Method for determining description text of rock thin section image, storage medium and processor

Abstract

The embodiment of the application provides a kind of rock slice image description text determination method, storage medium and processor.It includes: establishing the description database of rock slice image;Acquire the target rock slice image to be described;Determine the target prompt vector corresponding to the target rock slice image;According to target prompt vector, determine the rock slice image that is successfully matched in the description database with the target rock slice image;Get the target associated graphic-text vector of the rock slice image that is successfully matched with the target rock slice image;The target associated graphic-text vector is input to rock slice language model, to generate the description text corresponding to the target rock slice image by rock slice language model.The technical scheme described above can obtain more objective description text of rock slice image, improve the efficiency of oil and gas exploration and development, and reduce the labor cost.

Description

Technical Field

[0001] This application relates to the fields of rock casting thin section image identification and artificial intelligence, specifically to a method for determining descriptive text of rock thin section images, a storage medium, and a processor. Background Technology

[0002] The oil and gas industry requires extensive analysis of rock thin-section images. Traditional methods rely primarily on visual observation and description, necessitating expert interpretation and demanding a high level of experience from researchers. Current descriptions of rock thin-section images are highly dependent on the experience of interpreters, leading to potentially contradictory conclusions from different interpreters and a high degree of ambiguity. Existing technologies, reliant on specialized personnel, result in highly subjective rock thin-section description reports, failing to effectively guide oil and gas exploration and development. Summary of the Invention

[0003] The purpose of this application is to provide a method, storage medium, and processor for determining descriptive text of rock thin section images.

[0004] To achieve the above objectives, the first aspect of this application provides a method for determining descriptive text of a rock thin section image, comprising:

[0005] A descriptive database of rock thin section images is established, wherein the descriptive database includes multiple rock thin section images and initial descriptive data corresponding to each rock thin section image;

[0006] Acquire thin section images of the target rock to be described;

[0007] Determine the target cue vector corresponding to the target rock thin section image, wherein the cue vector includes the image features of the target rock thin section image;

[0008] Based on the target cue vector, identify the rock thin section image that successfully matches the target rock thin section image in the description database;

[0009] Obtain the target-related image-text vector of the rock thin section image that successfully matches the target rock thin section image. The target-related image-text vector is obtained through the intelligent description model of the rock thin section image.

[0010] The target associated image-text vector is input into the rock thin section language model to generate descriptive text corresponding to the target rock thin section image.

[0011] In this embodiment of the application, the determination method further includes: after establishing a description database of rock thin section images, sequentially inputting each rock thin section image in the description database and the initial description data corresponding to each rock thin section image into the intelligent description model of rock thin section images to obtain the associated image and text vector corresponding to each rock thin section image.

[0012] In this embodiment, the process of sequentially inputting each rock thin section image from the description database and the initial description data corresponding to each rock thin section image into the intelligent description model of the rock thin section image to obtain the associated graphic-text vector corresponding to each rock thin section image includes: sequentially inputting each rock thin section image from the description database into the intelligent description model of the rock thin section image to generate a prompt vector corresponding to the rock thin section image through the intelligent description model of the rock thin section image; sequentially inputting the initial description data corresponding to each rock thin section image into the intelligent description model of the rock thin section image to generate a text vector corresponding to the initial description data through the intelligent description model of the rock thin section image; and determining the associated graphic-text vector based on the prompt vector and the text vector.

[0013] In this embodiment of the application, each rock thin section image in the description database is sequentially input into the intelligent description model of rock thin section images. The generation of a prompt vector corresponding to the rock thin section image by the intelligent description model of rock thin section images includes: determining the image vector corresponding to the rock thin section image based on the rock thin section image; and determining the prompt vector based on the image vector.

[0014] In this embodiment of the application, the initial description data corresponding to each rock thin section image is sequentially input into the intelligent description model of the rock thin section image. The generation of the text vector corresponding to the initial description data by the intelligent description model of the rock thin section image includes: determining the word vector corresponding to the initial description data based on the initial description data; and determining the text vector based on the word vector.

[0015] In this embodiment of the application, determining the rock thin section image that successfully matches the target rock thin section image in the description database based on the target cue vector includes: comparing the target cue vector with the cue vector of each rock thin section image in the description database to obtain the similarity between the target cue vector and the cue vector of each rock thin section image in the description database; determining the cue vector corresponding to the largest similarity as the actual cue vector that successfully matches the target cue vector; and determining the actual rock thin section image corresponding to the actual cue vector as the rock thin section image that successfully matches the target rock thin section image.

[0016] In this embodiment of the application, the determination method further includes: after inputting the target associated image-text vector into the rock thin section language model to generate descriptive text corresponding to the target rock thin section image through the rock thin section language model, inputting the target rock thin section image and the descriptive text corresponding to the target rock thin section image into the description database to update the description database.

[0017] In this embodiment of the application, the determination method further includes a training step of a rock thin section image intelligent description model. The training step includes: acquiring multiple sample data, wherein each sample data includes a rock thin section image and initial description data corresponding to the rock thin section image; sequentially inputting each sample data into the rock thin section image intelligent description model, and acquiring a first model parameter and a second model parameter corresponding to each sample data, wherein the first model parameter is determined based on the rock thin section image in each sample data, and the second model parameter is determined based on the initial description data in each sample data; determining the prediction loss value of the rock thin section image intelligent description model based on the first model parameter and the second model parameter; determining that the rock thin section image intelligent description model training is complete when the prediction loss value is less than a preset value; and re-executing the steps of sequentially inputting each sample data into the rock thin section image intelligent description model and acquiring the first model parameter and the second model parameter corresponding to each sample data when the prediction loss value is greater than or equal to the preset value, until the prediction loss value is less than the preset value.

[0018] A second aspect of this application provides a processor configured to perform the method for determining descriptive text of a rock thin section image as described above.

[0019] A third aspect of this application provides a machine-readable storage medium storing instructions that, when executed by a processor, configure the processor to perform the method for determining descriptive text of a rock thin section image as described above.

[0020] The above technical solution involves establishing a descriptive database of rock thin section images; acquiring the target rock thin section image to be described; determining the target cue vector corresponding to the target rock thin section image; identifying rock thin section images in the description database that successfully match the target rock thin section image based on the target cue vector; acquiring the target associated image-text vector of the rock thin section image that successfully matches the target rock thin section image; and inputting the target associated image-text vector into a rock thin section language model to generate descriptive text corresponding to the target rock thin section image through the rock thin section language model. By adopting this technical solution, more objective descriptive text for rock thin section images can be obtained, improving the efficiency of oil and gas exploration and development and reducing labor costs.

[0021] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0022] The accompanying drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the following detailed description to explain the embodiments of this application, but do not constitute a limitation on the embodiments of this application. In the drawings:

[0023] Figure 1 The illustration shows a flowchart of a method for determining descriptive text for rock thin section images according to an embodiment of this application;

[0024] Figure 2 The illustration shows a schematic diagram of the training process of an intelligent description model for rock thin section images according to an embodiment of this application;

[0025] Figure 3 The illustration shows a schematic diagram of the application environment of the method for determining descriptive text of rock thin section images according to an embodiment of this application;

[0026] Figure 4 The diagram illustrates the internal structure of a computer device according to an embodiment of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0028] Figure 1 The illustration schematically shows a flowchart of a method for determining descriptive text for rock thin section images according to an embodiment of this application. For example... Figure 1 As shown in one embodiment of this application, a method for determining descriptive text for a rock thin section image is provided, comprising the following steps:

[0029] Step 101: Establish a description database of rock thin section images, wherein the description database includes multiple rock thin section images and initial description data corresponding to each rock thin section image.

[0030] Step 102: Obtain a thin section image of the target rock to be described.

[0031] Step 103: Determine the target cue vector corresponding to the target rock thin section image, wherein the cue vector includes the image features of the target rock thin section image.

[0032] Step 104: Determine the rock thin section image that successfully matches the target rock thin section image in the description database based on the target cue vector.

[0033] Step 105: Obtain the target associated image vector of the rock thin section image that successfully matches the target rock thin section image. The target associated image vector is obtained through the intelligent description model of the rock thin section image.

[0034] Step 106: Input the target associated image-text vector into the rock thin section language model to generate descriptive text corresponding to the target rock thin section image through the rock thin section language model.

[0035] Rock thin section images are a type of thin section of rock, obtained during geological work for rock and mineral identification. They are primarily used to observe and study the structure, texture, mineral composition, and associated assemblages of rocks, to study mineral metamorphism and alteration, to determine the names of rocks and minerals, and to correlate strata and rocks. Initial descriptive data consists of textual descriptions obtained by geological experts from analyzing rock thin section images. The processor can build a descriptive database of rock thin section images, which can include multiple rock thin section images and corresponding initial descriptive data for each image. Specifically, the processor can organize all rock thin section images and obtain the initial descriptive data for each image from geological experts. Then, each rock thin section image is processed with its corresponding initial descriptive data to obtain the descriptive database. The processor can acquire the target rock thin section image to be described and determine the target cue vector corresponding to it. The cue vector includes the image features of the target rock thin section image. Image features include mineral names, physical properties, and structural information. The processor can then determine the rock thin section images in the descriptive database that successfully match the target rock thin section image based on the target cue vector. The processor acquires the target-related image-text vector of the rock thin section image that successfully matches the target rock thin section image. This target-related image-text vector can be obtained through an intelligent description model of the rock thin section image. The target-related image-text vector can include image features and textual information of the rock thin section. The processor can input the target-related image-text vector into a rock thin section language model to generate descriptive text corresponding to the target rock thin section image. The descriptive text refers to a standardized, uniformly formatted statement.

[0036] In this embodiment of the application, determining the rock thin section image that successfully matches the target rock thin section image in the description database based on the target cue vector includes: comparing the target cue vector with the cue vector of each rock thin section image in the description database to obtain the similarity between the target cue vector and the cue vector of each rock thin section image in the description database; determining the cue vector corresponding to the largest similarity as the actual cue vector that successfully matches the target cue vector; and determining the actual rock thin section image corresponding to the actual cue vector as the rock thin section image that successfully matches the target rock thin section image.

[0037] After acquiring the target rock thin section image to be described, the processor can determine the target cue vector corresponding to the target rock thin section image. After obtaining the target cue vector, the processor can compare the similarity between the target cue vector and the cue vectors of each rock thin section image in the description database. After obtaining the similarity scores, the processor can evaluate all similarities to obtain the highest similarity. The cue vector corresponding to the highest similarity is then determined as the actual cue vector that successfully matches the target cue vector. The processor can then determine the actual rock thin section image corresponding to the actual cue vector as the rock thin section image that successfully matches the target rock thin section image.

[0038] For example, the description database includes rock thin section images A1, A2, A3, A4, and A5. The cue vectors for rock thin section images A1, A2, A3, A4, and A5 are rock thin section images a1, a2, a3, a4, and a5, respectively. After acquiring the target rock thin section image X1 to be described, the processor can determine the target cue vector x1 corresponding to the target rock thin section image X1. The processor can sequentially compare x1 with a1, x1 with a2, x1 with a3, x1 with a4, and x1 with a5 to obtain the similarity p1 between x1 and a1, the similarity p2 between x1 and a2, the similarity p3 between x1 and a3, the similarity p4 between x1 and a4, and the similarity p5 between x1 and a5. Among these, p3 > p2 > p5 > p1 > p4. The processor can determine the cue vector a3 corresponding to p3 as the actual cue vector that successfully matches the target cue vector x1, and determine the actual rock thin section image A3 corresponding to the actual cue vector as the rock thin section image that successfully matches the target rock thin section image X1.

[0039] In one embodiment, the determination method further includes: after establishing a description database of rock thin section images, sequentially inputting each rock thin section image in the description database and the initial description data corresponding to each rock thin section image into the intelligent description model of rock thin section images to obtain the associated image-text vector corresponding to each rock thin section image.

[0040] The processor can establish a descriptive database of rock thin section images. After establishing the descriptive database, the processor can sequentially input each rock thin section image in the descriptive database and the initial descriptive data corresponding to each rock thin section image into a rock thin section image intelligent description model to obtain an associated image-text vector corresponding to each rock thin section image. Specifically, in one embodiment, the processor can sequentially input each rock thin section image in the descriptive database into the rock thin section image intelligent description model to generate a cue vector corresponding to the rock thin section image. The processor can also sequentially input the initial descriptive data corresponding to each rock thin section image into the rock thin section image intelligent description model to generate a text vector corresponding to the initial descriptive data. The processor can determine the associated image-text vector corresponding to each rock thin section image based on the cue vector and the text vector.

[0041] For example, after establishing a description database of rock thin section images, the processor can sequentially input the rock thin section images A1, A2, A3, A4, and A5 from the description database, along with the initial description data Y1, Y2, Y3, Y4, and Y5 corresponding to the rock thin section images A1, A2, A3, A4, and A5, into the intelligent description model of the rock thin section images to obtain the associated image-text vectors m1, m2, m3, m4, and m5 corresponding to the rock thin section images A1, A2, A3, A4, and A5, respectively. Specifically, the processor can sequentially input the rock thin section images A1, A2, A3, A4, and A5 into the intelligent description model of the rock thin section images to generate the prompt vectors a1, a2, a3, a4, and a5 corresponding to the rock thin section images A1, A2, A3, A4, and A5, respectively. The processor can sequentially input initial description data Y1, Y2, Y3, Y4, and Y5 into the intelligent description model of rock thin section images. The model then generates text vectors y1, y2, y3, y4, and y5, corresponding to the initial description data Y1, Y2, Y3, Y4, and Y5, respectively. The processor can then determine the associated text vector m1 based on a1 and y1, m2 based on a2 and y2, m3 based on a3 and y3, m4 based on a4 and y4, and m5 based on a5 and y5.

[0042] In one embodiment, each rock thin section image in the description database is sequentially input into the intelligent description model for rock thin section images. Generating a prompt vector corresponding to the rock thin section image through the intelligent description model includes: determining an image vector corresponding to the rock thin section image based on the rock thin section image; and determining a prompt vector based on the image vector. An image vector is a vector that includes various image information of the rock thin section image.

[0043] After the processor sequentially inputs each rock thin section image from the description database into the intelligent description model for rock thin section images, it can determine the image vector corresponding to that rock thin section image. Once the corresponding image vector is determined, the processor can then determine the corresponding cue vector for the rock thin section image.

[0044] In one embodiment, initial description data corresponding to each rock thin section image is sequentially input into a rock thin section image intelligent description model. Generating text vectors corresponding to the initial description data through the rock thin section image intelligent description model includes: determining word vectors corresponding to the initial description data based on the initial description data; and determining text vectors based on the word vectors. Word vectors refer to vectors formed by processing the initial description data according to natural language rules and then extracting word information from the processed initial description data.

[0045] After the processor sequentially inputs the initial description data corresponding to each rock thin section image into the intelligent description model of the rock thin section image, it can determine the word vectors corresponding to the initial description data. After determining the corresponding word vectors, the processor can determine the text vectors corresponding to the initial description data based on the word vectors.

[0046] In one embodiment, the determination method further includes: after inputting the target associated image-text vector into a rock thin section language model to generate descriptive text corresponding to the target rock thin section image through the rock thin section language model, inputting the target rock thin section image and the descriptive text corresponding to the target rock thin section image into a description database to update the description database.

[0047] The processor can input the target associated image-text vector into a rock thin section language model to generate descriptive text corresponding to the target rock thin section image. After generating the descriptive text, the processor can input the target rock thin section image and the corresponding descriptive text into a description database to update the description database.

[0048] In one embodiment, the determination method further includes a training step of a rock thin section image intelligent description model. The training step includes: acquiring multiple sample data, wherein each sample data includes a rock thin section image and initial description data corresponding to the rock thin section image; sequentially inputting each sample data into the rock thin section image intelligent description model, and acquiring a first model parameter and a second model parameter corresponding to each sample data, wherein the first model parameter is determined based on the rock thin section image in each sample data, and the second model parameter is determined based on the initial description data in each sample data; determining the prediction loss value of the rock thin section image intelligent description model based on the first model parameter and the second model parameter; determining that the rock thin section image intelligent description model training is complete when the prediction loss value is less than a preset value; and repeating the steps of sequentially inputting each sample data into the rock thin section image intelligent description model and acquiring the first model parameter and the second model parameter corresponding to each sample data when the prediction loss value is greater than or equal to the preset value, until the prediction loss value is less than the preset value.

[0049] The processor can train an intelligent description model for rock thin-section images. The processor can acquire multiple sample data sets, each including a rock thin-section image and corresponding initial description data. The processor can sequentially input each sample data set into the intelligent description model and acquire first and second model parameters for each sample data set. The first model parameters can be determined based on the rock thin-section image in each sample data set, and the second model parameters can be determined based on the initial description data in each sample data set. The processor can determine the prediction loss value of the intelligent description model based on the first and second model parameters. After determining the prediction loss value, the processor can determine whether the prediction loss value is less than a preset value. If the prediction loss value is less than the preset value, the processor can determine that the training of the intelligent description model for rock thin-section images is complete. If the prediction loss value is greater than or equal to the preset value, the processor can repeat the steps of inputting each sample data set into the intelligent description model and acquiring the first and second model parameters for each sample data set until the prediction loss value is less than the preset value.

[0050] For example, such as Figure 2As shown, the processor can perform image data enhancement processing on the collected rock thin section images and process the rock image descriptions using a unified description template. The processor can establish a rock thin section image description database based on the processed rock thin section images and rock image descriptions. A rock thin section image description model is constructed using an image description artificial intelligence algorithm. The processor can optimize the model parameters by sequentially inputting the rock thin section images and rock image descriptions from the rock thin section image description database into the rock thin section image description model and obtaining the first model parameters corresponding to the rock thin section images and the second model parameters corresponding to the rock image descriptions. The model evaluation index of the rock thin section image description model is determined based on the first model parameters and the second model parameters. Specifically, the model evaluation index can be calculated using formula (1):

[0051]

[0052] Among them, L i Let I be the i-th evaluation metric for the rock thin section image description model. i T represents the first model parameter corresponding to the i-th rock thin section image. i Let r be the second model parameter corresponding to the i-th rock image description, r be a constant, and M be the degree of matching between the rock thin section image and the rock image description.

[0053] After obtaining the model evaluation metrics, the processor can analyze them to determine whether the metrics meet the requirements. If the requirements are met, a rock thin section image intelligent description model is obtained; if the requirements are not met, the model parameters are optimized until the model evaluation metrics meet the requirements. The processor can determine the intelligent description result of the rock thin section image using the rock thin section image intelligent description model.

[0054] In one embodiment, such as Figure 3As shown, the processor can input rock thin-section images into the intelligent rock thin-section image description model. Within the model, the processor can call the RiCim Encoder (rock thin-section image processing model) to process the input image and obtain the corresponding image vector. Then, it calls the RiCim (rock thin-section image feature editing model) to process the image vector and obtain the corresponding cue vector. After inputting the rock image label information corresponding to the thin-section image into the model, the processor calls the RiC-LanG Encoder (rock thin-section language processing model) to process the label information and obtain the corresponding word vector. Then, it calls the RiC-LanG (rock thin-section language editing model) to process the word vector and obtain the corresponding text vector. The processor can perform a rand-concat operation (rock thin-section image-text information association operation) on the cue vector and text vector to obtain the corresponding associated image-text vector. The processor can also calculate the loss of the intelligent rock thin-section image description model based on the word vector and cue vector.

[0055] The processor can call the RiCim Encoder (rock thin section image processing model) to process a new rock thin section image to obtain the corresponding predicted image vector. Then, it calls the RiCim (rock thin section image feature editing model) to process the predicted image vector to obtain the corresponding cue vector. The processor can then determine the associated text-image vector corresponding to the new rock thin section image based on the cue vector. Finally, the processor can call the RiC-LanG (rock thin section language editing model) to process the associated text-image vector to obtain the prediction result for the new rock thin section image.

[0056] The above technical solution involves establishing a descriptive database of rock thin section images; acquiring the target rock thin section image to be described; determining the target cue vector corresponding to the target rock thin section image; identifying rock thin section images in the description database that successfully match the target rock thin section image based on the target cue vector; acquiring the target associated image-text vector of the rock thin section image that successfully matches the target rock thin section image; and inputting the target associated image-text vector into a rock thin section language model to generate descriptive text corresponding to the target rock thin section image through the rock thin section language model. By adopting this technical solution, more objective descriptive text for rock thin section images can be obtained, improving the efficiency of oil and gas exploration and development and reducing labor costs.

[0057] Figure 1 , 2 This is a flowchart illustrating a method for determining descriptive text for a rock thin section image in one embodiment. It should be understood that, although... Figure 1 , 2The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 , 2 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but may be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0058] This application provides a storage medium storing a program that, when executed by a processor, implements the method for determining the descriptive text of the rock thin section image described above.

[0059] This application provides a processor for running a program, wherein the program executes a method for determining the descriptive text of the rock thin section image.

[0060] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor A01, a network interface A02, a memory (not shown), and a database (not shown) connected via a system bus. The processor A01 provides computing and control capabilities. The memory includes internal memory A03 and a non-volatile storage medium A04. The non-volatile storage medium A04 stores an operating system B01, a computer program B02, and a database (not shown). The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 stored in the non-volatile storage medium A04. The database stores initial descriptive data and descriptive text data for rock thin section images. The network interface A02 communicates with external terminals via a network connection. When executed by the processor A01, the computer program B02 implements a method for determining descriptive text for rock thin section images.

[0061] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0062] This application provides an apparatus including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs the following steps: establishing a description database of rock thin-section images, wherein the description database includes multiple rock thin-section images and initial description data corresponding to each rock thin-section image; acquiring a target rock thin-section image to be described; determining a target cue vector corresponding to the target rock thin-section image, wherein the cue vector includes image features of the target rock thin-section image; determining rock thin-section images in the description database that successfully match the target rock thin-section image based on the target cue vector; acquiring a target associated text-image vector of the rock thin-section image that successfully matches the target rock thin-section image, wherein the target associated text-image vector is obtained through a rock thin-section image intelligent description model; and inputting the target associated text-image vector into a rock thin-section language model to generate descriptive text corresponding to the target rock thin-section image through the rock thin-section language model.

[0063] In one embodiment, the determination method further includes: after establishing a description database of rock thin section images, sequentially inputting each rock thin section image in the description database and the initial description data corresponding to each rock thin section image into the intelligent description model of rock thin section images to obtain the associated image-text vector corresponding to each rock thin section image.

[0064] In one embodiment, sequentially inputting each rock thin section image from the description database and the initial description data corresponding to each rock thin section image into the intelligent description model of the rock thin section image to obtain the associated graphic-text vector corresponding to each rock thin section image includes: sequentially inputting each rock thin section image from the description database into the intelligent description model of the rock thin section image to generate a prompt vector corresponding to the rock thin section image through the intelligent description model of the rock thin section image; sequentially inputting the initial description data corresponding to each rock thin section image into the intelligent description model of the rock thin section image to generate a text vector corresponding to the initial description data through the intelligent description model of the rock thin section image; and determining the associated graphic-text vector based on the prompt vector and the text vector.

[0065] In one embodiment, each rock thin section image in the description database is sequentially input into the intelligent description model of rock thin section images. Generating a prompt vector corresponding to the rock thin section image through the intelligent description model of rock thin section images includes: determining the image vector corresponding to the rock thin section image based on the rock thin section image; and determining the prompt vector based on the image vector.

[0066] In one embodiment, initial description data corresponding to each rock thin section image is sequentially input into the intelligent description model of the rock thin section image. Generating a text vector corresponding to the initial description data through the intelligent description model of the rock thin section image includes: determining word vectors corresponding to the initial description data based on the initial description data; and determining the text vector based on the word vectors.

[0067] In one embodiment, determining the rock thin section image that successfully matches the target rock thin section image in the description database based on the target cue vector includes: comparing the target cue vector with the cue vector of each rock thin section image in the description database to obtain the similarity between the target cue vector and the cue vector of each rock thin section image in the description database; determining the cue vector corresponding to the largest similarity as the actual cue vector that successfully matches the target cue vector; and determining the actual rock thin section image corresponding to the actual cue vector as the rock thin section image that successfully matches the target rock thin section image.

[0068] In one embodiment, the determination method further includes: after inputting the target associated image-text vector into a rock thin section language model to generate descriptive text corresponding to the target rock thin section image through the rock thin section language model, inputting the target rock thin section image and the descriptive text corresponding to the target rock thin section image into a description database to update the description database.

[0069] In one embodiment, the determination method further includes a training step of a rock thin section image intelligent description model. The training step includes: acquiring multiple sample data, wherein each sample data includes a rock thin section image and initial description data corresponding to the rock thin section image; sequentially inputting each sample data into the rock thin section image intelligent description model, and acquiring a first model parameter and a second model parameter corresponding to each sample data, wherein the first model parameter is determined based on the rock thin section image in each sample data, and the second model parameter is determined based on the initial description data in each sample data; determining the prediction loss value of the rock thin section image intelligent description model based on the first model parameter and the second model parameter; determining that the rock thin section image intelligent description model training is complete when the prediction loss value is less than a preset value; and repeating the steps of sequentially inputting each sample data into the rock thin section image intelligent description model and acquiring the first model parameter and the second model parameter corresponding to each sample data when the prediction loss value is greater than or equal to the preset value, until the prediction loss value is less than the preset value.

[0070] This application also provides a computer program product that, when executed on a data processing device, is adapted to perform method steps for determining descriptive text such as a rock thin-section image.

[0071] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0072] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0073] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0074] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0075] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0076] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0077] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0078] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0079] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for determining descriptive text for a rock thin section image, characterized in that, The determination method includes: A descriptive database for the rock thin section images is established, wherein the descriptive database includes multiple rock thin section images and initial descriptive data corresponding to each rock thin section image; Each rock thin section image in the description database and the initial description data corresponding to each rock thin section image are sequentially input into the intelligent description model of the rock thin section image to obtain the associated image and text vector corresponding to each rock thin section image; Acquire thin section images of the target rock to be described; Determine a target cue vector corresponding to the target rock thin section image, wherein the cue vector includes image features of the target rock thin section image; Based on the target cue vector, determine the rock thin section image in the description database that successfully matches the target rock thin section image; Obtain the target associated image vector of the rock thin section image that successfully matches the target rock thin section image, wherein the target associated image vector is obtained through a rock thin section image intelligent description model; The target associated image-text vector is input into the rock thin section language model to generate descriptive text corresponding to the target rock thin section image through the rock thin section language model.

2. The determination method according to claim 1, characterized in that, The step of sequentially inputting each rock thin section image from the description database and the initial description data corresponding to each rock thin section image into the intelligent description model of the rock thin section image to obtain the associated image-text vector corresponding to each rock thin section image includes: Each rock thin section image in the description database is sequentially input into the intelligent description model of the rock thin section image, so as to generate a prompt vector corresponding to the rock thin section image through the intelligent description model of the rock thin section image; Initial description data corresponding to each rock thin section image is sequentially input into the intelligent description model of the rock thin section image, so as to generate a text vector corresponding to the initial description data through the intelligent description model of the rock thin section image; The associated image-text vector is determined based on the prompt vector and the text vector.

3. The determination method according to claim 2, characterized in that, The step of sequentially inputting each rock thin section image from the description database into the intelligent description model of the rock thin section image, and generating a prompt vector corresponding to the rock thin section image through the intelligent description model of the rock thin section image, includes: Determine the image vector corresponding to the rock thin section image based on the rock thin section image; The prompt vector is determined based on the image vector.

4. The determination method according to claim 2, characterized in that, The step of sequentially inputting initial description data corresponding to each rock thin section image into the intelligent description model of the rock thin section image, and generating a text vector corresponding to the initial description data through the intelligent description model of the rock thin section image includes: Determine the word vectors corresponding to the initial description data based on the initial description data; The text vector is determined based on the word vectors.

5. The determination method according to claim 1, characterized in that, The step of determining the rock thin section image in the description database that successfully matches the target rock thin section image based on the target cue vector includes: The target cue vector is compared with the cue vector of each rock thin section image in the description database to obtain the similarity between the target cue vector and the cue vector of each rock thin section image in the description database; The suggestion vector corresponding to the highest similarity is determined as the actual suggestion vector that successfully matches the target suggestion vector; The actual rock thin section image corresponding to the actual prompt vector is determined to be the rock thin section image that successfully matches the target rock thin section image.

6. The determination method according to claim 1, characterized in that, The determination method further includes: After the target associated image-text vector is input into the rock thin section language model to generate descriptive text corresponding to the target rock thin section image through the rock thin section language model, the target rock thin section image and the descriptive text corresponding to the target rock thin section image are input into the description database to update the description database.

7. The determination method according to claim 1, characterized in that, The determination method further includes a training step for the intelligent description model of the rock thin section image, the training step including: Multiple sample data are acquired, wherein each sample data includes a rock thin section image and initial description data corresponding to the rock thin section image; Each sample data is sequentially input into the intelligent description model of the rock thin section image, and the first model parameter and the second model parameter corresponding to each sample data are obtained. The first model parameter is determined based on the rock thin section image in each sample data, and the second model parameter is determined based on the initial description data in each sample data. The prediction loss value of the intelligent description model of the rock thin section image is determined based on the first model parameters and the second model parameters; If the predicted loss value is less than a preset value, the training of the intelligent description model for the rock thin section image is determined to be complete. If the predicted loss value is greater than or equal to the preset value, the steps of sequentially inputting each sample data into the intelligent description model of the rock thin section image and obtaining the first model parameter and the second model parameter corresponding to each sample data are executed again until the predicted loss value is less than the preset value.

8. A processor, characterized in that, The method is configured to perform the method for determining descriptive text of a rock thin section image according to any one of claims 1 to 7.

9. A machine-readable storage medium storing instructions thereon, characterized in that, When executed by a processor, this instruction causes the processor to be configured to perform a method for determining descriptive text of a rock thin section image according to any one of claims 1 to 7.

Images