A Weather Monitoring Method and System Based on Image Recognition

Abstract

This invention discloses a weather monitoring method and system based on image recognition, comprising: acquiring weather-related images and preprocessing them; extracting multiple image features from the preprocessed weather images, analyzing and evaluating them to obtain evaluation results; selecting image features closely related to weather as weather features from the multiple image features based on the evaluation results; analyzing the weather features, determining their weather categories, and labeling them with weather category tags; constructing and training an initial weather recognition model based on the weather features and weather category tags to finally obtain the weather recognition model; and performing weather recognition on monitored real-time weather images using the weather recognition model, and completing weather monitoring based on the recognition results. This invention constructs a weather recognition model to accurately identify weather conditions, thereby completing weather monitoring and helping users understand weather conditions in real time and make corresponding decisions and actions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a weather monitoring method and system based on image recognition. BACKGROUND

[0002] Weather monitoring is of great significance in meteorology, environmental protection, agriculture, aerospace and other fields. Weather monitoring method refers to the method of understanding and predicting weather phenomena by collecting, analyzing and interpreting weather data. With the rapid development of computer vision and deep learning, image recognition-based weather monitoring has become a popular method. It is a method of analyzing and processing weather image data using computer vision and deep learning technology, which has shown great potential in weather monitoring and prediction.

[0003] Traditional weather monitoring methods mainly rely on sensor networks to measure physical quantities to complete weather monitoring. However, this method uses a limited number of sensors arranged in specific locations, limiting the monitoring range, resulting in incomplete or inaccurate weather monitoring in some areas or remote areas. In addition, sensors in the sensor network are usually arranged at fixed intervals, resulting in uneven density and distribution of data, leading to inaccurate and incomplete weather monitoring data in some areas or locations. Image recognition-based weather monitoring can overcome the above difficulties to achieve accurate weather monitoring. SUMMARY

[0004] To solve the above technical problems, the present application provides a weather monitoring method and system based on image recognition, comprising:

[0005] Obtain weather images related to weather and preprocess the weather images;

[0006] Obtain a plurality of image features from the preprocessed weather images, analyze and evaluate the image features, and obtain evaluation results;

[0007] According to the evaluation results, screen out image features with close relationship with weather from the plurality of image features as weather features;

[0008] Analyze the weather features, determine the weather category of the weather features, and label the weather features with weather category labels;

[0009] Based on the weather features and weather category labels, an initial weather recognition model is constructed and trained to obtain a weather recognition model;

[0010] The weather recognition model is used to identify the real-time weather images monitored, and the weather is monitored according to the identification results.

[0011] Further, the weather-related weather image is obtained, and the weather image is preprocessed, comprising:

[0012] The weather-related weather image is obtained from the weather image database, and the weather image is subjected to Fourier transform to obtain a weather spectrum image;

[0013] The weather spectrum image is subjected to noise removal, smoothing and normalization processing;

[0014] The weather spectrum image after processing is subjected to clustering processing by a density clustering algorithm, a plurality of clustering centers are obtained, and the image features corresponding to each of the clustering centers are analyzed.

[0015] Further, the plurality of image features are obtained from the preprocessed weather image, and the image features are analyzed and evaluated to obtain an evaluation result, comprising:

[0016] The image features corresponding to each of the clustering centers are obtained from the preprocessed weather image, and the image features are analyzed to determine a plurality of evaluation indexes of the image features related to the weather;

[0017] The plurality of evaluation indexes of each of the image features are respectively evaluated and scored, and the scoring results are comprehensively calculated to obtain the evaluation value of the image features.

[0018] Further, the plurality of evaluation indexes of each of the image features are respectively evaluated and scored, and the scoring results are comprehensively calculated to obtain the evaluation value of the image features, comprising:

[0019] The plurality of evaluation indexes of each of the image features are respectively evaluated and scored to obtain the score of the evaluation index, and the score sequence A of the evaluation index of each of the image features is established, A=(a1, a2…a i…an), wherein n is the number of evaluation indexes, and a i is the score of the i th evaluation index.

[0020] The importance of each evaluation index is analyzed, and the weight coefficient of each evaluation index is determined according to the importance to establish the weight coefficient sequence B of the evaluation index of each of the image features, B=(b1, b2…b i…bn), wherein b i is the weight coefficient of the i th evaluation index.

[0021] Based on the score and the weight coefficient of each evaluation index, the evaluation value s of the image features is obtained by weighted calculation,

[0022] Further, the image features closely related to the weather are screened out from the plurality of image features according to the evaluation results as weather features, comprising:

[0023] An evaluation limit value is set based on the evaluation values of all the image features, and the evaluation limit value is used as a limit condition for determining whether the image features are image features closely related to the weather;

[0024] The image features with evaluation values higher than the evaluation limit value are screened out as weather features.

[0025] Further, the weather features are analyzed, the weather categories of the weather features are determined, and the weather features are labeled with weather category labels, comprising:

[0026] The correlation of the weather features with each weather category is analyzed, and the correlation coefficients of the weather features with each weather category are calculated respectively;

[0027] The weather features with the highest correlation coefficients with a certain weather category are classified into this weather category;

[0028] Each of the weather features is labeled with a corresponding weather category label.

[0029] Further, the initial weather recognition model is constructed based on the weather features and weather category labels, and the initial weather recognition model is trained to obtain a weather recognition model, comprising:

[0030] The weather features are used as input data of the model, and the weather category labels are used as output data of the model;

[0031] The initial weather recognition model is constructed based on the weather features, weather category labels and a preset neural network model;

[0032] A plurality of sets of the model input data and the model output data are obtained as training data of the model, the initial weather recognition model is trained until the initial weather recognition model meets a preset convergence condition, and a weather recognition model is obtained.

[0033] The application also provides a weather monitoring system based on image recognition, comprising:

[0034] An acquisition module is configured to acquire weather images related to the weather and pre-process the weather images;

[0035] An evaluation module is configured to acquire a plurality of image features from the pre-processed weather images, analyze and evaluate the image features, and obtain evaluation results;

[0036] a screening module configured to screen, according to the evaluation result, an image feature having a close correlation with weather from a plurality of the image features as a weather feature;

[0037] a judging module configured to analyze the weather feature, judge a weather category of the weather feature, and label the weather feature with a weather category label;

[0038] a modeling module configured to construct an initial weather recognition model based on the weather feature and the weather category label, and train the initial weather recognition model to obtain a weather recognition model;

[0039] a monitoring module configured to perform weather recognition on a monitored real-time weather image by using the weather recognition model, and complete the monitoring of weather according to the recognition result.

[0040] Compared with the prior art, the weather monitoring method and system based on image recognition according to the embodiments of the present application have the following beneficial effects:

[0041] The present application constructs a weather recognition model to accurately recognize weather conditions, thereby completing the monitoring of weather, helping users to know the weather conditions in real time and make corresponding decisions and actions. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 is a flow structure diagram of the weather monitoring method based on image recognition in the embodiments of the present application;

[0043] Figure 2 is a composition diagram of the weather monitoring system based on image recognition in the embodiments of the present application. DETAILED DESCRIPTION

[0044] The specific embodiments of the present application will be further described in detail below in combination with the drawings and embodiments. The following embodiments are used to illustrate the present application, but not to limit the scope of the present application.

[0045] In the description of the present application, it should be understood that the terms "center", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the present application and simplify the description, and therefore cannot be understood as indicating or implying that the platform or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0046] The terms "first", "second", "third", etc. are used only for descriptive purposes and do not imply or suggest a relative importance or an implicit indication of the number of the technical features indicated. Thus, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0047] In the description of the present application, it should be noted that unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integrally connected; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, or the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0048] As shown in Figure 1 In the embodiments of the present application, a weather monitoring method based on image recognition is provided, which comprises: S100, acquiring a weather image related to weather, and preprocessing the weather image; S200, acquiring a plurality of image features from the preprocessed weather image, and analyzing and evaluating the image features to obtain an evaluation result; S300, screening an image feature having close correlation with weather from the plurality of image features as a weather feature according to the evaluation result; S400, analyzing the weather feature, judging the weather category of the weather feature, and labeling the weather feature with a weather category label; S500, constructing an initial weather recognition model based on the weather feature and the weather category label, and training the initial weather recognition model to obtain a weather recognition model; S600, performing weather recognition on a monitored real-time weather image through the weather recognition model, and completing the monitoring of weather according to the recognition result.

[0049] Further, the present application constructs a weather recognition model to accurately recognize weather conditions, so as to complete the monitoring of weather, help users to understand weather conditions in real time, and make corresponding decisions and actions.

[0050] In the embodiments of the present application, a weather monitoring method based on image recognition is provided, which comprises: acquiring a weather image related to weather, and preprocessing the weather image, comprising: acquiring a weather image related to weather from a weather image database, and performing Fourier transform on the weather image to obtain a weather spectrum image; performing noise removal, smoothing and normalization processing on the weather spectrum image; performing clustering processing on the processed weather spectrum image again through a density clustering algorithm to obtain a plurality of clustering centers, and analyzing the image features corresponding to each of the clustering centers.

[0051] Specifically, weather-related image data is selected from a pre-prepared weather image database, which can be image data from meteorological satellites, weather cameras, and other devices; Fourier transform is applied to the selected weather images to convert the images from the time domain to the frequency domain, which can represent the images as spectral images containing information of different frequencies; noise removal, smoothing, and normalization are performed on the processed weather spectral images, which can improve the quality of the images and reduce noise and interference; a density clustering algorithm such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is used to cluster the processed weather spectral images, which can group similar image features together to form multiple cluster centers; for each cluster center, the corresponding image features are analyzed, and by analyzing the features represented by the cluster centers, image features related to different weather types can be obtained.

[0052] In an embodiment of the present application, an image recognition-based weather monitoring method is provided, which obtains multiple image features from pre-processed weather images and analyzes and evaluates the image features to obtain evaluation results, including: obtaining the image features corresponding to each cluster center from the pre-processed weather images, and analyzing the image features to determine multiple evaluation indicators of the image features that are related to the weather; evaluating each of the multiple evaluation indicators of the image features and comprehensively calculating the evaluation scores to obtain the evaluation value of the image features.

[0053] Specifically, from the pre-processed weather images, the image features corresponding to each cluster center are extracted; for each image feature, multiple evaluation indicators related to the weather are determined according to professional knowledge and experience in the field of weather, which are used to measure the degree of association between the image features and different weather types; experts in the field of weather are invited to evaluate and score the multiple evaluation indicators of each image feature, and the experts can evaluate each indicator according to their experience and professional knowledge and give the corresponding score, which reflects the degree of correlation between the image features and the weather; for each image feature, the evaluation scores of the experts in the field of weather are comprehensively calculated to obtain the evaluation value of the image features, which reflects the degree of association between the image features and the weather.

[0054] In the embodiments of the present application, a weather monitoring method based on image recognition is provided, and the evaluation indexes of each image feature are respectively evaluated and scored, and the scoring results are comprehensively calculated to obtain the evaluation value of the image feature, including: the evaluation indexes of each image feature are respectively evaluated and scored to obtain the scores of the evaluation indexes, and the score sequence A of the evaluation indexes of each image feature is established, A=(a1, a2…a i…an), wherein n is the number of evaluation indexes, and a i is the score of the i th evaluation index; the importance of each evaluation index is analyzed, and the weight coefficient of each evaluation index is determined according to the importance, and the weight coefficient sequence B of the evaluation indexes of each image feature is established, B=(b1, b2…b i…bn), wherein b i is the weight coefficient of the i th evaluation index; based on the score and the weight coefficient of each evaluation index, the evaluation value s of the image feature is obtained through weighted calculation,

[0055] In the embodiments of the present application, a weather monitoring method based on image recognition is provided, and the evaluation indexes of each image feature are respectively evaluated and scored, and the scoring results are comprehensively calculated to obtain the evaluation value of the image feature, including: the evaluation indexes of each image feature are respectively evaluated and scored to obtain the scores of the evaluation indexes, and the score sequence A of the evaluation indexes of each image feature is established, A=(a1, a2…a i…an), wherein n is the number of evaluation indexes, and a i is the score of the i th evaluation index; the importance of each evaluation index is analyzed, and the weight coefficient of each evaluation index is determined according to the importance, and the weight coefficient sequence B of the evaluation indexes of each image feature is established, B=(b1, b2…b i…bn), wherein b i is the weight coefficient of the i th evaluation index; based on the score and the weight coefficient of each evaluation index, the evaluation value s of the image feature is obtained through weighted calculation,

[0056] Specifically, by comprehensively analyzing the evaluation values of all image features, an evaluation limit value is finally set, which will be used as a limit condition for judging whether the image feature is closely related to the weather; for the evaluation value of each image feature, the image features with evaluation values higher than the evaluation limit value are screened out, and these image features are considered to have close correlation with the weather; the screened image features are taken as weather features.

[0057] In the embodiments of the present application, a weather monitoring method based on image recognition is provided, and the evaluation indexes of each image feature are respectively evaluated and scored, and the scoring results are comprehensively calculated to obtain the evaluation value of the image feature, including: the evaluation indexes of each image feature are respectively evaluated and scored to obtain the scores of the evaluation indexes, and the score sequence A of the evaluation indexes of each image feature is established, A=(a1, a2…a i…an), wherein n is the number of evaluation indexes, and a i is the score of the i th evaluation index; the importance of each evaluation index is analyzed, and the weight coefficient of each evaluation index is determined according to the importance, and the weight coefficient sequence B of the evaluation indexes of each image feature is established, B=(b1, b2…b i…bn), wherein b i is the weight coefficient of the i th evaluation index; based on the score and the weight coefficient of each evaluation index, the evaluation value s of the image feature is obtained through weighted calculation,

[0058] Specifically, a dataset containing weather categories is obtained, and for each weather feature, the correlation coefficient with each weather category is calculated respectively, which can include Pearson correlation coefficient and Spearman correlation coefficient, where the correlation coefficient measures the strength of linear or nonlinear relationship between the weather feature and each weather category; for each weather category, determine the weather feature with the highest correlation coefficient with the category, and classify the weather feature into the corresponding weather category; for each weather feature, label it with the corresponding weather category, and in subsequent applications, different weather types can be identified and classified according to the label of the weather feature.

[0059] In an embodiment of the present application, a weather monitoring method based on image recognition is provided, which constructs an initial weather recognition model based on the weather features and weather category labels, and trains the initial weather recognition model to obtain a weather recognition model, including: taking the weather features as input data of the model, and taking the weather category labels as output data of the model; constructing an initial weather recognition model based on the weather features and weather category labels and a preset neural network model; obtaining multiple groups of model input data and model output data as training data of the model, training the initial weather recognition model until the initial weather recognition model meets a preset convergence condition, and obtaining a weather recognition model.

[0060] Specifically, the weather features are taken as input data, the weather category labels are taken as output data, and a training data set is prepared, which should contain multiple groups of weather images and their corresponding weather category labels; according to a preset neural network model structure, an initial weather recognition model is established, where the preset neural network model structure can use a convolutional neural network (CNN) or other suitable model structure for image classification; the prepared data set is divided into a training data set and a validation data set, the training data set is used for the training process of the model, and the validation data set is used to evaluate the performance of the model; the initial weather recognition model is trained using the training data set, in the training process, the input data (weather features) are provided to the model and compared with the output data (weather category labels), and the weights and biases of the model are optimized through a backpropagation algorithm to enable the model to better predict the weather category; a preset convergence condition is set, such as the number of training rounds, the change of loss function, etc., when the model meets the preset convergence condition, the training process stops, and finally a weather recognition model is obtained.

[0061] As Figure 2As shown, in the embodiments of the present application, a weather monitoring system based on image recognition is provided, comprising: an acquisition module, configured to acquire weather images related to weather, and pre-process the weather images; an evaluation module, configured to acquire a plurality of image features from the pre-processed weather images, and analyze and evaluate the image features to obtain evaluation results; a screening module, configured to screen out image features having close correlation with weather from the plurality of image features as weather features according to the evaluation results; a judgment module, configured to analyze the weather features, judge weather categories of the weather features, and label the weather features with weather category labels; a modeling module, configured to construct an initial weather recognition model based on the weather features and weather category labels, and train the initial weather recognition model to obtain a weather recognition model; and a monitoring module, configured to perform weather recognition on monitored real-time weather images through the weather recognition model, and complete weather monitoring according to the recognition results.

[0062] In summary, the embodiments of the present application provide a weather monitoring method and system based on image recognition, which comprises: acquiring weather images related to weather, and pre-processing the weather images; acquiring a plurality of image features from the pre-processed weather images, and analyzing and evaluating the image features to obtain evaluation results; screening out image features having close correlation with weather from the plurality of image features as weather features according to the evaluation results; analyzing the weather features, judging weather categories of the weather features, and labeling the weather features with weather category labels; constructing and training an initial weather recognition model based on the weather features and weather category labels, and finally obtaining a weather recognition model; performing weather recognition on monitored real-time weather images through the weather recognition model, and completing weather monitoring according to the recognition results. The present application constructs a weather recognition model, which is used to accurately recognize weather conditions, so as to complete weather monitoring, help users to understand weather conditions in real time, and make corresponding decisions and actions.

[0063] Finally, it should be noted that: obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

[0064] The above description is only one embodiment of the present application, but cannot limit the scope of the present application, and any structural changes made according to the present application should be considered as falling within the scope of the present application.

[0065] The term "comprising" or any other similar term is intended to encompass the inclusion of one or more stated elements or steps but not preclude the inclusion of additional elements or steps. The term "comprising" is intended to mean that the process, platform, article, or apparatus that "comprises" one or more elements can also comprise other elements not expressly listed or inherent to such process, platform, article, or apparatus.

[0066] The technical solutions of the present application have been described in combination with the further embodiments shown in the drawings, but it is easy for those skilled in the art to understand that the protection scope of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to the related technical features without departing from the principles of the present application, and the technical solutions after the changes or replacements will all fall within the protection scope of the present application.

[0067] The above description is merely preferred embodiments of the present application, but not intended to limit the protection scope of the present application.

Claims

1. An image recognition-based weather monitoring method, characterized by, The method comprises the following steps: obtaining weather images related to weather and preprocessing the weather images; obtaining a plurality of image features from the preprocessed weather images, and analyzing and evaluating the image features to obtain evaluation results; screening out image features with close correlation with weather from the plurality of image features as weather features according to the evaluation results; analyzing the weather features, judging the weather categories of the weather features, and labeling the weather features with weather category labels; constructing an initial weather recognition model based on the weather features and the weather category labels, and training the initial weather recognition model to obtain a weather recognition model; recognizing weather through the weather recognition model based on the real-time weather images monitored, and completing the monitoring of the weather according to the recognition results; the step of obtaining weather images related to weather and preprocessing the weather images comprises the following steps: obtaining weather images related to weather from a weather image database, and performing Fourier transform on the weather images to obtain weather spectrum images; performing noise removal, smoothing and normalization processing on the weather spectrum images; performing clustering processing on the processed weather spectrum images again through a density clustering algorithm to obtain a plurality of clustering centers, and analyzing the image features corresponding to each of the clustering centers; the step of analyzing the weather features, judging the weather categories of the weather features, and labeling the weather features with weather category labels comprises the following steps: analyzing the correlation between the weather features and each weather category, and calculating the correlation coefficients between the weather features and each weather category respectively; grouping the weather features with the highest correlation coefficient with a certain weather category into this weather category; labeling each of the weather features with a corresponding weather category label. 2.The weather monitoring method based on image recognition of claim 1, wherein, the step of obtaining a plurality of image features from the preprocessed weather images, and analyzing and evaluating the image features to obtain evaluation results comprises the following steps: obtaining the image features corresponding to each of the clustering centers from the preprocessed weather images, and analyzing the image features to determine a plurality of evaluation indexes of the image features related to weather; evaluating and scoring the plurality of evaluation indexes of each of the image features respectively, and comprehensively calculating the scoring results to obtain the evaluation value of the image features. 3.The weather monitoring method based on image recognition of claim 2, wherein, the step of evaluating and scoring the plurality of evaluation indexes of each of the image features respectively, and comprehensively calculating the scoring results to obtain the evaluation value of the image features comprises the following steps: evaluating and scoring the plurality of evaluation indexes of each of the image features respectively to obtain the scores of the evaluation indexes, and establishing an evaluation index score sequence A of each of the image features, A=(a1, a2…ai…an), wherein n is the number of evaluation indexes, and ai is the score of the ith evaluation index; analyzing the importance of each evaluation index, and determining the weight coefficient of each evaluation index according to the importance to establish an evaluation index weight coefficient sequence B of each of the image features, B=(b1, b2…bi…bn), wherein bi is the weight coefficient of the ith evaluation index; Based on the score of each evaluation index and the weight coefficient, the evaluation value s of the image feature is obtained by weighted calculation, 4.The weather monitoring method based on image recognition of claim 2, wherein, The method comprises the following steps: An evaluation boundary value is set based on evaluation values of all the image features, and the evaluation boundary value is used as a boundary condition for determining whether the image features are weather-related image features; The image features with evaluation values higher than the evaluation boundary value are selected as weather features. 5.The weather monitoring method based on image recognition of claim 1, wherein, The method comprises the following steps: The weather features are used as input data of the model, and the weather category labels are used as output data of the model; An initial weather recognition model is constructed based on the weather features, the weather category labels, and a preset neural network model; A plurality of sets of the model input data and the model output data are obtained as training data of the model, and the initial weather recognition model is trained until the initial weather recognition model meets a preset convergence condition, thereby obtaining a weather recognition model.

6. An image recognition based weather monitoring system characterized in that, The method comprises the following steps: An acquisition module is configured to acquire weather images related to weather and pre-process the weather images; An evaluation module is configured to acquire a plurality of image features from the pre-processed weather images, analyze and evaluate the image features, and obtain evaluation results; A selection module is configured to select weather-related image features from the plurality of image features as weather features based on the evaluation results; A judgment module is configured to analyze the weather features, determine weather categories of the weather features, and label the weather features with weather category labels; A modeling module is configured to construct an initial weather recognition model based on the weather features and the weather category labels, and train the initial weather recognition model to obtain a weather recognition model; A monitoring module is configured to perform weather recognition on monitored real-time weather images by using the weather recognition model, and complete weather monitoring based on the recognition results. The method comprises the following steps: Weather images related to weather are acquired from a weather image database, and Fourier transform is performed on the weather images to obtain weather spectrum images; The weather spectrum images are processed by removing noise, smoothing, and normalization; The processed weather spectrum images are further clustered by using a density clustering algorithm to obtain a plurality of cluster centers, and image features corresponding to each cluster center are analyzed; The method comprises the following steps: The correlation of the weather features with each weather category is analyzed, and a correlation coefficient of the weather features with each weather category is calculated; The weather features with the highest correlation coefficient with a certain weather category are classified into this weather category; Each weather feature is labeled with a corresponding weather category label.

Images