 2, 2, 2, 1. An activation function layer and an inverse pooling layer are set between each layer. A normalization layer needs to be added after each activation function layer of the entire network, which can speed up the convergence speed of the model training, make the model training process more stable, avoid gradient explosion or gradient disappearance, and play a certain regularization role.

[0093] Customizing the loss function is the key to implementing unsupervised learning algorithms. Figure 4As shown, a custom layer is set after the last layer of the decoder, that is, the initial microstructure image is subjected to a two-dimensional Fourier transform (FFT2) to obtain a simulated scattering image. Then the loss function value between the original scattering image and the simulated scattering image is calculated. The loss function uses (1+NPCC) to refine the parameter. The closer the loss function value is to 0, the closer the Pearson correlation coefficient is to 1, that is, the more similar the simulated scattering pattern after the two-dimensional Fourier transform of the initial microstructure image is to the original scattering pattern. Based on this, when the loss function curve tends to stabilize and no longer decreases with the increase of the iteration cycle, it is considered that the loss function curve has reached convergence. At this time, the output of the last layer of the decoder is the real space image corresponding to this original scattering image, that is, the target microstructure image.

[0094] The setting of parameters needs to be considered in combination with the target resolution and the desired output efficiency. Taking the output resolution of 128×128 as an example, the optimizer selects the Adam optimizer, and the learning rate mainly affects the convergence speed and the smoothness of the convergence loss function curve. The larger the learning rate, the faster the convergence speed and the greater the fluctuation of the loss function curve. Therefore, the learning rate needs to consider these two factors comprehensively and take a value that converges as fast and stable as possible. After testing, the comprehensive performance is best when the learning rate is 0.00005. It converges after about 2000 rounds, and each round takes about 0.1s. Therefore, it takes 3-4 minutes to complete the processing of each original scattering image. Since only one original scattering image is input for iterative calculation each time, the requirements for hardware performance are not high, and it can be completed using the CPU (Central Processing Unit) without using high-performance graphics processors such as GPU (Graphics Processing Unit). At the same time, it also means that if better hardware is used, the calculation speed can be further improved.

[0095] Figure 5A The figure schematically shows the convergence diagram of the loss function curve according to the embodiment of the present disclosure.

[0096] like Figure 5A As shown in the figure, the horizontal axis represents the iteration rounds (Epochs), and the vertical axis represents the loss function value (Loss). It can be seen from the figure that as the loss function value decreases and converges at any time, the output image texture of the prediction model is clearer and the image information it contains is richer.

[0097] Figure 5B An original scattering image, a characteristic image, a Fourier image and a target microstructure image according to an embodiment of the present disclosure are schematically shown.

[0098] like Figure 5BAs shown, the upper left image is the original scattering image, the upper right image is the feature image, the lower left image is the Fourier image, and the lower right image is the target microstructure image.

[0099] During runtime, batch processing of SAXS images is generally required. We need to let the algorithm read each SAXS data to be processed in turn and then save the output results. When the model is trained, it is first necessary to output the training error curve to check whether the algorithm has converged. If it has not converged, the number of rounds must be increased. Then, the original image of each SAXS data and the reconstructed electron density distribution map of the last layer of the decoder output by the model and the simulated SAXS pattern of the custom two-dimensional Fourier transform layer must be output. The purpose of outputting this item is to facilitate the inspection of the rationality of the reconstructed image, and the high-dimensional feature map of the hidden layer can also be output as a reference.

[0100] The batch processing process can be completed through a simple for loop. Various parameters of the iteration are set in advance, and an iteration number and a SAXS image with the same iteration number are specified. The prediction model starts running in the nesting. The prediction model is reconstructed each time a SAXS image is read. In each loop, the relevant iteration data and images are saved in the specified directory.

[0101] The present invention will be described in detail below by taking an example of synchrotron radiation small-angle x-ray scattering data processing for a common thin film material.

[0102] In this example, the extrusion grade iPP (F401) used for the samples. The weight average (Mw) and number average (Mn) molecular weights were about 537 kg / mol and 162 kg / mol, respectively. The iPP film was made by a homemade extrusion casting machine with a width of 160 mm, a slot die thickness of 1 mm, and extrusion temperatures set at 180°C, 195°C, 220°C, and 235°C. The die temperature was 220°C. The tempering temperatures of the casting and cooling rolls were maintained at 110 and 60°C, respectively. In order to cool the melt and make it fit closely with the casting roll, an air knife was used outside the die. After that, the extruded cast film was placed in a hot oven with a fan for annealing. The annealing lasted for one hour at 140°C. Through these processing steps, we obtained a hard and elastic iPP precursor film for the preparation of microporous membranes.

[0103] The above samples were used for in situ experiments in this study. In situ experiments require a variable temperature film stretching device that can be used with an X-ray light source. Uniaxial tensile deformation was performed using a homemade uniaxial stretching device. The length and thickness of the original sample were 20 mm and 0.015 mm, respectively. In order to obtain a high signal-to-noise ratio scattering signal, eight layers of hard elastic iPP precursor film were stacked. The time resolution of SAXS signal acquisition was set to 1 s. The total stretching rate of the sample was about 0.2 mm / s, and the stretching temperature ranged from 30 to 160 °C with a step size of 10 °C. Both in situ and ex situ experiments were performed on the beamline of a synchrotron radiation facility. In the in situ experiment, a total of 5046 SAXS patterns were collected by the detector, which had 487619 pixels and a pixel size of 172 μm. The sample-to-detector distance for SAXS measurement was 4030 mm.

[0104] Figure 6 The original scattering images of different stretching temperature regions according to an embodiment of the present disclosure are schematically shown.

[0105] like Figure 6 As shown, the first stretching temperature region (Region I) is 40 degrees Celsius to 60 degrees Celsius. The second stretching temperature region (Region II) is 90 degrees Celsius to 120 degrees Celsius. The third stretching temperature region (Region III) is 150 degrees Celsius. The scattering intensity can be expressed according to the three primary color light values ​​(Red, Green, Blue, RGB). The closer to red, the higher the intensity. After grayscale value conversion, the brightness of the grayscale scattering image is the scattering intensity. The scattering intensity of the original scattering image under different stretching temperature regions is different.

[0106] Based on the above set of in-situ SAXS images, some SAXS scattering data at some main characteristic temperatures were selected, and the microstructure image prediction method of small-angle X-ray scattering disclosed in the present invention was applied to perform image analysis. We took in-situ stretching data at 30, 90, 130, and 160°C.

[0107] Figure 7 Target microstructure images of different stretching temperature regions according to an embodiment of the present disclosure are schematically shown.

[0108] At each temperature, an image is taken at a certain distance, for a total of 34 images, such as Figure 7 As shown, after encoding, batch calculations are performed according to the microstructure image prediction method of small-angle X-ray scattering disclosed in the present invention to output target microstructure images in different stretching temperature regions.

[0109] The SAXS images are reconstructed in sequence by the microstructure image prediction method of small-angle X-ray scattering proposed in the present invention, and an intuitive real-space electron density distribution image can be obtained, which can quickly and intuitively obtain the microstructure evolution law of the sample when it is stretched to different stretching ratios at different temperatures. The displayed structural evolution image can describe the structural characteristics of each stage.

[0110] like Figure 7 As shown in the figure, whether it is the microphase separation phenomenon that occurs when the stretching is relatively low at low temperature, or the fiber crystal signal that appears after the stretching is relatively large at 130°C, corresponding similar structural images can be found.

[0111] Figure 8 The block diagram of the microstructure image prediction device of small-angle X-ray scattering according to an embodiment of the present disclosure is schematically shown.

[0112] like Figure 8 As shown, the microstructure image prediction device 800 of small-angle X-ray scattering includes an input module 810 , a Fourier transform module 820 , an iteration module 830 and a determination module 840 .

[0113] The input module 810 is used to input the original scattering image into the prediction model to obtain a predicted initial microstructure image.

[0114] The Fourier transform module 820 is used to perform Fourier transform on the initial microstructure image to obtain a simulated scattering image.

[0115] The iteration module 830 is used to iterate the prediction model based on the loss function value between the original scattering image and the simulated scattering image until the loss function curve converges, and the loss function curve is constructed based on the loss function values ​​obtained in each round of multiple iterations.

[0116] The determination module 840 is used to use the output image of the prediction model obtained by iterating the original scattering image as the target microstructure image.

[0117] According to an embodiment of the present disclosure, the input module 810 includes a first input submodule and a second input submodule. The first input submodule is used to input the original scattering image into the encoder of the prediction model to obtain a feature image. The second input submodule is used to input the feature image into the decoder of the prediction model to obtain an initial microstructure image.

[0118] According to an embodiment of the present disclosure, the encoder includes multiple convolutional network layers, and the convolutional network layers include convolutional layers, activation function layers and pooling layers.

[0119] According to an embodiment of the present disclosure, the decoder includes multiple deconvolutional network layers, and the deconvolutional network layers include deconvolution layers, activation function layers and pooling layers.

[0120] According to an embodiment of the present disclosure, the Fourier transform module 820 includes a two-dimensional Fourier transform 830 and a pixel conversion module 840. The two-dimensional Fourier transform 830 is used to perform a two-dimensional Fourier transform on the initial microstructure image to obtain a Fourier image, and the Fourier image represents the frequency domain information of the initial microstructure image. The pixel conversion module 840 is used to perform pixel conversion on the Fourier transform image according to the mapping relationship between the frequency domain information and the scattering intensity information of the small-angle X-ray to obtain a simulated scattering image.

[0121] According to an embodiment of the present disclosure, the parameters of the prediction model are updated by a gradient-based optimizer.

[0122] According to an embodiment of the present disclosure, the above also includes a grayscale module, a normalization module and an adjustment module. The grayscale module is used to grayscale the small-angle X-ray scattering image to obtain a grayscale scattering image. The normalization module is used to normalize the grayscale scattering image to obtain a normalized scattering image. The adjustment module is used to adjust the normalized scattering image based on a preset image size to obtain an original scattering image.

[0123] According to an embodiment of the present disclosure, the prediction model further includes a normalization layer, and the input image of the normalization layer is the output image of the activation function layer.

[0124] According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits, or at least part of the functions of any one of them can be implemented in one module. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be split into multiple modules for implementation. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be at least partially implemented as hardware circuits, such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, application specific integrated circuits (ASICs), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging circuits, or can be implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, according to the embodiments of the present invention, one or more of the modules, submodules, units, and subunits can be at least partially implemented as computer program modules, and when the computer program modules are run, the corresponding functions can be performed.

[0125] For example, any multiple of the input module 810, the Fourier transform module 820, the iteration module 830, and the determination module 840 can be combined in one module / unit / sub-unit for implementation, or any one of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of these modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit. According to an embodiment of the present disclosure, at least one of the input module 810, the Fourier transform module 820, the iteration module 830, and the determination module 840 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware or in a suitable combination of any of them. Alternatively, at least one of the input module 810, the Fourier transform module 820, the iteration module 830 and the determination module 840 may be at least partially implemented as a computer program module, and when the computer program module is executed, the corresponding function may be performed.

[0126] It should be noted that the part of the microstructure image prediction device of small-angle X-ray scattering in the embodiment of the present disclosure corresponds to the part of the microstructure image prediction method of small-angle X-ray scattering in the embodiment of the present disclosure. The description of the part of the microstructure image prediction device of small-angle X-ray scattering specifically refers to the part of the microstructure image prediction method of small-angle X-ray scattering, which will not be repeated here.

[0127] Fig. 9 A block diagram of an electronic device suitable for implementing a microstructure image prediction method of small-angle X-ray scattering according to an embodiment of the present disclosure is schematically shown. Fig. 9 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0128] like Fig. 9As shown, the electronic device 900 according to an embodiment of the present disclosure includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage part 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include an onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0129] In RAM 903, various programs and data required for the operation of electronic device 900 are stored. Processor 901, ROM 902 and RAM 903 are connected to each other via bus 904. Processor 901 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 902 and / or RAM 903. It should be noted that the program can also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0130] According to an embodiment of the present disclosure, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to the bus 904. The electronic device 900 may further include one or more of the following components connected to the input / output (I / O) interface 905: an input portion 906 including a keyboard, a mouse, etc.; an output portion 907 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 908 including a hard disk, etc.; and a communication portion 909 including a network interface card such as a LAN card, a modem, etc. The communication portion 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 910 as needed, so that a computer program read therefrom is installed into the storage portion 908 as needed.

[0131] According to an embodiment of the present disclosure, the method flow according to an embodiment of the present disclosure can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, the above-mentioned functions defined in the system of the embodiment of the present disclosure are executed. According to an embodiment of the present disclosure, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0132] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0133] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium. For example, it may include, but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0134] For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the ROM 902 and / or the RAM 903 described above and / or one or more memories other than the ROM 902 and the RAM 903 .

[0135] An embodiment of the present disclosure also includes a computer program product, which includes a computer program, which contains program code for executing the method provided by the embodiment of the present disclosure. When the computer program product is run on an electronic device, the program code is used to enable the electronic device to implement the microstructure image prediction method of small-angle X-ray scattering provided by the embodiment of the present disclosure.

[0136] When the computer program is executed by the processor 901, the above functions defined in the system / device of the embodiment of the present disclosure are executed. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0137] In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and downloaded and installed through the communication part 909, and / or installed from a removable medium 911. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0138] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0139] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It can be understood by those skilled in the art that the features recorded in the various embodiments of the present disclosure can be combined and / or combined in a variety of ways, even if such a combination or combination is not explicitly recorded in the present disclosure. In particular, without departing from the spirit and teaching of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0140] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A method for predicting microstructure images of small-angle X-ray scattering, comprising: Inputting the original scattering image into the prediction model to obtain a predicted initial microstructure image, wherein the original scattering image represents the scattering pattern of small-angle X-rays of the target object made of polymer material, and the initial microstructure image represents the electron density distribution of the target object; Performing Fourier transformation on the initial microstructure image to obtain a simulated scattering image; Iterating the prediction model based on a loss function value between the original scattering image and the simulated scattering image until a loss function curve converges, wherein the loss function curve is constructed based on the loss function values ​​obtained in each round of multiple rounds of iterations; The output image of the prediction model obtained by iterating the original scattering image is used as the target microstructure image.

2. The method according to claim 1, characterized in that The step of inputting the original scattering image into the prediction model to obtain the predicted initial microstructure image comprises: Inputting the original scattering image into an encoder of the prediction model to obtain a feature image; The feature image is input into the decoder of the prediction model to obtain the initial microstructure image.

3. The method according to claim 2, characterized in that The encoder includes multiple convolutional network layers, and the convolutional network layers include convolutional layers, activation function layers and pooling layers.

4. The method according to any one of claims 2 to 3, characterized in that The decoder includes multiple deconvolutional network layers, and the deconvolutional network layer includes a deconvolution layer, the activation function layer and the inverse pooling layer.

5. The method according to claim 1, characterized in that The performing Fourier transform on the initial microstructure image to obtain a simulated scattering image comprises: Performing a two-dimensional Fourier transform on the initial microstructure image to obtain a Fourier image, wherein the Fourier image represents frequency domain information of the initial microstructure image; According to the mapping relationship between the frequency domain information and the scattering intensity information of the small-angle X-ray, the Fourier transform image is pixel-converted to obtain the simulated scattering image.

6. The method according to claim 1, characterized in that The loss function of the prediction model is: NPCC represents the negative of the Pearson correlation coefficient between the original scattering image A and the simulated scattering image B, A i represents the pixel value of the pixel point in the original scattering image A, represents the pixel average value of the original scattering image A, B i represents the pixel value of the pixel point in the simulated scattering image B, represents the pixel average value of the simulated scattering image B.

7. The method according to claim 1, characterized in that The parameters of the prediction model are updated by a gradient-based optimizer.

8. The method according to claim 1, characterized in that The method further comprises: Gray-scale the small-angle X-ray scattering image to obtain a gray-scale scattering image; Normalizing the grayscale scattering image to obtain a normalized scattering image; The normalized scattering image is adjusted based on a preset image size to obtain the original scattering image.

9. The method according to claim 4, characterized in that The prediction model also includes a normalization layer, and the input image of the normalization layer is the output image of the activation function layer.

10. A microstructure image prediction device for small-angle X-ray scattering, comprising: An input module, used to input an original scattering image into a prediction model to obtain a predicted initial microstructure image, wherein the original scattering image represents a scattering pattern of small-angle X-rays of a target object made of a polymer material, and the initial microstructure image represents an electron density distribution of the target object; A Fourier transform module, used for performing Fourier transform on the initial microstructure image to obtain a simulated scattering image; an iteration module, configured to iterate the prediction model based on a loss function value between the original scattering image and the simulated scattering image until a loss function curve converges, wherein the loss function curve is constructed based on the loss function values ​​obtained in each round of multiple iterations; A determination module is used to use the output image of the prediction model iteratively obtained by using the original scattering image as the target microstructure image.

Citation Information

Cited By

  • X-ray small-angle scattering pattern reconstruction method based on physical information neural network

    CN121788724A

  • A method for reconstructing small-angle X-ray scattering patterns based on physical information neural networks

    CN121788724B