End-to-end pulsar candidate identification method and device, computer equipment

Abstract

The application relates to an end-to-end pulsar candidate identification method and device and a computer device. The method comprises the following steps: acquiring an original pulsar candidate image corresponding to a pulsar candidate signal, performing enhancement processing on the original pulsar candidate image to obtain a pulsar candidate image; labeling a sub-image in the pulsar candidate image to construct a target data set for identifying a pulsar signal type; constructing a positioning data set by labeling position information of the sub-image in the original pulsar candidate image; training a preset pulsar identification model according to the positioning data set to obtain a positioning model for identifying the position information of the sub-image; acquiring a preset inference prompt, training the positioning model according to the preset inference prompt and the target data set, and obtaining a pulsar identification model for identifying the pulsar signal type. The method can improve the pulsar candidate identification accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image processing technology of artificial intelligence, in particular to an end-to-end pulsar candidate identification method and device and computer equipment. BACKGROUND

[0002] A pulsar is a neutron star with highly periodic pulsar radiation. By identifying and utilizing the location and signal of the pulsar, a high-precision cosmic navigation system can be established to help spacecraft determine position and navigation in deep space exploration and improve navigation accuracy.

[0003] With the rapid development of deep learning technology, pulsar search technology is also continuously improving. Applying deep learning technology to the identification of pulsar candidates can achieve efficient and automated screening of pulsar candidates. However, in related technologies, local features in candidate images are extracted for identification, which cannot accurately identify pulsar candidates. SUMMARY

[0004] Therefore, it is necessary to provide an end-to-end pulsar candidate identification method, device, computer equipment, computer readable storage medium and computer program product that can improve the accuracy of end-to-end pulsar candidate identification to solve the above technical problems.

[0005] In a first aspect, the present application provides an end-to-end pulsar candidate identification method, comprising:

[0006] Obtaining an original pulsar candidate image corresponding to a pulsar candidate signal, performing enhancement processing on the original pulsar candidate image to obtain a pulsar candidate image;

[0007] Labeling subgraphs in the pulsar candidate image to construct a target data set for identifying the type of pulsar signal;

[0008] Constructing a positioning data set by labeling the position information of the subgraphs in the original pulsar candidate image;

[0009] Training a preset pulsar identification model according to the positioning data set to obtain a positioning model for identifying subgraph position information;

[0010] Obtaining a preset inference prompt, training the positioning model according to the preset inference prompt and the target data set to obtain a pulsar identification model for identifying the type of pulsar signal.

[0011] In one embodiment, the original pulsar candidate image is enhanced to obtain a pulsar candidate image, comprising:

[0012] Identifying the subgraph frequency phase graph in the original pulsar candidate image;

[0013] perform data enhancement processing on the subgraph frequency phase graph to obtain a pulsar candidate image.

[0014] In one of the embodiments, the labeling of the subgraph in the original pulsar candidate image to construct a target data set for identifying the pulsar signal type comprises:

[0015] determining the subgraph frequency phase graph and the subgraph dispersion curve graph in the original pulsar candidate image;

[0016] identifying a first morphological feature of the subgraph frequency phase graph, and determining first labeling information of the subgraph frequency phase graph according to the identification result of the first morphological feature;

[0017] identifying a second morphological feature of the subgraph dispersion curve graph, and determining second labeling information of the subgraph dispersion curve graph according to the identification result of the second morphological feature and the pulsar candidate image;

[0018] constructing a target data set for identifying the pulsar signal type according to the first labeling information, the second labeling information, and the corresponding pulsar candidate image.

[0019] In one of the embodiments, the identification of the first morphological feature of the subgraph frequency phase graph and the determination of the first labeling information of the subgraph frequency phase graph according to the identification result of the first morphological feature comprise:

[0020] identifying the first morphological feature of the subgraph frequency phase graph, and determining a phase difference corresponding to the intersection of the target region of the preset shape in the identification result of the first morphological feature and the horizontal and vertical coordinates of the subgraph frequency phase graph;

[0021] if the phase difference is within a preset range, determining the first labeling information of the subgraph frequency phase graph as a first preset label, and the first preset label is used to represent that the subgraph frequency phase graph meets the characteristics of the real pulsar signal frequency phase graph;

[0022] if the phase difference is not within the preset range, determining the first labeling information as a second preset label, and the second preset label is used to represent that the subgraph frequency phase graph does not meet the characteristics of the real pulsar signal frequency phase graph.

[0023] In one of the embodiments, the identification of the second morphological feature of the subgraph dispersion curve graph and the determination of the second labeling information of the subgraph dispersion curve graph according to the identification result of the second morphological feature and the pulsar candidate image comprise:

[0024] identify a second morphological feature of the subgraph dispersion curve diagram, and if a preset change trend exists in the identification result of the second morphological feature, the number of wave crests of the preset change trend is a first preset value, and a target dispersion value in the pulsar candidate image is not a second preset value, determine that the second label information of the subgraph dispersion curve diagram is a third preset label; the third preset label is used to represent that the real pulsar signal dispersion curve diagram feature is met;

[0025] If the identification result of the second morphological feature does not exist a preset change trend and / or the target dispersion value is the second preset value, determine that the second label information of the subgraph dispersion curve diagram is a fourth preset label; the fourth preset label is used to represent that the real pulsar signal dispersion curve diagram feature is not met.

[0026] In one of the embodiments, the constructing the positioning data set by labeling the position information of the subgraph in the original pulsar candidate image comprises:

[0027] Labeling the first position information of the subgraph in the original pulsar candidate image;

[0028] Performing a scale transformation processing on the original pulsar candidate image to generate a plurality of size-adjusted images with different resolutions;

[0029] Determining the scale ratio between each of the size-adjusted images and the original pulsar candidate image, mapping the first position information according to the scale ratio, and determining the second position information of the subgraph in each of the size-adjusted images;

[0030] Constructing the positioning data set according to the first position information and the second position information.

[0031] In one of the embodiments, the preset inference prompt comprises:

[0032] If the first preset label and the third preset label exist in the pulsar candidate image, it is determined that the pulsar candidate signal corresponding to the original pulsar candidate image is a pulsar signal;

[0033] If the first preset label and the third preset label do not exist in the pulsar candidate image at the same time, it is determined that the pulsar candidate signal is radio frequency interference.

[0034] In one of the embodiments, the method further comprises:

[0035] Obtaining a target pulsar candidate image corresponding to a to-be-identified pulsar candidate signal;

[0036] input the pulsar candidate image to be identified into the pulsar identification model, and output a signal type of the pulsar candidate signal to be identified.

[0037] In a second aspect, the present application further provides an end-to-end pulsar candidate identification device, comprising:

[0038] An image processing module is configured to acquire an original pulsar candidate image corresponding to a pulsar candidate signal, and perform enhancement processing on the original pulsar candidate image to obtain a pulsar candidate image.

[0039] A first data set construction module is configured to label a sub-image in the pulsar candidate image, and construct a target data set for identifying a signal type of a pulsar.

[0040] A second data set construction module is configured to label position information of the sub-image in the original pulsar candidate image, and construct a positioning data set.

[0041] A first training module is configured to train a preset pulsar identification model according to the positioning data set, and obtain a positioning model for identifying the position information of the sub-image.

[0042] A second training module is configured to acquire a preset inference prompt, train the positioning model according to the preset inference prompt and the target data set, and obtain a pulsar identification model for identifying the signal type of the pulsar.

[0043] In a third aspect, the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method according to any one of the above aspects when executing the computer program.

[0044] In a fourth aspect, the present application further provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the method according to any one of the above aspects when executed by a processor.

[0045] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, and the computer program implements the steps of the method according to any one of the above aspects when executed by a processor.

[0046] The end-to-end pulsar candidate body identification method, device, computer device, computer readable storage medium and computer program product balance the data set, increase the diversity of the data set, finely label the subgraphs in the image, and improve the identification capability of the model on the pulsar. The position information of the subgraph in the original pulsar candidate image is labeled to obtain the positioning data set. The preset pulsar identification model is trained using the positioning data set, so that the model can accurately identify and frame the region of the subgraph, and the positioning capability of the model on the subgraph is improved. The positioning model is trained based on the preset inference prompt and the target data set, so that the model can more effectively learn the characteristics of the pulsar, and guide the model to output the identification result based on the image characteristics. Compared with the information fragmentation problem caused by the cutting method of the related art, the comprehensiveness and accuracy of identification are further improved. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the description of the embodiments of the present application or the related art will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other related drawings can be obtained by those skilled in the art without creative labor.

[0048] Figure 1 An application environment diagram of the end-to-end pulsar candidate body identification method in an embodiment;

[0049] Figure 2 A flowchart of the end-to-end pulsar candidate body identification method in an embodiment;

[0050] Figure 3 A schematic diagram of the original pulsar candidate image in an embodiment;

[0051] Figure 4 A flowchart of the 204 step in an embodiment;

[0052] Figure 5 A flowchart of the construction method of the positioning data set in an embodiment;

[0053] Figure 6 A schematic diagram of the target pulsar candidate image in an embodiment;

[0054] Figure 7 A flowchart of the end-to-end pulsar candidate body identification method in another embodiment;

[0055] Figure 8A structural block diagram of an end-to-end pulsar candidate identification device in an embodiment;

[0056] Figure 9 An internal structural diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0057] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0058] With the rapid development of deep learning technology, researchers have begun to explore the use of deep learning methods to directly extract identification features from images of pulsar candidates. For example, the PICS (Pulsar Image-based Classification System), PICS-ResNet and CCNN (Concat Convolutional Neural Network) methods are all based on convolutional neural networks to automatically extract local features in candidate images, achieving efficient and automated screening of pulsar candidates. These studies have confirmed the great potential of image pattern recognition technology in processing astronomical big data.

[0059] However, traditional CNN (Convolutional Neural Network)-based identification methods mainly rely on local receptive fields and are difficult to fully model the global context information in the image, which may lead to an incomplete understanding of the overall structure of the pulsar signal. These methods usually take subgraphs as input and only focus on visual features, ignoring equally important numerical features. This further leads to low accuracy of the cutout model.

[0060] Based on this, an end-to-end pulsar candidate intelligent identification method based on a multimodal large language model that integrates visual reasoning is proposed. This approach can integrate multimodal information such as subgraphs and numerical data, avoiding the loss of relevance caused by data segmentation. It should be noted that the end-to-end identification method can be understood as directly identifying the entire pulsar candidate image, rather than cutting the image into multiple subgraphs for separate analysis.

[0061] The end-to-end pulsar candidate identification method provided by the embodiments of the present application can be applied to, for example Figure 1The application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data required by the server 104 to process. The data storage system can be integrated on the server 104, or placed on the cloud or other network servers. The original pulsar candidate image corresponding to the pulsar candidate signal is obtained, the original pulsar candidate image is enhanced to obtain the pulsar candidate image, the subgraph in the pulsar candidate image is labeled, and the target data set for identifying the type of pulsar signal is constructed. By labeling the position information of the subgraph in the original pulsar candidate image, a positioning data set is constructed. According to the positioning data set, the preset pulsar recognition model is trained to obtain a positioning model for identifying subgraph position information; obtain a preset inference prompt, and train the positioning model according to the preset inference prompt and the target data set to obtain a pulsar recognition model for identifying the type of pulsar signal.

[0062] Among them, the terminal 102 can be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, Internet of Things devices and portable wearable devices, etc. The server 104 can be a standalone physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0063] In an exemplary embodiment, as Figure 2 shown, an end-to-end pulsar candidate identification method is provided, which is applied to the terminal in Figure 1 The following steps 202 to 210 are described by way of example. Among them:

[0064] Step 202, obtaining the original pulsar candidate image corresponding to the pulsar candidate signal, and enhancing the original pulsar candidate image to obtain the pulsar candidate image.

[0065] Among them, the image acquisition method can be determined by existing methods, which will not be described here. The original pulsar candidate image includes a variety of modal information, for example, it can include a dispersion curve graph, an accumulation profile graph, a frequency phase graph, a time phase graph, and numerical information on the original pulsar candidate image. The numerical information on the original pulsar candidate image, for example, can be the best dispersion value. As Figure 3 shown, the original pulsar candidate image includes a dispersion curve graph, an accumulation profile graph, a frequency phase graph, a time phase graph, and a best dispersion value. The dispersion curve graph, the accumulation profile graph, the frequency phase graph, and the time phase graph are subgraphs of the original pulsar candidate image, so the corresponding subgraphs can also be called subgraph dispersion curve graph, subgraph accumulation profile graph, subgraph frequency phase graph, and subgraph time phase graph. Among them, the horizontal coordinate "DM" in the dispersion curve graph represents the dispersion value, with a unit of pc / cm 3 ; the vertical coordinate " "ChiSqRed" represents the reduced chi-square value; "2Pulses of Best Profile" in the cumulative profile plot represents 2 pulses of the best profile; "Sub-band", "Phase", and "Frequency" in the frequency-phase plot represent a sub-band, a phase, and a frequency in MHz, respectively; and "Phase" and "Time" in the time-phase plot represent a phase and a time, respectively. It should be noted that the specific meanings of other subplots can be understood with reference to related technologies in the art, which are not listed here.

[0066] It can be understood that the image enhancement processing technology has relevant applications in image processing. However, in the embodiment, the image enhancement processing is considered in view of the scarcity of real pulsar signals and the serious imbalance between positive and negative samples. Therefore, the original pulsar candidate image as a whole or a part thereof (for example, data enhancement processing is performed on the frequency-phase plot) is enhanced to better generalize the signal position changes that may occur in real observation data, thereby increasing the diversity of data and improving the generalization ability of the model.

[0067] The image enhancement processing mode includes but is not limited to one or more of the following operations: translation, flipping, rotation, and contrast transformation. The translation can be a translation of the image by several pixels in the horizontal direction (phase axis). When translating to the right, the right pixel values are used to fill the blank area on the left. When translating to the left, the left pixel values are used to fill the blank area on the right. The flipping can be a mirror operation of the image in the horizontal or vertical direction. The rotation can be a rotation of the image by 180 degrees clockwise or counterclockwise around the center point. The contrast transformation can be a gamma transformation, histogram equalization, etc., to adjust the contrast of the image, making the bright area brighter or the dark area darker. Alternatively, the training model requires the construction of a corresponding data set, and the number of original pulsar candidate images obtained is multiple. For ease of understanding, the present example is described with reference to one original pulsar candidate image or a pulsar candidate image.

[0068] Illustratively, the pulsar candidate data set is obtained from the server, and the pulsar candidate data set includes a training set Dtrain and a test set Dtest. The original pulsar candidate image corresponding to each pulsar candidate signal in the training set Dtrain is enhanced to obtain a pulsar candidate image, and a new training sample Dtraind required for constructing a pulsar recognition model is constructed according to the plurality of pulsar candidate images.

[0069] In step 204, the subplots in the pulsar candidate image are labeled to construct a target data set for identifying the type of pulsar signal.

[0070] The sub-image annotation can be adding labels in the image data so that the model can understand the content of the image. The purpose of annotation is to convert the image data into structured data, so that the model can learn the relationship between the image and the label. The label can be category, attribute, position, etc. For example, in the image classification task, a category label is given to each image; in the target detection task, the position and category of the target are annotated.

[0071] Optionally, the annotation of the sub-image in the pulsar candidate image can be annotation of the characteristics of the pulsar, such as morphological characteristics. The morphological characteristics can be pulse period, pulse profile, and curve trend, etc. The target data set can also be understood as an annotated data set, which can be used for model training and model performance evaluation, etc. The annotation method can be but not limited to one or more of self-defined annotation tools, image annotation tools, or Web-based annotation tools, and the implementation method can be realized by existing methods, which will not be described here.

[0072] It can be understood that the annotated sub-image can be a sub-image with a morphological differentiation degree meeting a preset condition, and the preset condition is determined according to actual demand or historical data.

[0073] Step 206, constructing a positioning data set by annotating the position information of the sub-image in the original pulsar candidate image.

[0074] Illustratively, the position information of the sub-image in the original pulsar candidate image is annotated to obtain the coordinate position of the sub-image in the original pulsar candidate image. The original pulsar candidate image is processed by scale transformation to generate a plurality of images with different resolutions. For each size-transformed image, the new coordinates of the sub-image are calculated and mapped according to the size ratio of the image to the original image, thereby constructing the corresponding annotation information for the sub-image of different sizes to obtain the corresponding positioning data set.

[0075] Step 208, training a preset pulsar recognition model according to the positioning data set to obtain a positioning model for recognizing the position information of the sub-image.

[0076] The preset pulsar recognition model can be a base model, for example, a multi-modal large language model. The preset pulsar recognition model can be fine-tuned.

[0077] Exemplarily, a multimodal large language model qwen2.5-vl is used as a base model, a preset pulsar identification model is trained according to a positioning data set, so that the position information of a subgraph in a pulsar image is learned, and when a preset training end condition is met, a positioning model for identifying the position information of the subgraph is obtained. Further, the positioning data set can be verified based on the constructed positioning data set to ensure the correctness and consistency of the subgraph coordinate positions verified by different size images, thereby ensuring the reliability and practicality of the identification result.

[0078] The multimodal large language model uses a Transformer architecture and can process multiple data modalities, and is suitable for tasks that require the integration of multiple types of data. For example, it can maintain high recognition performance under various observation conditions and different distributions of data, demonstrating better generalization ability than traditional deep learning models, and has strong adaptability, making it suitable for actual pulsar search tasks.

[0079] In step 210, a preset reasoning prompt is obtained, and the positioning model is trained according to the preset reasoning prompt and the target data set to obtain a pulsar identification model for identifying the type of pulsar signal.

[0080] During the training process, the reasoning method with "thinking" ability is used to guide the model to output the identification result based on the image features. The preset reasoning prompt can be determined according to actual needs to guide the positioning model to gradually reason, rather than directly giving the answer. The preset reasoning prompt can be that if the morphological features of the subgraph in the pulsar candidate image meet the morphological features of the real pulsar signal and the numerical information related to the subgraph in the pulsar candidate image is not the preset value, then the signal corresponding to the pulsar candidate image is identified as a pulsar signal, otherwise it is identified as radio frequency interference. The training of the pulsar identification model can be fine-tuning of the weights of the positioning model.

[0081] Exemplarily, a preset reasoning prompt is obtained, and during the training process of the positioning model based on the target data set, the preset reasoning prompt is used, that is, a reasoning method with "thinking" ability is used to guide the model to output the identification result based on the image features, and when a preset condition for ending the training of the positioning model is met, a pulsar identification model for identifying the type of pulsar signal is obtained. The training can be fine-tuning of the weights of the positioning model to generate the pulsar identification model. Further, after the training is completed, the pulsar identification model is verified based on the test set, the output identification result is analyzed, and the accuracy and reliability of the pulsar identification model are verified. For example, the correct classification of the pulsar signal and the radio frequency interference is verified.

[0082] In the aforementioned end-to-end pulsar candidate recognition method, the entire original pulsar candidate image is enhanced, and sub-images are annotated based on the enhanced pulsar candidate image to obtain the target dataset. This approach balances the dataset, increases its diversity, and provides fine-grained annotation of sub-images within the image, improving the model's ability to recognize pulsars. A localization dataset is obtained by annotating the positional information of sub-images within the entire original pulsar candidate image. This localization dataset is then used to train a pre-defined pulsar recognition model, enabling the model to accurately identify and define regions within sub-images, thus improving its localization capabilities. Finally, the localization model is trained based on pre-defined inference prompts and the target dataset, allowing it to more effectively learn pulsar features and guiding the model to output recognition results based on image features. Compared to the information fragmentation problem caused by image slicing methods in related technologies, this approach further enhances the comprehensiveness and accuracy of recognition.

[0083] The original pulsar candidate image contains multiple sub-images, but the morphological discrimination of different sub-images is different. Therefore, the image enhancement process can be performed on only a part, and the sub-image annotation can be performed on only those whose morphological discrimination meets the preset conditions.

[0084] In an exemplary embodiment, enhancing a pulsar candidate image to obtain a pulsar candidate image includes: identifying a sub-image frequency-phase map in the original pulsar candidate image; and performing data enhancement processing on the sub-image frequency-phase map to obtain the pulsar candidate image.

[0085] For example, the frequency-phase map of a sub-image in the original pulsar candidate image is identified; operations such as translation, flipping, rotation, and contrast transformation are performed on the frequency-phase map of the sub-image to obtain the pulsar candidate image. It is understandable that, with the pre-set multimodal large language model qwen2.5-vl as the base model, the base model's ability to recognize dispersion curves meets practical needs and can be fine-tuned based on a small amount of data. However, its ability to recognize frequency-phase maps cannot meet practical requirements because it is necessary to enhance the frequency-phase map of the sub-image in the original pulsar candidate image and expand the data volume.

[0086] The above method amplifies the real pulsar signal, and new training samples are obtained based on the enhanced original pulsar candidate image. In other words, data augmentation is performed on the frequency-phase map to better generalize to possible signal position variations in real observation data. This increases data diversity and improves the model's generalization ability.

[0087] In one exemplary embodiment, such as Figure 4 As shown, step 204 includes steps 402 to 408. Wherein:

[0088] In step 402, a sub-band frequency phase diagram and a sub-band dispersion curve diagram in the original pulsar candidate image are determined.

[0089] In step 404, a first morphological feature of the sub-band frequency phase diagram is identified, and first labeling information of the sub-band frequency phase diagram is determined according to a recognition result of the first morphological feature.

[0090] The first morphological feature of the sub-band frequency phase diagram can be an image feature of the sub-band frequency phase diagram, whether there is a target region of a preset shape, and a phase difference corresponding to an intersection of the target region and horizontal and vertical coordinates of the sub-band frequency phase diagram.

[0091] Optionally, the first morphological feature of the sub-band frequency phase diagram is identified, and the first labeling information of the sub-band frequency phase diagram is determined according to a recognition result of the first morphological feature, including:

[0092] The first morphological feature of the sub-band frequency phase diagram is identified, and if the recognition result of the first morphological feature includes a target region of a preset shape, a phase difference corresponding to an intersection of the target region and horizontal and vertical coordinates of the sub-band frequency phase diagram is determined. If the phase difference is within a preset range, the first labeling information of the sub-band frequency phase diagram is determined as a first preset label, and the first preset label is used to represent that the sub-band frequency phase diagram meets the characteristics of the real pulsar signal frequency phase diagram. If the phase difference is not within the preset range, the first labeling information is determined as a second preset label, and the second preset label is used to represent that the sub-band frequency phase diagram does not meet the characteristics of the real pulsar signal frequency phase diagram.

[0093] The frequency phase diagram shows the superimposed pulsar signal at different observation frequencies, and usually presents the relationship between frequency and phase in a two-dimensional form, and the color depth represents the size of the signal amplitude. A real pulsar usually emits a wideband signal, so it appears as a continuous bright vertical stripe in the sub-band diagram, that is, a black vertical line. Therefore, the target region of the preset shape can be a region representing a wideband signal, for example, it can be a black vertical line or a vertical strip structure.

[0094] The frequency phase diagram is divided into two categories, that is, the first preset label and the second preset label, by judging whether the frequency phase diagram meets the characteristics of the real pulsar signal frequency phase diagram.

[0095] For example, the first morphological feature of the sub-band frequency phase diagram is identified, if a black vertical line is identified in the sub-band frequency phase diagram, and the phase difference between the vertical line and the upper and lower intersection points of the horizontal coordinate phase axis is within a preset range, that is, not more than 0.4, the frequency phase diagram is classified as “yes”, that is, the first preset label. Otherwise, the frequency phase diagram is classified as “no” category, that is, the second preset label.

[0096] At step 406, a second morphological feature of the subgraph dispersion curve graph is identified, and second labeling information of the subgraph dispersion curve graph is determined according to the identification result of the second morphological feature and the pulsar candidate image.

[0097] The dispersion curve graph represents the integration of electron density from the earth to the pulsar, and is used to determine the best dispersion measurement. The true pulsar signal is from outer space, and its dispersion curve usually reaches a peak at a non-zero dispersion value. The rising trend on the left side of the peak and the falling trend on the right side may indicate that the electron density increases first and then decreases when the pulsar signal passes through the interstellar medium. This feature is an important basis for verifying the pulsar candidate. The second labeling information can be the matching degree of the subgraph dispersion curve graph and the true pulsar signal dispersion curve graph feature. The identification result of the second morphological feature can include the change trend.

[0098] Optionally, the second morphological feature of the subgraph dispersion curve graph is identified, and the second labeling information of the subgraph dispersion curve graph is determined according to the identification result of the second morphological feature and the pulsar candidate image, including: identifying the second morphological feature of the subgraph dispersion curve graph, if the identification result of the second morphological feature includes a preset change trend and the number of peaks in the preset change trend is a first preset value, and the target dispersion value in the pulsar candidate image is not a second preset value, then determining the second labeling information of the subgraph dispersion curve graph as a third preset label; the third preset label is used to represent that the true pulsar signal dispersion curve graph feature is satisfied; if the identification result of the second morphological feature does not include the preset change trend and / or the target dispersion value is the second preset value, then determining the second labeling information of the subgraph dispersion curve graph as a fourth preset label; the fourth preset label is used to represent that the true pulsar signal dispersion curve graph feature is not satisfied.

[0099] It can be understood that the monotone increasing may be that the electron density gradually increases when the signal propagates. The monotone decreasing may be that the electron density gradually decreases when the signal propagates, which is not consistent with the characteristics of the pulsar signal, because the pulsar signal usually does not monotonically change with frequency, and reaches a peak at a non-zero dispersion value.

[0100] The preset change trend can be that the dispersion curve graph has one relatively obvious peak, and has an upward trend on the left side of the peak and a downward trend on the right side. The identification result of the second morphological feature does not include the preset change trend can be that the overall is a monotone increasing trend, a monotone decreasing trend, or the dispersion curve fluctuates frequently, and there can be a case of having 0 or more relatively obvious peaks. The target dispersion value can be the best dispersion value of the pulsar candidate in the original pulsar candidate image. The first preset value can be but is not limited to 1, and the second preset value can be but is not limited to 0. In this embodiment, the first preset value is 1 and the second preset value is 0.

[0101] For example, if the dispersion curve graph has a relatively obvious peak, the left side of the peak has an upward trend, and the right side has a downward trend; and combined with whether the best dispersion value of the pulsar candidate in the image is non-“0”, if it is non-“0”, it is classified as “yes”, that is, the third preset label.

[0102] If the dispersion curve graph is, for example, monotonically increasing as a whole, monotonically decreasing, the dispersion curve frequently fluctuates with 0 or more relatively obvious peaks, etc., the best dispersion value of the pulsar candidate in the image is “0”. If any of the above conditions is met, it is classified as “no”, that is, the fourth preset label.

[0103] Step 408, constructing a target data set for identifying the type of pulsar signal according to the first labeling information, the second labeling information, and the corresponding pulsar candidate image.

[0104] In this embodiment, by finely labeling the subgraphs in the pulsar candidate image, a target data set is constructed, and the model is trained with the target data set, which can improve the recognition ability of the model for pulsars. Moreover, the image information and data features are effectively integrated, avoiding the information fragmentation problem caused by the traditional cutting method, and further improving the comprehensiveness and accuracy of the recognition.

[0105] In one exemplary embodiment, a method for constructing a positioning data set is provided, as shown in Figure 5 The method comprises the following steps:

[0106] Step 502, labeling the first position information of the subgraph in the original pulsar candidate image.

[0107] The first position information can be understood as a coordinate position. The subgraph includes a frequency-phase graph and a dispersion curve graph.

[0108] For example, the coordinate positions of the frequency-phase graph and the dispersion curve graph in the original pulsar candidate image are accurately labeled by artificial means to obtain the first position information corresponding to the frequency-phase graph and the dispersion curve graph respectively.

[0109] Step 504, performing scale transformation processing on the original pulsar candidate image to generate a plurality of size-adjusted images with different resolutions.

[0110] The scale transformation processing can be scaling the original pulsar candidate image according to a preset size ratio to obtain a plurality of size-adjusted images with different resolutions.

[0111] Step 506, determining the scale ratio between each size-adjusted image and the original pulsar candidate image, mapping the first position information according to the scale ratio to determine the second position information of the subgraph in each size-adjusted image.

[0112] At step 508, a positioning dataset is constructed according to the first position information and the second position information.

[0113] In the above embodiment, a positioning dataset containing images of diversified sizes and corresponding subgraph coordinates is constructed. This can enable the model to adapt to the identification of subgraph positions in images of different sizes, and enhance the robustness of the model to size changes.

[0114] In an exemplary embodiment, the preset inference prompt includes: if the morphological features corresponding to the first preset label and the third preset label exist in the pulsar candidate image, the pulsar candidate signal corresponding to the original pulsar candidate image is identified as a pulsar signal; if the morphological features corresponding to the first preset label and the third preset label do not exist at the same time in the pulsar candidate image, it is determined that the pulsar candidate signal is identified as radio frequency interference.

[0115] In an exemplary embodiment, based on the application of the above-mentioned pulsar identification model, a target pulsar candidate image corresponding to a to-be-identified pulsar candidate signal can be obtained; the to-be-identified pulsar candidate image is input into the pulsar identification model, and the signal type of the to-be-identified pulsar candidate signal is output.

[0116] For example, to obtain a target pulsar candidate image corresponding to a to-be-identified pulsar candidate signal, first, determine whether there is a black vertical line in the subgraph frequency-phase graph in the target pulsar candidate image, and whether the phase difference between the upper and lower intersection points of the vertical line and the horizontal coordinate phase axis is not more than 0.4; second, determine whether the subgraph dispersion curve graph has a relatively obvious wave peak, and whether the left side of the wave peak has an upward trend and the right side has a downward trend. Moreover, whether the best dispersion value of the target pulsar candidate image is non-zero, and whether the non-zero DM value is an important feature of a pulsar signal. If there is a black vertical line and the phase difference is not more than 0.4, the subgraph dispersion curve graph has a relatively obvious wave peak, and the left side of the wave peak has an upward trend and the right side has a downward trend, and whether the best dispersion value of the target pulsar candidate image is non-zero, the signal is identified as a pulsar signal; otherwise, it is identified as radio frequency interference.

[0117] For example, input the target pulsar candidate image shown in the following figure: Figure 6 The following identification result will be output:

[0118] <Thinking>1. Check the third subgraph, which is a frequency-phase graph. For a real pulsar signal, you should see a black vertical line or vertical band structure running through the frequency axis. I see a black vertical line or vertical band structure in the subgraph, which is a characteristic of a pulsar signal.

[0119] 2. Check the fourth subgraph, which is a dispersion curve graph. Since the pulsar is a non-terrestrial signal, we expect the signal to peak at a non-zero DM value (i.e., a non-zero dispersion value). I find a clear peak, and the best dispersion value is non-zero, which is a characteristic of a pulsar signal. <Think>

[0120] <Answer> The signal is a pulsar signal. <Answer>

[0121] It should be noted that Figure 6 the target pulsar candidate image shown includes a frequency-phase graph, a dispersion curve graph, and a best dispersion value, Figure 6 the types of graphs shown are the same as Figure 3 the types of graphs shown, and the meanings of the parameters in the graphs can be understood by referring to Figure 3 and related technologies in the art, which will not be described here.

[0122] Based on the above model, in the case of actual model identification, the pulsar candidate image to be identified can be directly input to the model to output the result, without the need for labeled segmentation enhancement. The model will give the reasoning process, including the characteristics of the subgraph, and the final result.

[0123] In an exemplary embodiment, as Figure 7 shown, an end-to-end pulsar candidate identification method is provided. Taking the terminal in Figure 1 as an example, the method includes the following steps 702 to 714. Among them:

[0124] Step 702, obtaining the original pulsar candidate image corresponding to the pulsar candidate signal, and performing enhancement processing on the original pulsar candidate image to obtain the pulsar candidate image.

[0125] Step 704, labeling the subgraph in the pulsar candidate image, and constructing a target data set for identifying the type of pulsar signal.

[0126] Step 706, by labeling the position information of the subgraph in the original pulsar candidate image, a positioning data set is constructed.

[0127] Step 708, training a preset pulsar identification model according to the positioning data set to obtain a positioning model for identifying the position information of the subgraph.

[0128] Step 710, obtaining a preset reasoning prompt, and training the positioning model according to the preset reasoning prompt and the target data set to obtain a pulsar identification model for identifying the type of pulsar signal.

[0129] Step 712, obtaining a target pulsar candidate image corresponding to a pulsar candidate signal to be identified.

[0130] At step 714, input the pulsar candidate image to be identified into the pulsar identification model, and output a signal type of the pulsar candidate image to be identified.

[0131] It should be noted that the specific implementation of the present embodiment can be realized by the above defined manner, which will not be described here.

[0132] In the above embodiments, by enhancing the original pulsar candidate image, balancing the data set, and finely labeling the subgraphs in the image, a high-quality training data set is constructed, so that the model can more effectively learn the characteristics of the pulsar, and the recognition accuracy and positioning accuracy are improved; in the training process, the reasoning mechanism simulating the thinking of scientists is introduced, so that the model can analyze the image features step by step, reason, and greatly improve the understanding ability of complex image patterns, thereby improving the accuracy and reliability of the identification result; the modeling ability of the multi-modal large language model with the Transformer architecture is used to effectively integrate image information and data features, avoid the information fragmentation problem caused by the traditional cutting method, and further improve the comprehensiveness and accuracy of the identification. In addition, the large model with the Transformer architecture has better learning ability, can maintain high recognition performance under various observation conditions and different distributed data, and has better generalization ability than traditional deep learning models, strong adaptability, and is suitable for actual pulsar search tasks.

[0133] It should be noted that the annotations of the frequency phase diagram and the dispersion curve diagram in the original pulsar candidate image and the corresponding pulsar candidate image obtained after enhancement processing can be the same, and the best dispersion value in the original pulsar candidate image and the corresponding pulsar candidate image obtained after enhancement processing is also consistent. In the above embodiments, the best dispersion value can be obtained from the original pulsar candidate image or from the corresponding pulsar candidate image obtained after enhancement processing. It should be understood that although each step in the flowchart involved in each of the above embodiments is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified in this document, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowchart involved in each of the above embodiments can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least part of other steps or steps or stages in other steps.

[0134] Based on the same inventive concept, the embodiments of the present application also provide an end-to-end pulsar candidate identification device for implementing the end-to-end pulsar candidate identification method described above. The implementation scheme of the device for solving the problem is similar to the implementation scheme described in the above method, so the specific limitations in one or more end-to-end pulsar candidate identification device embodiments provided below can refer to the limitations of the end-to-end pulsar candidate identification method described above, which will not be repeated here.

[0135] In one exemplary embodiment, as shown in Figure 8 An end-to-end pulsar candidate identification device is provided, comprising: an image processing module 802, a first data set construction module 804, a second data set construction module 806, a first training module 808 and a second training module 810, wherein:

[0136] The image processing module 802 is configured to obtain an original pulsar candidate image corresponding to a pulsar candidate signal, perform enhancement processing on the original pulsar candidate image, and obtain a pulsar candidate image.

[0137] The first data set construction module 804 is configured to label a sub-image in the pulsar candidate image, and construct a target data set for identifying a pulsar signal type.

[0138] The second data set construction module 806 is configured to label position information of the sub-image in the original pulsar candidate image, and construct a positioning data set.

[0139] The first training module 808 is configured to train a preset pulsar identification model according to the positioning data set, and obtain a positioning model for identifying the position information of the sub-image.

[0140] The second training module 810 is configured to obtain a preset inference prompt, train the positioning model according to the preset inference prompt and the target data set, and obtain a pulsar identification model for identifying the pulsar signal type.

[0141] The end-to-end pulsar candidate identification device balances the data set, increases the diversity of the data set, and finely labels the subgraphs in the image, thereby improving the recognition ability of the model for pulsars. The position information of the subgraphs in the original pulsar candidate image is labeled to obtain a positioning data set. The positioning data set is used to train a preset pulsar recognition model, so that the model can accurately identify and frame the region of the subgraph, thereby improving the positioning ability of the model for the subgraph. The positioning model is trained based on the preset inference prompt and the target data set, so that the model can more effectively learn the features of the pulsar and guide the model to output the recognition result based on the image features. Compared with the information fragmentation problem caused by the cutting method in the related art, this method further improves the comprehensiveness and accuracy of the recognition.

[0142] In an example embodiment, the image processing module 802 is configured to identify a subgraph frequency-phase graph in the original pulsar candidate image; and perform data enhancement processing on the subgraph frequency-phase graph to obtain a pulsar candidate image.

[0143] In an example embodiment, the first data set construction module 804 is configured to determine a subgraph frequency-phase graph and a subgraph dispersion curve graph in the original pulsar candidate image.

[0144] identify a first morphological feature of the subgraph frequency-phase graph, and determine first labeling information of the subgraph frequency-phase graph based on the recognition result of the first morphological feature;

[0145] identify a second morphological feature of the subgraph dispersion curve graph, and determine second labeling information of the subgraph dispersion curve graph based on the recognition result of the second morphological feature and the pulsar candidate image;

[0146] construct a target data set for identifying the type of pulsar signal based on the first labeling information, the second labeling information, and the corresponding pulsar candidate image.

[0147] In an example embodiment, the first data set construction module 804 is configured to identify a first morphological feature of the subgraph frequency-phase graph, and if a target region with a preset shape exists in the recognition result of the first morphological feature, determine a phase difference corresponding to the intersection of the horizontal and vertical coordinates of the target region and the subgraph frequency-phase graph.

[0148] If the phase difference is within a preset range, the first labeling information of the subgraph frequency-phase graph is determined as a first preset label, and the first preset label is used to represent that the subgraph frequency-phase graph satisfies the characteristics of the true pulsar signal frequency-phase graph.

[0149] If the phase difference is not within the preset range, the first annotation information is determined as a second preset label, and the second preset label is used to represent that the subgraph frequency phase graph does not satisfy the characteristics of the real pulsar signal frequency phase graph.

[0150] In an exemplary embodiment, the first data set construction module 804 is configured to identify a second morphological feature of the subgraph dispersion curve graph. If a preset change trend exists in the identification result of the second morphological feature, and the number of the preset change trend is a first preset value, and the target dispersion value in the original pulsar candidate image is not a second preset value, the second annotation information of the subgraph dispersion curve graph is determined as a third preset label. The third preset label is used to represent that the characteristics of the real pulsar signal dispersion curve graph are satisfied.

[0151] If the identification result of the second morphological feature does not exist the preset change trend, and / or the target dispersion value is the second preset value, the second annotation information of the subgraph dispersion curve graph is determined as a fourth preset label. The fourth preset label is used to represent that the characteristics of the real pulsar signal dispersion curve graph are not satisfied.

[0152] In an exemplary embodiment, the second data set construction module 806 is configured to annotate the first position information of the subgraph in the original pulsar candidate image.

[0153] The original pulsar candidate image is subjected to a scale transformation process to generate a plurality of size-adjusted images with different resolutions.

[0154] The scale ratio between each size-adjusted image and the original pulsar candidate image is determined, and the first position information is mapped according to the scale ratio to determine the second position information of the subgraph in each size-adjusted image.

[0155] The positioning data set is constructed according to the first position information and the second position information.

[0156] In an exemplary embodiment, the preset inference prompt includes:

[0157] If the morphological features corresponding to the first preset label and the third preset label exist in the pulsar candidate image, the pulsar candidate body signal corresponding to the original pulsar candidate image is identified as a pulsar signal.

[0158] If the morphological features corresponding to the first preset label and the third preset label do not exist at the same time in the pulsar candidate image, it is determined that the pulsar candidate body signal is identified as radio frequency interference.

[0159] In an exemplary embodiment, the end-to-end pulsar candidate body identification device described above further includes an identification module, which is configured to obtain a target pulsar candidate body image corresponding to a to-be-identified pulsar candidate body signal.

[0160] The pulsar candidate image to be identified is input into the pulsar identification model, and a signal type of the pulsar candidate to be identified is output.

[0161] The modules in the end-to-end pulsar candidate identification apparatus can be implemented by software, hardware, or a combination thereof. The modules can be embedded in or independent of a processor in a computer device in hardware form, or stored in a memory in the computer device in software form, so as to be invoked and executed by the processor to perform operations corresponding to the modules.

[0162] In an exemplary embodiment, a computer device, which can be a terminal, has an internal structure as shown in Figure 9 The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to perform wired or wireless communication with external terminals. The wireless communication can be achieved through WIFI, mobile cellular network, near field communication (NFC), or other technologies. The computer program is executed by the processor to implement an end-to-end pulsar candidate identification method. The display unit of the computer device is configured to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball, or touchpad arranged on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0163] Those skilled in the art can understand that Figure 9 The structure shown in the above description is only a block diagram of part of the structure related to the scheme of the present application, and does not limit the computer device to which the scheme of the present application is applied. Specifically, the computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0164] In an embodiment, a computer device is also provided, comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0165] In an embodiment, a computer readable storage medium is provided, storing a computer program, and the computer program implementing the steps in the above method embodiments when executed by a processor.

[0166] In an embodiment, a computer program product is provided, comprising a computer program, and the computer program implementing the steps in the above method embodiments when executed by a processor.

[0167] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0168] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., without being limited thereto.

[0169] The technical features of the above embodiments can be combined in any manner. To make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present application.

[0170] The above-described embodiments are merely illustrative of several embodiments of the present application, which are described in more detail and in a specific manner, but should not be construed as limiting the scope of the patent of the present application. It should be noted that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. An end-to-end pulsar candidate identification method, characterized by, The method comprises: obtaining an original pulsar candidate image corresponding to a pulsar candidate signal, performing enhancement processing on a subgraph frequency phase graph in the original pulsar candidate image to obtain a pulsar candidate image; annotating the subgraph in the pulsar candidate image to construct a target data set for identifying the type of pulsar signal; constructing a positioning data set by annotating the position information of the subgraph in the original pulsar candidate image; training a preset pulsar identification model according to the positioning data set to obtain a positioning model for identifying the position information of the subgraph; obtaining a preset inference prompt, training the positioning model according to the preset inference prompt and the target data set to obtain a pulsar identification model for identifying the type of pulsar signal; the preset inference prompt is used to guide the pulsar identification model to perform inference based on the morphological features of the subgraph and output an identification result; wherein, annotating the subgraph in the pulsar candidate image to construct a target data set for identifying the type of pulsar signal comprises: determining the subgraph frequency phase graph and the subgraph dispersion curve graph in the original pulsar candidate image; identifying a first morphological feature of the subgraph frequency phase graph, and determining first annotation information of the subgraph frequency phase graph according to the identification result of the first morphological feature; the first annotation information comprises a first preset label and a second preset label, the first preset label is used to represent that the subgraph frequency phase graph meets the characteristics of a real pulsar signal frequency phase graph; the second preset label is used to represent that the subgraph frequency phase graph does not meet the characteristics of a real pulsar signal frequency phase graph; identifying a second morphological feature of the subgraph dispersion curve graph, and determining second annotation information of the subgraph dispersion curve graph according to the identification result of the second morphological feature and the pulsar candidate image; the second annotation information comprises a third preset label and a fourth preset label, the third preset label is used to represent that the subgraph dispersion curve graph meets the characteristics of a real pulsar signal dispersion curve graph; the fourth preset label is used to represent that the subgraph dispersion curve graph does not meet the characteristics of a real pulsar signal dispersion curve graph; constructing a target data set for identifying the type of pulsar signal according to the first annotation information, the second annotation information and the corresponding pulsar candidate image.

2. The method of claim 1, wherein, performing enhancement processing on the subgraph frequency phase graph in the pulsar candidate image to obtain a pulsar candidate image, comprising: identifying the subgraph frequency phase graph in the original pulsar candidate image; performing data enhancement processing on the subgraph frequency phase graph to obtain a pulsar candidate image.

3. The method of claim 1, wherein, The identification of the first morphological feature of the subgraph frequency phase graph and the determination of the first annotation information of the subgraph frequency phase graph according to the identification result of the first morphological feature comprises: identifying the first morphological feature of the subgraph frequency phase graph, and if there is a target region of a preset shape in the identification result of the first morphological feature, determining a phase difference corresponding to the intersection of the target region and the horizontal and vertical coordinates of the subgraph frequency phase graph; if the phase difference is within a preset range, the first annotation information of the subgraph frequency phase graph is determined as the first preset label; If the phase difference is not within the preset range, the first annotation information is determined as a second preset label.

4. The method of claim 3, wherein, The second morphological feature of the subgraph dispersion curve graph is identified, and second annotation information of the subgraph dispersion curve graph is determined according to an identification result of the second morphological feature and the pulsar candidate image. The second morphological feature of the subgraph dispersion curve graph is identified, and second annotation information of the subgraph dispersion curve graph is determined according to an identification result of the second morphological feature and the pulsar candidate image. If the identification result of the second morphological feature does not include the preset change trend, and / or the target dispersion value is the second preset value, the second annotation information of the subgraph dispersion curve graph is determined as a fourth preset label.

5. The method of claim 1, wherein, The first position information of the subgraph in the original pulsar candidate image is annotated. The first position information of the subgraph in the original pulsar candidate image is annotated. The original pulsar candidate image is subjected to a scale transformation process to generate a plurality of size-adjusted images with different resolutions. The scale ratio between each size-adjusted image and the original pulsar candidate image is determined, the first position information is mapped according to the scale ratio, and the second position information of the subgraph in each size-adjusted image is determined. The first position information and the second position information are used to construct a positioning data set.

6. The method of claim 5, wherein, The preset inference prompt includes: If the morphological features corresponding to the first preset label and the third preset label exist in the pulsar candidate image, the pulsar candidate signal corresponding to the original pulsar candidate image is identified as a pulsar signal. If the morphological features corresponding to the first preset label and the third preset label do not exist at the same time in the pulsar candidate image, it is determined that the pulsar candidate signal is identified as radio frequency interference.

7. The method according to any one of claims 1 to 6, characterized in that, The method further includes: Obtaining a target pulsar candidate image corresponding to a pulsar candidate signal to be identified; Inputting the pulsar candidate image to be identified into the pulsar identification model to output the signal type of the pulsar candidate signal to be identified.

8. An end-to-end pulsar candidate identification apparatus, characterized by, The device includes: An image processing module is configured to obtain an original pulsar candidate image corresponding to a pulsar candidate signal, and to perform enhancement processing on a subgraph frequency phase graph in the original pulsar candidate image to obtain a pulsar candidate image; A first data set construction module is configured to determine a subgraph frequency phase graph and a subgraph dispersion curve graph in the original pulsar candidate image; identify a first morphological feature of the subgraph frequency phase diagram, and determine first labeling information of the subgraph frequency phase diagram according to a recognition result of the first morphological feature; the first labeling information includes a first preset label and a second preset label, the first preset label is used to represent that the subgraph frequency phase diagram meets a characteristic of a real pulsar signal frequency phase diagram, and the second preset label is used to represent that the subgraph frequency phase diagram does not meet the characteristic of the real pulsar signal frequency phase diagram; identify a second morphological feature of the subgraph dispersion curve diagram, and determine second labeling information of the subgraph dispersion curve diagram according to a recognition result of the second morphological feature and the candidate pulsar image; the second labeling information includes a third preset label and a fourth preset label, the third preset label is used to represent that the subgraph dispersion curve diagram meets a characteristic of a real pulsar signal dispersion curve diagram, and the fourth preset label is used to represent that the subgraph dispersion curve diagram does not meet the characteristic of the real pulsar signal dispersion curve diagram; construct a target data set for identifying a pulsar signal type according to the first labeling information, the second labeling information and the corresponding candidate pulsar image; a second data set construction module is configured to construct a positioning data set by labeling position information of a subgraph in the original candidate pulsar image; a first training module is configured to train a preset pulsar identification model according to the positioning data set, to obtain a positioning model for identifying subgraph position information; a second training module is configured to obtain a preset inference prompt, and train the positioning model according to the preset inference prompt and the target data set, to obtain a pulsar identification model for identifying a pulsar signal type; the preset inference prompt is used to guide the pulsar identification model to perform inference based on morphological features of the subgraph, and output an identification result. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor executes the computer program to implement the steps of the method in any one of claims 1 to 6.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 7.

11. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 7. The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 7.

Images