Atmospheric Pollution Simulation and Evaluation Method, Device and Electronic Equipment Based on Image Processing

Abstract

The present invention provides a method, apparatus and electronic device for simulating and evaluating air pollution based on image processing. The method includes: obtaining target forecast data and obtaining target observation data; generating a target forecast image based on the target forecast data and generating a target observation image based on the target observation data, where the image sizes of the target forecast image and the target observation image are the same; extracting features from the target forecast image to obtain a forecast feature extraction result of the target forecast image, and extracting features from the target observation image to obtain an observation feature extraction result of the target observation image; determining target similarity judgment information between the target forecast data and the target observation data based on the forecast feature extraction result and the observation feature extraction result. The embodiments of the present invention can effectively improve the accuracy of air pollution simulation and evaluation.

Description

Technical Field

[0001] The present invention relates to the technical field of air quality, and in particular, to a method, device, and electronic device for simulating and evaluating air pollution based on image processing. Background Art

[0002] The evaluation and verification of numerical weather prediction are important components of the application of numerical models. However, related technologies usually achieve the simulation and evaluation of air pollution through objective statistical tests with observed data as the true values, such as absolute measurement methods based on root mean square error, etc. That is, the simulation and evaluation of air pollution are usually achieved through point-to-point tests (such as the point-to-point comparison between station data and forecast data), making it difficult to evaluate the spatial distribution characteristics of pollution objects, resulting in relatively low accuracy of the simulation and evaluation results of air pollution. Based on this, there is currently no good solution to how to improve the accuracy of the simulation and evaluation of air pollution. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a method, device, and electronic device for simulating and evaluating air pollution based on image processing to solve problems such as relatively low accuracy of the simulation and evaluation of air pollution caused by related technologies. That is, the embodiments of the present invention can perform the simulation and evaluation of air pollution through a target forecast image and a target observation image, that is, can evaluate the spatial distribution characteristics of a target pollution object through the forecast feature extraction result and the observation feature extraction result, and then determine a target air pollution simulation and evaluation result with relatively high accuracy through the forecast feature extraction result and the observation feature extraction result, effectively improving the accuracy of the simulation and evaluation of air pollution.

[0004] According to one aspect of the embodiments of the present invention, a method for simulating and evaluating air pollution based on image processing is provided, and the method includes:

[0005] Obtain target forecast data and obtain target observation data, where the target forecast data includes forecast information of a target pollution object in each grid area of a target area, the target forecast data is forecast by a target numerical model, and the target observation data includes observation information of the target pollution object in each grid area;

[0006] Generate a target forecast image based on the target forecast data and generate a target observation image based on the target observation data, where the target forecast image and the target observation image have the same image size;

[0007] Extract features from the target forecast image to obtain a forecast feature extraction result of the target forecast image, and extract features from the target observation image to obtain an observation feature extraction result of the target observation image;

[0008] Based on the predicted feature extraction result and the observed feature extraction result, determine the target similarity judgment information between the target prediction data and the target observation data;

[0009] Based on the target similarity judgment information, determine the target air pollution simulation evaluation result, and the target air pollution simulation evaluation result supports indicating the simulation effect of the target numerical model for the target pollution object.

[0010] According to another aspect of the embodiments of the present invention, there is provided an air pollution simulation evaluation device based on image processing, and the device includes:

[0011] An acquisition unit, configured to acquire target prediction data and acquire target observation data, where the target prediction data includes prediction information of a target pollution object in each grid area in a target area, the target prediction data is predicted by a target numerical model, and the target observation data includes observation information of the target pollution object in each grid area;

[0012] A processing unit, configured to generate a target prediction image based on the target prediction data and generate a target observation image based on the target observation data, and the image sizes of the target prediction image and the target observation image are the same;

[0013] The processing unit is further configured to perform feature extraction on the target prediction image to obtain a prediction feature extraction result of the target prediction image, and perform feature extraction on the target observation image to obtain an observation feature extraction result of the target observation image;

[0014] The processing unit is further configured to determine the target similarity judgment information between the target prediction data and the target observation data based on the prediction feature extraction result and the observation feature extraction result;

[0015] The processing unit is further configured to determine a target air pollution simulation evaluation result based on the target similarity judgment information, and the target air pollution simulation evaluation result supports indicating the simulation effect of the target numerical model for the target pollution object.

[0016] According to another aspect of the embodiments of the present invention, there is provided an electronic device, which includes a processor and a memory storing a program, where the program includes instructions, and when the instructions are executed by the processor, the processor executes the method mentioned above.

[0017] According to another aspect of the embodiments of the present invention, there is provided a non-transitory computer-readable storage medium storing computer instructions, and the computer instructions are used to cause a computer to execute the method mentioned above.

[0018] In an embodiment of the present invention, after obtaining target forecast data and target observation data, a target forecast image may be generated based on the target forecast data, and a target observation image may be generated based on the target observation data; the target forecast data includes forecast information of a target pollution object in each grid area of a target area, the target forecast data is forecast by a target numerical model, and the target observation data includes observation information of the target pollution object in each grid area, and the image sizes of the target forecast image and the target observation image are the same. Then, feature extraction may be performed on the target forecast image to obtain a forecast feature extraction result of the target forecast image, and feature extraction may be performed on the target observation image to obtain an observation feature extraction result of the target observation image. Further, target similarity judgment information between the target forecast data and the target observation data may be determined based on the forecast feature extraction result and the observation feature extraction result; and based on the target similarity judgment information, a target air pollution simulation evaluation result may be determined, and the target air pollution simulation evaluation result is used to indicate the simulation effect of the target numerical model for the target pollution object. It can be seen that the embodiment of the present invention can perform air pollution simulation evaluation through the target forecast image and the target observation image, that is, the spatial distribution characteristics of the target pollution object can be evaluated through the forecast feature extraction result and the observation feature extraction result, and then a target air pollution simulation evaluation result with relatively high accuracy can be determined through the forecast feature extraction result and the observation feature extraction result, which can effectively improve the accuracy of air pollution simulation evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In the following description of exemplary embodiments with reference to the accompanying drawings, more details, features, and advantages of the present invention are disclosed. In the drawings:

[0020] Figure 1 A flowchart showing a method for air pollution simulation evaluation based on image processing according to an exemplary embodiment of the present invention is shown;

[0021] Figure 2 A flowchart showing another method for air pollution simulation evaluation based on image processing according to an exemplary embodiment of the present invention is shown;

[0022] Figure 3 A flowchart showing still another method for air pollution simulation evaluation based on image processing according to an exemplary embodiment of the present invention is shown;

[0023] Figure 4 A schematic block diagram showing an apparatus for air pollution simulation evaluation based on image processing according to an exemplary embodiment of the present invention is shown;

[0024] Figure 5 A structural block diagram showing an exemplary electronic device capable of implementing an embodiment of the present invention is shown. Detailed Embodiments

[0025] Embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Instead, these embodiments are provided to more thoroughly and completely understand the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not used to limit the protection scope of the present invention.

[0026] It should be understood that the various steps recited in the method embodiments of the present invention can be executed in different orders and / or executed in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this regard.

[0027] The term "including" and its variations used herein are open-ended, that is, "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description. It should be noted that the concepts such as "first" and "second" mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order of the functions performed by these devices, modules or units or their interdependent relationships.

[0028] It should be noted that the modifications of "one" and "plural" mentioned in the present invention are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly specified in the context, it should be understood as "one or more".

[0029] The names of the messages or information exchanged between multiple devices in the embodiments of the present invention are only for illustrative purposes and are not used to limit the scope of these messages or information.

[0030] It should be noted that the execution subject of the atmospheric pollution simulation and evaluation method based on image processing provided in the embodiments of the present invention can be one or more electronic devices, and the present invention does not limit this; among them, the electronic device can be a terminal (i.e., a client) or a server. Then, when the execution subject includes multiple electronic devices, and at least one terminal and at least one server are included in the multiple electronic devices, the atmospheric pollution simulation and evaluation method based on image processing provided in the embodiments of the present invention can be jointly executed by the terminal and the server. Correspondingly, the terminals mentioned here can include, but are not limited to: smart phones, tablet computers, laptop computers, desktop computers, etc.; the servers mentioned here can be independent physical servers, or a server cluster or distributed system composed of multiple physical servers, or can also be cloud servers providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms, etc.

[0031] Based on the above description, the embodiments of the present invention propose an atmospheric pollution simulation and evaluation method based on image processing. The atmospheric pollution simulation and evaluation method based on image processing can be executed by the above-mentioned electronic devices (terminals or servers); or, the atmospheric pollution simulation and evaluation method based on image processing can be jointly executed by the terminal and the server. For the convenience of elaboration, in the following, it is taken as an example that the electronic device executes the atmospheric pollution simulation and evaluation method based on image processing; as Figure 1 shown, the atmospheric pollution simulation and evaluation method based on image processing may include the following steps S101 - S105:

[0032] S101, obtain target forecast data, and obtain target observation data. The target forecast data includes the forecast information of the target pollution object in each grid area in the target area. The target forecast data is forecast by a target numerical model, and the target observation data includes the observation information of the target pollution object in each grid area.

[0033] Optionally, the target pollution object can be any pollutant (such as PM 10 (inhalable particulate matter), PM 2.5(Fine particulate matter), CO (carbon monoxide), O3 (ozone), etc., or it can also be AQI (Air Quality Index), etc. The embodiments of the present invention do not limit this. Optionally, the target area can be any area, and the embodiments of the present invention do not limit this; Exemplarily, the target area can be the national scope, or the scope of a province, etc. Optionally, the target forecast data and the target observation data can be data for the target area within the target time range. The target time range can be any time range, that is, the target forecast data and the target observation data can be data for the target area within the same time range; That is to say, the target forecast data can be the forecast data of the target area within the target time range through the target numerical model, and the target observation data can be the monitoring data (i.e., observation data) of the target area within the target time range, etc.; The embodiments of the present invention do not limit this. Optionally, the target numerical model can be any numerical model, that is, it can be any air quality numerical model, and the embodiments of the present invention do not limit this.

[0034] Optionally, the target area may include multiple grid areas, and each grid area in the target area can be each of the multiple grid areas; Optionally, the resolution of the target area (i.e., the grid area division method) can be set according to experience or according to actual needs, and the embodiments of the present invention do not limit this.

[0035] In the embodiments of the present invention, the acquisition methods of the target forecast data may include but are not limited to the following several:

[0036] The first acquisition method: The storage space of the electronic device may store the forecast data of the target area within multiple time ranges. In this case, the electronic device may select the forecast data of the target area within the target time range from the forecast data of the target area within multiple time ranges, so as to use the forecast data of the target area within the target time range as the target forecast data. The multiple time ranges may include the target time range.

[0037] The second acquisition method: The electronic device may obtain the target forecast data download link and download the target forecast data based on the target forecast data download link to achieve the acquisition of the target forecast data, etc.

[0038] Correspondingly, the acquisition methods of the target observation data may include but are not limited to the following several:

[0039] The first acquisition method: The storage space of the electronic device may store the observation data of the target area in multiple time ranges. In this case, the electronic device can select the observation data of the target area in the target time range from the observation data of the target area in multiple time ranges, and use the observation data of the target area in the target time range as the target observation data.

[0040] The second acquisition method: The electronic device can obtain the download link of the target observation data and download the target observation data based on the download link of the target observation data to obtain the target observation data. Optionally, the electronic device can also obtain the data integration download link, and use the forecast data in the data downloaded based on the data integration download link as the target forecast data, and use the observation data in the data downloaded based on the data integration download link as the target observation data.

[0041] The third acquisition method: The electronic device can obtain the observation site data of the target area in the target time range. The observation site data includes the observation values of the target pollution object at each observation site among multiple observation sites included in the target area. Based on this, the target observation data can be determined based on the observation site data. That is to say, through the spatial interpolation method, the observation site data can be interpolated into spatial data with the same resolution as the target forecast data (also called the target forecast field), so as to obtain the target observation data, and so on. Exemplarily, for any grid area in the target area, the electronic device can determine the top W observation sites closest to the any grid area from multiple observation sites, and perform weighted summation on the observation values of the target pollution object at each of the top W observation sites to obtain the observation information of the target pollution object in the any grid area, so as to obtain the target observation data. W is a positive integer. Optionally, the weights of each of the top W observation sites can be the same (such as mean operation, etc.), or can be the reciprocal of the distance to the any grid area, etc. The embodiments of the present invention do not limit this.

[0042] S102, generate a target forecast image based on the target forecast data, and generate a target observation image based on the target observation data. The image sizes of the target forecast image and the target observation image are the same.

[0043] In an embodiment of the present invention, the electronic device can determine the target image size and determine the mapping pixel points of each grid region according to the target image size; optionally, the target image size can be set according to experience or according to actual requirements, and the embodiments of the present invention do not limit this. Based on this, for any grid region in the target region, the electronic device can determine the mapping pixel points of any grid region based on the region position information of any grid region and the target image size; that is to say, the mapping abscissa and mapping ordinate of any grid region can be determined, and the pixel point closest to the coordinate point formed by the mapping abscissa and mapping ordinate of any grid region can be determined from the pixel coordinate system, so as to use the determined pixel point as the mapping pixel point of any grid region. Among them, the target image size can be used to indicate the maximum abscissa and maximum ordinate of the image (for example, when the target image size is 1000×900, the maximum abscissa of the image can be 1000, and the maximum ordinate of the image can be 900).

[0044] Optionally, when determining the mapping abscissa and mapping ordinate of any grid region, the mapping abscissa of any grid region can be calculated based on the abscissa in the region position information of any grid region, the maximum abscissa of the image, and the maximum abscissa of the target region; and the mapping ordinate of any grid region can be calculated based on the ordinate in the region position information of any grid region, the maximum ordinate of the image, and the maximum ordinate of the target region. Optionally, the region position information of any grid region can be the center point coordinates of any grid region (that is, the coordinates of the center point in the plane rectangular coordinate system). At this time, the maximum abscissa of the target region can be the maximum abscissa among the center point coordinates of each grid region, and the maximum ordinate of the target region can be the maximum ordinate among the center point coordinates of each grid region; or, the region position information of any grid region can be the grid label of any grid region (that is, the grid coordinates). For example, when any grid region is the grid region in the i-th row and the j-th column, the grid label of any grid region can be (i, j). At this time, the abscissa in the region position information of any grid region can be i, the ordinate in the region position information of any grid region can be j, the maximum abscissa of the target region can be I, and the maximum ordinate of the target region can be J, etc.; the embodiments of the present invention do not limit this. Among them, I is the number of grid regions in the target region in the horizontal direction, and J is the number of grid regions in the target region in the vertical direction.

[0045] Optionally, when calculating the mapped abscissa of any grid region based on the abscissa in the region position information of any grid region, the maximum abscissa of the image, and the maximum abscissa of the target region, the ratio between the maximum abscissa of the image and the maximum abscissa of the region can be multiplied by the abscissa in the region position information of any grid region to obtain the mapped abscissa of any grid region. Optionally, when the region position information of any grid region is the center point coordinates of any grid region, the ratio between the maximum ordinate of the image and the maximum ordinate of the region can be multiplied by the ordinate in the region position information of any grid region to obtain the result of the ordinate multiplication operation, and the difference between the maximum ordinate of the image and the result of the ordinate multiplication operation can be used as the mapped ordinate of any grid region; when the region position information of any grid region is the grid label of any grid region, the ratio between the maximum ordinate of the image and the maximum ordinate of the region can be multiplied by the ordinate in the region position information of any grid region to obtain the mapped ordinate of any grid region (i.e., the result of the ordinate multiplication operation can be used as the mapped ordinate of any grid region).

[0046] Further, the electronic device can determine the predicted pixel values of each pixel in the target prediction image based on the mapped pixel points of each grid region and the target prediction data (i.e., determine the predicted pixel values of each pixel among the multiple pixel points indicated by the target image size, and use the predicted pixel values of each pixel as the pixel values of the corresponding pixels in the target prediction image) to generate the target prediction image; wherein, the predicted pixel value of a pixel is determined based on the prediction information in each grid region of at least one grid region where the mapped pixel point is the corresponding pixel. In other words, for any pixel in the target prediction image (i.e., any pixel among the multiple pixels), the electronic device can determine at least one grid region in the target area where the mapped pixel point is the any pixel based on the mapped pixel points of each grid region, and determine the predicted pixel value of the any pixel based on the prediction information in each grid region of at least one grid region where the mapped pixel point is the any pixel. Optionally, the electronic device can use the mean value among the prediction information in each grid region of at least one grid region where the mapped pixel point is the any pixel as the predicted pixel value of the any pixel; or, can use the maximum value among the prediction information in each grid region of at least one grid region where the mapped pixel point is the any pixel as the predicted pixel value of the any pixel; or, if the mean value among the prediction information in each grid region of at least one grid region where the mapped pixel point is the any pixel is greater than or equal to the preset pollution object concentration threshold, use the mean value among the prediction information in each grid region where the mapped pixel point is the any pixel as the predicted pixel value of the any pixel, if the mean value among the prediction information in each grid region where the mapped pixel point is the any pixel is less than the preset pollution object concentration threshold, use the first preset pixel value as the predicted pixel value of the any pixel; or, if the mean value among the prediction information in each grid region where the mapped pixel point is the any pixel is greater than or equal to the preset pollution object concentration threshold, use the second preset pixel value as the predicted pixel value of the any pixel, if the mean value among the prediction information in each grid region where the mapped pixel point is the any pixel is less than the preset pollution object concentration threshold, use the first preset pixel value as the predicted pixel value of the any pixel, and so on; the embodiments of the present invention do not limit this. Optionally, the preset pollution object concentration threshold, the first preset pixel value, and the second preset pixel value can all be set according to experience or according to actual requirements, and the embodiments of the present invention do not limit this; exemplarily, the first preset pixel value can be 255, and the second preset pixel value can be 0. Among them, the above preset pollution object concentration threshold can be the preset pollution object concentration threshold corresponding to the target pollution object; the preset pollution object concentration thresholds corresponding to different pollution objects can be the same or different, and the embodiments of the present invention do not limit this.

[0047] Correspondingly, the electronic device can determine the observed pixel values of each pixel in the target observation image (i.e., determine the pixel values of each pixel in the target observation image) based on the mapped pixel points of each grid region and the target observation data, so as to generate the target observation image; wherein, the observed pixel value of a pixel is determined based on the observed information in each grid region of at least one grid region where the mapped pixel point is the corresponding pixel point. It should be understood that the determination method of the observed pixel values of each pixel in the target observation image can be the same as the determination method of the predicted pixel values of each pixel in the target prediction image, and the embodiments of the present invention will not elaborate herein.

[0048] It should be understood that an image (such as the target prediction image or the target observation image, etc.) can include multiple pixels, that is, an image can include multiple pixels indicated by the target image size. That is to say, each pixel in the target prediction image and each pixel in the target observation image can be each pixel among the multiple pixels. Thus, the predicted pixel values of each pixel can be respectively used as the pixel values of the corresponding pixels in the target prediction image, and the observed pixel values of each pixel can be respectively used as the pixel values of the corresponding pixels in the target observation image.

[0049] S103. Extract features from the target prediction image to obtain the prediction feature extraction result of the target prediction image, and extract features from the target observation image to obtain the observation feature extraction result of the target observation image.

[0050] S104. Based on the prediction feature extraction result and the observation feature extraction result, determine the target similarity judgment information between the target prediction data and the target observation data.

[0051] S105. Based on the target similarity judgment information, determine the target air pollution simulation evaluation result, and the target air pollution simulation evaluation result is used to indicate the simulation effect of the target numerical model for the target pollution object.

[0052] Optionally, if the target similarity judgment information is greater than or equal to the preset similarity judgment threshold, the electronic device can use the first simulation effect indication information as the target air pollution simulation evaluation result; if the target similarity judgment information is less than the preset similarity judgment threshold, then use the second simulation effect indication information as the target air pollution simulation evaluation result; wherein, the first simulation effect indication information can be used to indicate a good simulation effect (i.e., it can be used to indicate a high similarity between the target prediction data and the target observation data), and the second simulation effect indication information can be used to indicate a poor simulation effect (i.e., it can be used to indicate a low similarity between the target prediction data and the target observation data). Optionally, the preset similarity judgment threshold, the first simulation effect indication information, and the second simulation effect indication information can all be set according to experience or according to actual requirements, and the embodiments of the present invention do not limit this.

[0053] In other embodiments, the electronic device may also use the target similarity judgment information as the target air pollution simulation evaluation result. For example, if the target air pollution simulation evaluation result is larger, it may indicate a better simulation effect. Then, the quality of the simulation effect may be determined based on the size of the target air pollution simulation evaluation result, and so on.

[0054] In an embodiment of the present invention, after obtaining the target forecast data and the target observation data, a target forecast image may be generated based on the target forecast data, and a target observation image may be generated based on the target observation data. The target forecast data includes the forecast information of the target pollution object in each grid area of the target area. The target forecast data is forecast by the target numerical model, and the target observation data includes the observation information of the target pollution object in each grid area. The image sizes of the target forecast image and the target observation image are the same. Then, feature extraction may be performed on the target forecast image to obtain the forecast feature extraction result of the target forecast image, and feature extraction may be performed on the target observation image to obtain the observation feature extraction result of the target observation image. Further, the target similarity judgment information between the target forecast data and the target observation data may be determined based on the forecast feature extraction result and the observation feature extraction result. And based on the target similarity judgment information, the target air pollution simulation evaluation result may be determined. The target air pollution simulation evaluation result is used to indicate the simulation effect of the target numerical model for the target pollution object. It can be seen that the embodiment of the present invention can perform air pollution simulation evaluation through the target forecast image and the target observation image, that is, the spatial distribution characteristics of the target pollution object can be evaluated through the forecast feature extraction result and the observation feature extraction result. Furthermore, a target air pollution simulation evaluation result with higher accuracy can be determined through the forecast feature extraction result and the observation feature extraction result, which can effectively improve the accuracy of air pollution simulation evaluation.

[0055] Based on the above description, an embodiment of the present invention further proposes a more specific method for air pollution simulation evaluation based on image processing. Correspondingly, the method for air pollution simulation evaluation based on image processing may be executed by the above-mentioned electronic device (terminal or server); or, the method for air pollution simulation evaluation based on image processing may be jointly executed by the terminal and the server. For the sake of convenience of description, in the following, it is taken as an example that the electronic device executes the method for air pollution simulation evaluation based on image processing; please refer to Figure 2 , the method for air pollution simulation evaluation based on image processing may include the following steps S201 - S208:

[0056] S201, Obtain target forecast data and target observation data. The target forecast data includes the forecast information of the target pollution object in each grid area within the target region. The target forecast data is forecast by a target numerical model, and the target observation data includes the observation information of the target pollution object in each grid area.

[0057] S202, Generate a target forecast image based on the target forecast data, and generate a target observation image based on the target observation data. The image sizes of the target forecast image and the target observation image are the same.

[0058] S203, Perform blob detection on the target forecast image to obtain the blob features of each forecast blob in at least one forecast blob.

[0059] Among them, a blob generally refers to connected pixel points in an image that have similar attributes (such as color or brightness, etc.), that is, a blob usually refers to an area with a color and grayscale difference from the surrounding area.

[0060] Optionally, the electronic device can perform blob detection on the target forecast image through a target blob detection algorithm to obtain the blob features of each forecast blob in at least one forecast blob. Each forecast blob is a blob in the target forecast image. Optionally, the target blob detection algorithm can be the SimpleBlobDetector algorithm (a blob detection algorithm), or the Laplace of Gaussian (LOG) blob detection algorithm, etc.; the embodiments of the present invention do not limit this. Exemplarily, taking the SimpleBlobDetector algorithm as an example for illustration, the SimpleBlobDetector algorithm is good at detecting blobs of various shapes and sizes. For blobs with clear edges and obvious contrast with the surrounding area, the SimpleBlobDetector algorithm can accurately identify them. Among them, when using the SimpleBlobDetector algorithm for blob detection, first, the minimum and maximum thresholds can be set to perform threshold processing on the image to convert the image into multiple binary images. These thresholds are incremented at a certain step size until the maximum threshold is reached. Secondly, in each binary image, the connected areas can be grouped together to form binary blobs. Then, the centers of the binary blobs are calculated, and the blobs with closer centers are merged together. Finally, the algorithm performs feature description, calculates and returns the centers and radii of the newly merged blobs, etc., to achieve blob detection.

[0061] Optionally, the blob features of a blob (also referred to as a blob area) may include but are not limited to at least one of the following: the position (i.e., the center) of the corresponding blob, the size (i.e., the radius), the direction, the color, the area, and the standard deviation of the pixel values in the covered area, etc.; the embodiments of the present invention do not limit this.

[0062] S204. Determine the prediction feature extraction result of the target prediction image based on the feature of each prediction spot.

[0063] In one implementation, the electronic device may determine the spot indication image of the target prediction image based on the feature of each prediction spot and the target prediction image. Based on this, key point detection may be performed on the spot indication image of the target prediction image to obtain the key point features of each of at least one prediction key point in the spot indication image of the target prediction image, and the prediction feature extraction result of the target prediction image may be determined based on the key point features of each prediction key point in the spot indication image of the target prediction image.

[0064] Optionally, when determining the prediction feature extraction result of the target prediction image based on the key point features of each prediction key point in the spot indication image of the target prediction image, the electronic device may also perform key point detection on the target prediction image to obtain the key point features of each of at least one prediction key point in the target prediction image, and add the key point features of each prediction key point in the target prediction image and the key point features of each prediction key point in the spot indication image of the target prediction image to the prediction feature extraction result of the target prediction image to achieve the determination of the prediction feature extraction result. At this time, the feature points of the target prediction image can be fully extracted (i.e., it can include each prediction key point in the target prediction image and each prediction key point in the spot indication image of the target prediction image. At this time, the prediction feature extraction result may include the key point features of each prediction key point in the target prediction image and the key point features of each prediction key point in the spot indication image of the target prediction image), thereby further improving the accuracy of the feature extraction result; or, the key point features of each prediction key point in the spot indication image of the target prediction image may be used as the prediction feature extraction result, that is, the key point features of each prediction key point may be added to the prediction feature extraction result. At this time, the prediction feature extraction result may include the key point features of each prediction key point, and so on. The embodiments of the present invention do not limit this. Based on this, the approximate fall area of the target pollution object in the corresponding data (such as the target prediction data) can be represented by the contaminated image (such as the target prediction image) and / or the spot indication image, so as to represent the contaminated area in detail through key point detection and spot detection to achieve sufficient feature extraction and improve the accuracy of feature extraction.

[0065] Optionally, the electronic device can perform key point detection on the speckle indication image and / or the target prediction image of the target prediction image through a target key point detection algorithm to obtain the key point features of each of at least one predicted key point (such as the key point features of each predicted key point in the target prediction image and the key point features of each predicted key point in the speckle indication image of the target prediction image). Each predicted key point in the speckle indication image of the target prediction image is a key point in the speckle indication image of the target prediction image, and each predicted key point in the target prediction image is a key point in the target prediction image. Optionally, the target key point detection algorithm can be a Scale-invariant feature transform (SIFT) algorithm, etc., and the embodiments of the present invention are not limited thereto. Optionally, the key point features of a key point may include, but are not limited to, at least one of the following: the position, size, direction, and descriptor (which can be a representation vector) of the corresponding key point, etc.; the embodiments of the present invention are not limited thereto. Exemplarily, when the prediction feature extraction result includes the key point features of each predicted key point, the key point features of a key point may include the descriptor of the corresponding key point. Among them, the SIFT algorithm can work stably under a variety of different image conditions and is good at extracting key points of corners, edges, and repetitive textures. The first step of this algorithm is to search for image positions at all scales and identify potential key points that are invariant to scale and rotation through the Difference of Gaussian function; the second step is to locate the key points. At each candidate position, a refined model is fitted to determine the position and scale, and the key points are selected based on their stability; the third step is to determine the key point direction. Based on the local gradient direction of the image, one or more directions are assigned to each key point position, and all subsequent operations on the image data are transformed relative to the direction, scale, and position of the key point, thus ensuring invariance to these transformations; the fourth step is to describe the key points. In the neighborhood around each key point, the local gradient of the image is measured at the selected scale; finally, the SIFT algorithm can output the position, scale, direction, and descriptor of the key point, etc., to achieve key point detection.

[0066] Optionally, the spot feature of a spot includes the center point and radius of the corresponding spot. Then, in a specific implementation, when determining the spot indication image of the target prediction image based on the spot features of each predicted spot and the target prediction image, the electronic device may, based on the spot features of each predicted spot, retain the pixel value (herein referred to as the predicted pixel value) of each pixel located on each predicted spot in the target prediction image, and set the pixel value of each pixel not located on any predicted spot in the target prediction image to a first preset pixel value, so as to determine the spot indication image of the target prediction image; that is, for any pixel in the target prediction image (i.e., any one of the above-mentioned multiple pixels), it may be determined, based on the spot features of each predicted spot, whether the pixel is located on any one of at least one predicted spot. If the pixel is located on any one of the predicted spots, the pixel value of the pixel in the target prediction image is used as the pixel value of the pixel in the spot indication image of the target prediction image. If the pixel is not located on any one of at least one predicted spot (i.e., the pixel is not located on each predicted spot), the first preset pixel value is used as the pixel value of the pixel in the spot indication image of the target prediction image. In another specific implementation, the electronic device may also determine whether the pixel is located on any one of at least one predicted spot. If the pixel is located on any one of the predicted spots, the second preset pixel value (such as 0) is used as the pixel value of the pixel in the spot indication image of the target prediction image. If the pixel is not located on any one of at least one predicted spot, the first preset pixel value is used as the pixel value of the pixel in the spot indication image of the target prediction image, and so on. Herein, a pixel being located on a spot may mean that the distance between the corresponding pixel and the center point (i.e., the position) of the corresponding spot is less than or equal to the radius of the corresponding spot. That is, the pixels located on a spot may include all the pixels among the multiple pixels whose distance from the center point of the corresponding spot is less than or equal to the radius of the corresponding spot; correspondingly, a pixel not being located on a spot may mean that the distance between the corresponding pixel and the center point of the corresponding spot is greater than the radius of the corresponding spot.

[0067] In another implementation, the electronic device can also perform key point detection on the target prediction image to obtain the key point features of each target prediction key point in at least one target prediction key point, and add the spot features of each prediction spot and the key point features of each target prediction key point to the prediction feature extraction result, so that each prediction spot and each target prediction key point are prediction feature points in the prediction feature extraction result, thereby realizing the determination of the prediction feature extraction result, and so on. In this case, the prediction feature extraction result can include the spot features of each prediction spot and the key point features of each target prediction key point, and at this time, the features of a feature point can include but are not limited to at least one of the following: the position, size, and direction of the corresponding feature point, etc. A feature point is a spot or a key point (i.e., spots and key points can be collectively referred to as feature points), and the feature of a feature point can be the spot feature of a spot or the key point feature of a key point, that is, at this time, the spot feature of a spot and the key point feature of a key point can both include at least one of the following: the corresponding position, size, and direction, etc.

[0068] In yet another implementation, the electronic device can use the spot features of each prediction spot as the prediction feature extraction result of the target prediction image; that is, the spot features of each prediction spot can be added to the prediction feature extraction result of the target prediction image to realize the determination of the prediction feature extraction result of the target prediction image. In this case, the features of a feature point can include but are not limited to at least one of the following: the position, size, and direction of the corresponding feature point.

[0069] In other embodiments, the electronic device may not perform spot detection on the target prediction image, but only perform key point detection on the target prediction image to obtain the key point features of each prediction key point in at least one prediction key point in the target prediction image, and thus add the key point features of each prediction key point in the target prediction image to the prediction feature extraction result of the target prediction image to obtain the prediction feature extraction result of the target prediction image.

[0070] Optionally, when the feature points in the feature extraction result are all key points, a key point feature can be a descriptor of the corresponding key point, so as to more accurately represent the corresponding key point, and further improve the accuracy of the feature extraction result.

[0071] S205, perform spot detection on the target observation image to obtain the spot features of each observation spot in at least one observation spot.

[0072] It should be noted that the implementation of performing spot detection on the target observation image can be the same as the implementation of performing spot detection on the target prediction image described above, and the embodiments of the present invention will not be elaborated here.

[0073] S206. Determine the observation feature extraction result of the target observation image based on the spot features of each observation spot.

[0074] It should be noted that the determination method of the observation feature extraction result can be the same as that of the prediction feature extraction result, and will not be elaborated in this embodiment of the present invention. Exemplarily, the electronic device can determine the spot indication image of the target observation image based on the spot features of each observation spot and the target observation image; then perform key point detection on the spot indication image of the target observation image to obtain the key point features of each observation key point in at least one observation key point of the spot indication image of the target observation image, and determine the observation feature extraction result of the target observation image based on the key point features of each observation key point in the spot indication image of the target observation image (for example, key point detection can also be performed on the target observation image to obtain the key point features of each observation key point in at least one observation key point of the target observation image, and add the key point features of each observation key point in the target observation image and the key point features of each observation key point in the spot indication image of the target observation image to the observation feature extraction result of the target observation image), etc.; it should be understood that the specific implementation manner of determining the observation feature extraction result of the target observation image based on the key point features of each observation key point in the spot indication image of the target observation image can be the same as the specific implementation manner of determining the prediction feature extraction result of the target prediction image based on the key point features of each prediction key point in the spot indication image of the target prediction image, and will not be elaborated in this embodiment of the present invention.

[0075] S207. Determine the target similarity judgment information between the target prediction data and the target observation data based on the prediction feature extraction result and the observation feature extraction result.

[0076] In this embodiment of the present invention, the electronic device can perform feature matching on the prediction feature extraction result and the observation feature extraction result to obtain a feature matching result; and determine the target similarity judgment information between the target prediction data and the target observation data based on the feature matching result.

[0077] In one implementation, the prediction feature extraction result may include the key point features of each prediction key point in the prediction key point set, and the observation feature extraction result may include the key point features of each observation key point in the observation key point set. That is, at this time, the prediction feature points in the prediction feature extraction result may include the prediction key point set, and the observation feature points in the observation feature extraction result may include the prediction key point set. Based on this, when performing feature matching on the prediction feature extraction result and the observation feature extraction result to obtain a feature matching result, the electronic device may perform feature matching on the key point features of each prediction key point in the prediction key point set and the key point features of each observation key point in the observation key point set to obtain a feature matching result. Correspondingly, when determining the target similarity judgment information between the target prediction data and the target observation data based on the feature matching result, at least one matching point in the prediction key point set may be determined based on the feature matching result, and the number of matching points in the at least one matching point may be used to calculate the target similarity judgment information between the target prediction data and the target observation data. Optionally, the prediction key point set may include at least one prediction key point in the target prediction image and at least one prediction key point in the speckle indication image of the target prediction image, and the observation key point set may include at least one observation key point in the target observation image and at least one observation key point in the speckle indication image of the target observation image; or, the prediction key point set may include at least one prediction key point in the speckle indication image of the target prediction image, and the observation key point set may include at least one observation key point in the speckle indication image of the target observation image; or, the prediction key point set may include all the prediction key points located on any one of the at least one prediction speckles included in at least one prediction key point in the target prediction image, and the observation key point set may include all the observation key points located on any one of the at least one observation speckles included in at least one observation key point in the target observation image; or, the prediction key point set may include at least one prediction key point in the target prediction image, and the observation key point set may include at least one observation key point in the target observation image, and so on; the embodiments of the present invention do not limit this.

[0078] In a specific implementation, when performing feature matching on the key-point features of each predicted key point in the predicted key-point set and the key-point features of each observed key point in the observed key-point set to obtain a feature matching result, the electronic device may divide the key-point features of each predicted key point in the predicted key-point set into at least one predicted feature subset, and divide the key-point features of each observed key point in the observed key-point set into at least one observed feature subset. A feature subset includes the key-point features of each key point in at least one key point, that is, a predicted feature subset may include the key-point features of each key point in at least one key point in the predicted key-point set, and an observed feature subset may include the key-point features of each key point in at least one key point in the observed key-point set. Then, based on at least one observed feature subset, approximate nearest neighbor search may be performed on each predicted feature subset in the at least one predicted feature subset respectively to obtain the nearest neighbor observed feature subset of each predicted feature subset. The nearest neighbor observed feature subset of a predicted feature subset is an observed feature subset in the at least one observed feature subset. Based on this, the nearest neighbor observed feature subsets of each predicted feature subset may be used as the feature matching result. At this time, the feature matching result may include the nearest neighbor observed feature subsets of each predicted feature subset. Optionally, for any predicted feature subset in the at least one predicted feature subset, the electronic device may calculate the distance (also referred to as the approximate distance) between any predicted feature subset and each observed feature subset in the at least one observed feature subset respectively, and use the observed feature subset with the smallest distance between any predicted feature subset and the at least one observed feature subset as the nearest neighbor observed feature subset of any predicted feature subset. Optionally, for any observed feature subset in the at least one observed feature subset, the mean value (i.e., the mean vector) between the key-point features in any predicted feature subset and the mean value between the key-point features in any observed feature subset may be used to calculate the distance between any predicted feature subset and any observed feature subset, such as the Euclidean distance or the Mahalanobis distance between the two mean values, etc.

[0079] Accordingly, when determining at least one matching point in the predicted key point set based on the feature matching result, for any predicted key point in the predicted key point set, if there is a key point feature of the matching observed key point corresponding to any predicted key point in the nearest neighbor observed feature subset of the predicted feature subset to which any predicted key point belongs (i.e., the predicted feature subset where the key point feature of any predicted key point is located), then any predicted key point can be used as a matching point. If there is no key point feature of the matching observed key point corresponding to any predicted key point in the nearest neighbor observed feature subset of the predicted feature subset to which any predicted key point belongs, then any predicted key point may not be used as a matching point. The matching observed key point corresponding to any predicted key point refers to an observed key point whose distance from the key point feature of any predicted key point is less than the preset key point feature threshold; and / or, if the distance between the predicted feature subset to which any predicted key point belongs and the nearest neighbor observed feature subset of the predicted feature subset to which any predicted key point belongs is less than the preset feature subset distance threshold, then any predicted key point can be used as a matching point. If the distance between the predicted feature subset to which any predicted key point belongs and the nearest neighbor observed feature subset of the predicted feature subset to which any predicted key point belongs is greater than or equal to the preset feature subset distance threshold, then any predicted key point may not be used as a matching point, so as to determine at least one matching point in the predicted key point set, and so on. Optionally, both the preset key point feature threshold and the preset feature subset distance threshold can be set according to experience or according to actual requirements, and the embodiments of the present invention do not limit this.

[0080] In another specific implementation, when performing feature matching on the key point features of each predicted key point in the predicted key point set and the key point features of each observed key point in the observed key point set to obtain a feature matching result, for any predicted key point in the predicted key point set, the electronic device can calculate the distances between the key point feature of any predicted key point and the key point features of each observed key point in the observed key point set respectively, and based on the distances between the key point feature of any predicted key point and the key point features of each observed key point, determine the nearest neighbor observed key point of any predicted key point (i.e., the observed key point with the smallest distance between the key point feature and the key point feature of any predicted key point) from the observed key point set, so as to use the nearest neighbor observed key points of each predicted key point as the feature matching result, that is, at this time the feature matching result can include the nearest neighbor observed key points of each predicted key point in the predicted key point set.

[0081] Correspondingly, when determining at least one matching point in the set of predicted key points based on the feature matching result, if the distance between the key point feature of any predicted key point and the nearest neighbor observed key point of any predicted key point is less than the preset key point feature threshold, then any predicted key point is taken as a matching point; if the distance between the key point feature of any predicted key point and the nearest neighbor observed key point of any predicted key point is greater than or equal to the preset key point feature threshold, then any predicted key point is not taken as a matching point, so as to determine at least one matching point in the set of predicted key points.

[0082] Further, when calculating the target similarity judgment information between the target predicted data and the target observed data by using the number of matching points among at least one matching point, the electronic device may use the ratio between the number of matching points among at least one matching point (i.e., the number of matching points among at least one matching point) and the number of key points in the set of predicted key points (i.e., the number of key points in the set of predicted key points) as the target similarity judgment information between the target predicted data and the target observed data; or, the ratio between the number of matching points among at least one matching point and the number of key points in the set of observed key points may be used as the target similarity judgment information between the target predicted data and the target observed data, and so on.

[0083] In another implementation manner, the predicted feature extraction result may include the features of each predicted feature point among multiple predicted feature points, and the observed feature extraction result may include the features of each observed feature point among multiple observed feature points. Optionally, the multiple predicted feature points may include at least one predicted spot and at least one target predicted key point, the multiple observed feature points may include at least one observed spot and at least one target observed key point, at least one target observed key point may be obtained by performing key point detection on the target observed image (i.e., at least one observed key point in the target observed image), and at least one target predicted key point may be obtained by performing key point detection on the target predicted image (i.e., at least one predicted key point in the target predicted image); or, the multiple predicted feature points may include at least one predicted spot, and the multiple observed feature points may include at least one observed spot, and so on; the embodiments of the present invention do not limit this.

[0084] Based on this, when performing feature matching on the forecast feature extraction result and the observation feature extraction result to obtain the feature matching result, the features of each forecast feature point and the features of each observation feature point can be subjected to feature matching to obtain the feature matching result; correspondingly, when determining the target similarity judgment information between the target forecast data and the target observation data based on the feature matching result, at least one feature matching point among the multiple forecast feature points can be determined based on the feature matching result, and the number of feature matching points among the at least one feature matching point can be used to calculate the target similarity judgment information between the target forecast data and the target observation data. For example, the target similarity judgment information can be calculated using the number of feature matching points of the at least one feature matching point and the number of feature points among the multiple forecast feature points. It should be noted that the specific implementation manner of performing feature matching on the features of each forecast feature point and the features of each observation feature point can be the same as the specific implementation manner of performing feature matching on the key point features of each forecast key point in the forecast key point set and the key point features of each observation key point in the observation key point set; moreover, the specific implementation manner of determining at least one feature matching point among the multiple forecast feature points based on the feature matching result can be the same as the specific implementation manner of determining at least one matching point in the forecast key point set based on the feature matching result as described above, and the embodiments of the present invention will not be elaborated herein.

[0085] Optionally, the multiple forecast feature points can also be a forecast key point set (i.e., at this time, the multiple forecast feature points can include all the forecast key points in the forecast key point set), and the multiple observation feature points can also be an observation key point set (i.e., at this time, the multiple observation feature points can include all the observation key points in the observation key point set), and so on.

[0086] S208. Determine the target air pollution simulation evaluation result based on the target similarity judgment information, and the target air pollution simulation evaluation result supports indicating the simulation effect of the target numerical model for the target pollution object.

[0087] Optionally, the embodiments of the present invention can implement the method for simulating and evaluating air pollution based on image processing through Python (a high-level scripting language that combines interpretability, compilation, interactivity, and object orientation), or can also be implemented through C language (a general-purpose, procedural computer programming language), etc.; the embodiments of the present invention do not limit this. Exemplarily, taking Python software as an example, the embodiments of the present invention can install Python software on an electronic device and import various libraries required for calculation, such as including but not limited to at least one of the following: the OpenCV library (an open-source computer vision library) cv2 in Python version and the library numpy (an extension library of the Python language that supports a large number of dimensional array and matrix operations) for numerical calculation, etc.; based on this, after reading the images (i.e., the target forecast image and the target observation image) through Python software, subsequent air pollution simulation and evaluation can be carried out, or the target forecast data and the target observation data can be read to carry out subsequent air pollution simulation and evaluation after the reading of the target forecast data and the target observation data is completed, etc., the embodiments of the present invention do not limit this; it can be seen that the embodiments of the present invention can use OpenCV to perform corresponding processing on the images, and finally use the image similarity judgment method of OpenCV to conduct similarity comparison, so as to evaluate the quality of the simulation effect. Optionally, when the embodiments of the present invention perform air pollution simulation and evaluation through the OpenCV library, the SimpleBlobDetector algorithm can be used for blob detection, the SIFT algorithm can be used for key point detection, and the FLANN (Fast Library for Approximate Nearest Neighbors) library can be used for feature matching, etc.; among them, the FLANN library is a library for approximate nearest neighbor search.

[0088] In summary, taking the example that the forecast feature extraction result includes the key point features of each forecast key point in the target forecast image and the key point features of each forecast key point in the blob indication image of the target forecast image, and the observation feature extraction result includes the key point features of each observation key point in the target observation image and the key point features of each observation key point in the blob indication image of the target observation image, as Figure 3As shown, after the electronic device obtains the target forecast image and the target observation image, it can respectively perform image feature point recognition and extraction on the target forecast image and the target observation image (that is, respectively perform speckle detection and key point detection on the target forecast image and the target observation image, and respectively perform key point detection on the speckle indication image of the target forecast image and the speckle indication image of the target observation image), so as to obtain the forecast feature extraction result and the observation feature extraction result. Furthermore, the forecast feature extraction result and the observation feature extraction result can be subjected to feature matching (that is, similar feature matching) to determine the target similarity judgment information, and thus the target similarity judgment information can be used for similarity judgment to determine the target air pollution simulation evaluation result; Exemplarily, when the preset similarity judgment threshold is 70%, if the target similarity judgment information is greater than or equal to 70%, it can be determined that the target air pollution simulation evaluation result is used to indicate good simulation effect. If the target similarity judgment information is less than 70%, it can be determined that the target air pollution simulation evaluation result is used to indicate poor simulation effect, and so on.

[0089] In an embodiment of the present invention, after obtaining target forecast data and target observation data, a target forecast image can be generated based on the target forecast data, and a target observation image can be generated based on the target observation data. The image sizes of the target forecast image and the target observation image are the same. Further, spot detection can be performed on the target forecast image to obtain the spot features of each of at least one forecast spot, and based on the spot features of each forecast spot, a forecast feature extraction result of the target forecast image can be determined; and spot detection can be performed on the target observation image to obtain the spot features of each of at least one observation spot, and based on the spot features of each observation spot, an observation feature extraction result of the target observation image can be determined. Based on this, target similarity judgment information between the target forecast data and the target observation data can be determined based on the forecast feature extraction result and the observation feature extraction result; and based on the target similarity judgment information, a target air pollution simulation evaluation result can be determined, and the target air pollution simulation evaluation result supports indicating the simulation effect of the target numerical model for the target pollution object. It can be seen that the embodiment of the present invention can evaluate the spatial distribution characteristics of the target pollution object through the forecast feature extraction result and the observation feature extraction result (i.e., the spatial distribution characteristics can be evaluated through the detection of spots and key points), that is, the air pollution simulation evaluation method proposed in the embodiment of the present invention can also be called an air pollution simulation spatial distribution evaluation method. That is to say, the embodiment of the present invention can screen out the values of the pollution areas to be inspected to output images of the pollution areas (such as the target forecast image and the target observation image), so that the shape of the simulated pollution area indicated by the target forecast data and the shape of the observed pollution area indicated by the target observation data can be compared for similarity through the forecast feature extraction result and the observation feature extraction result (the shape of the pollution area can be represented by spots and / or key points), thereby effectively evaluating the quality of the simulation effect to improve the accuracy of air pollution simulation evaluation; based on this, the embodiment of the present invention can apply the spatial evaluation method to the field of air pollution and can evaluate the air pollution simulation effect in a way of image recognition, which can avoid problems such as large subjective uncertainty and lack of intelligence caused by manually assigning spatial attribute weights.

[0090] Based on the description of the related embodiments of the above air pollution simulation evaluation method based on image processing, an embodiment of the present invention also proposes an air pollution simulation evaluation device based on image processing. The air pollution simulation evaluation device based on image processing can be a computer program (including program code) running in an electronic device; as Figure 4 shown, the air pollution simulation evaluation device based on image processing can include an acquisition unit 401 and a processing unit 402. The air pollution simulation evaluation device based on image processing can execute Figure 1 or Figure 2The atmospheric pollution simulation and evaluation method based on image processing shown, that is, the atmospheric pollution simulation and evaluation device based on image processing can run the above units:

[0091] An acquisition unit 401, configured to acquire target forecast data and acquire target observation data, where the target forecast data includes forecast information of a target pollution object in each grid area of a target area, the target forecast data is forecast by a target numerical model, and the target observation data includes observation information of the target pollution object in each grid area;

[0092] A processing unit 402, configured to generate a target forecast image based on the target forecast data, and generate a target observation image based on the target observation data, where the target forecast image and the target observation image have the same image size;

[0093] The processing unit 402 is further configured to perform feature extraction on the target forecast image to obtain a forecast feature extraction result of the target forecast image, and perform feature extraction on the target observation image to obtain an observation feature extraction result of the target observation image;

[0094] The processing unit 402 is further configured to determine target similarity judgment information between the target forecast data and the target observation data based on the forecast feature extraction result and the observation feature extraction result;

[0095] The processing unit 402 is further configured to determine a target atmospheric pollution simulation and evaluation result based on the target similarity judgment information, and the target atmospheric pollution simulation and evaluation result is used to indicate the simulation effect of the target numerical model for the target pollution object.

[0096] In an implementation manner, when the processing unit 402 performs feature extraction on the target forecast image to obtain a forecast feature extraction result of the target forecast image, it may specifically be configured to:

[0097] Perform speckle detection on the target forecast image to obtain speckle features of each of at least one forecast speckle;

[0098] Determine a forecast feature extraction result of the target forecast image based on the speckle features of each of the forecast speckles;

[0099] When the processing unit 402 performs feature extraction on the target observation image to obtain an observation feature extraction result of the target observation image, it may specifically be configured to:

[0100] Perform speckle detection on the target observation image to obtain speckle features of each of at least one observation speckle;

[0101] Determine the observation feature extraction result of the target observation image based on the feature of each observed spot.

[0102] In another implementation, when the processing unit 402 determines the prediction feature extraction result of the target prediction image based on the feature of each predicted spot, it may specifically be used for:

[0103] Determine the spot indication image of the target prediction image based on the feature of each predicted spot and the target prediction image;

[0104] Perform key point detection on the spot indication image of the target prediction image to obtain the key point features of each prediction key point in at least one prediction key point in the spot indication image of the target prediction image, and determine the prediction feature extraction result of the target prediction image based on the key point features of each prediction key point in the spot indication image of the target prediction image;

[0105] When the processing unit 402 determines the observation feature extraction result of the target observation image based on the feature of each observed spot, it may specifically be used for:

[0106] Determine the spot indication image of the target observation image based on the feature of each observed spot and the target observation image;

[0107] Perform key point detection on the spot indication image of the target observation image to obtain the key point features of each observation key point in at least one observation key point in the spot indication image of the target observation image, and determine the observation feature extraction result of the target observation image based on the key point features of each observation key point in the spot indication image of the target observation image.

[0108] In another implementation, when the processing unit 402 determines the target similarity judgment information between the target prediction data and the target observation data based on the prediction feature extraction result and the observation feature extraction result, it may specifically be used for:

[0109] Perform feature matching on the prediction feature extraction result and the observation feature extraction result to obtain a feature matching result;

[0110] Determine the target similarity judgment information between the target prediction data and the target observation data based on the feature matching result.

[0111] In another implementation, the prediction feature extraction result includes the key point features of each prediction key point in the prediction key point set, and the observation feature extraction result includes the key point features of each observation key point in the observation key point set; when the processing unit 402 performs feature matching on the prediction feature extraction result and the observation feature extraction result to obtain a feature matching result, it can be specifically used for:

[0112] Perform feature matching on the key point features of each prediction key point in the prediction key point set and the key point features of each observation key point in the observation key point set to obtain a feature matching result;

[0113] When the processing unit 402 determines the target similarity judgment information between the target prediction data and the target observation data based on the feature matching result, it can be specifically used for:

[0114] Based on the feature matching result, determine at least one matching point in the prediction key point set, and use the number of matching points in the at least one matching point to calculate the target similarity judgment information between the target prediction data and the target observation data.

[0115] In another implementation, when the processing unit 402 performs feature matching on the key point features of each prediction key point in the prediction key point set and the key point features of each observation key point in the observation key point set to obtain a feature matching result, it can be specifically used for:

[0116] Divide the key point features of each prediction key point in the prediction key point set into at least one prediction feature subset, and divide the key point features of each observation key point in the observation key point set into at least one observation feature subset. One feature subset includes the key point features of each key point in at least one key point;

[0117] Based on the at least one observation feature subset, perform approximate nearest neighbor search on each prediction feature subset in the at least one prediction feature subset to obtain the nearest neighbor observation feature subset of each prediction feature subset. The nearest neighbor observation feature subset of one prediction feature subset is one observation feature subset in the at least one observation feature subset; and use the nearest neighbor observation feature subsets of the respective prediction feature subsets as the feature matching result;

[0118] When the processing unit 402 determines at least one matching point in the prediction key point set based on the feature matching result, it can be specifically used for:

[0119] For any prediction key point in the set of prediction key points, if there is a key point feature of the matching observation key point corresponding to the any prediction key point in the nearest neighbor observation feature subset of the prediction feature subset to which the any prediction key point belongs, then the any prediction key point is taken as a matching point, and the matching observation key point corresponding to the any prediction key point refers to an observation key point whose distance from the key point feature of the any prediction key point is less than a preset key point feature threshold; and / or,

[0120] If the distance between the prediction feature subset to which the any prediction key point belongs and the nearest neighbor observation feature subset of the prediction feature subset to which the any prediction key point belongs is less than a preset feature subset distance threshold, then the any prediction key point is taken as a matching point to determine at least one matching point in the set of prediction key points.

[0121] In another embodiment, when generating a target prediction image based on the target prediction data, the processing unit 402 may be specifically configured to:

[0122] Determine the target image size, and determine the mapped pixel points of each grid region according to the target image size;

[0123] Based on the mapped pixel points of each grid region and the target prediction data, determine the predicted pixel values of each pixel point in the target prediction image to generate the target prediction image; wherein, the predicted pixel value of a pixel point is determined based on the prediction information in each grid region of at least one grid region where the mapped pixel point is the corresponding pixel point.

[0124] When generating a target observation image based on the target observation data, the processing unit 402 may be specifically configured to:

[0125] Based on the mapped pixel points of each grid region and the target observation data, determine the observed pixel values of each pixel point in the target observation image to generate the target observation image; wherein, the observed pixel value of a pixel point is determined based on the observation information in each grid region of at least one grid region where the mapped pixel point is the corresponding pixel point.

[0126] According to an embodiment of the present invention, Figure 4Each unit in the atmospheric pollution simulation and evaluation device based on image processing shown can be separately or all combined into one or several other units to form, or some of them can be further split into multiple smaller units in terms of function to form, which can achieve the same operation without affecting the realization of the technical effects of the embodiments of the present invention. The above units are divided based on logical functions. In practical applications, the function of one unit can also be realized by multiple units, or the functions of multiple units can be realized by one unit. In other embodiments of the present invention, any atmospheric pollution simulation and evaluation device based on image processing can also include other units. In practical applications, these functions can also be assisted by other units and can be realized by the cooperation of multiple units.

[0127] According to another embodiment of the present invention, it can be achieved by running a computer program (including program code) capable of executing the respective steps involved in the corresponding method shown in Figure 1 or Figure 2 on a general electronic device such as a computer including processing elements and storage elements such as a central processing unit (CPU), a random access storage medium (RAM), and a read-only storage medium (ROM), to construct the atmospheric pollution simulation and evaluation device based on image processing shown in Figure 4 and to implement the atmospheric pollution simulation and evaluation method based on image processing of the embodiments of the present invention. The computer program can be recorded on, for example, a computer storage medium, loaded into the above-mentioned electronic device through the computer storage medium, and run therein.

[0128] Based on the descriptions of the above method embodiments and device embodiments, an exemplary embodiment of the present invention further provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, and the computer program, when executed by the at least one processor, is used to cause the electronic device to execute the method according to the embodiments of the present invention.

[0129] An exemplary embodiment of the present invention further provides a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to execute the method according to the embodiments of the present invention.

[0130] An exemplary embodiment of the present invention further provides a computer program product, including a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to execute the method according to the embodiments of the present invention.

[0131] Refer to Figure 5, a block diagram of an electronic device 500 that can be a server or a client of the present invention will now be described. It is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device is intended to represent various forms of digital electronic computer devices, such as, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0132] As Figure 5 shown, the electronic device 500 includes a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 can also be stored. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0133] Multiple components in the electronic device 500 are connected to the I / O interface 505, including: an input unit 506, an output unit 507, a storage unit 508, and a communication unit 509. The input unit 506 can be any type of device that can input information into the electronic device 500. The input unit 506 can receive input digital or character information, and generate key signal inputs related to the user settings and / or function controls of the electronic device. The output unit 507 can be any type of device that can present information, and can include, but is not limited to, a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 508 can include, but is not limited to, magnetic disks, optical disks. The communication unit 509 allows the electronic device 500 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks, and can include, but is not limited to, a modem, a network card, an infrared communication device, a wireless communication transceiver, and / or a chipset, such as a BluetoothTM device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0134] The computing unit 501 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 501 executes the various methods and processes described above. For example, in some embodiments, the method for simulating and evaluating air pollution based on image processing can be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as the storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 500 via the ROM 502 and / or the communication unit 509. In some embodiments, the computing unit 501 can be configured to execute the method for simulating and evaluating air pollution based on image processing in any other suitable manner (e.g., by means of firmware).

[0135] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to the processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing devices, such that when the program codes are executed by the processor or controller, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as an independent software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0136] In the context of the present invention, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0137] As used in this invention, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and / or device (e.g., a magnetic disk, an optical disk, a memory, a programmable logic device (PLD)) used to provide machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0138] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0139] Also, it should be understood that the above-disclosed is only the preferred embodiment of the present invention, and of course cannot be used to limit the scope of rights of the present invention. Therefore, equivalent changes made according to the claims of the present invention still fall within the scope covered by the present invention.

Claims

1. An atmospheric pollution simulation and assessment method based on image processing, characterized in that: include: Acquiring target forecast data and target observation data, wherein the target forecast data includes forecast information of the target pollutant in each grid area in the target area, the target forecast data is forecasted by a target numerical model, and the target observation data includes observation information of the target pollutant in each grid area; Based on the target prediction data, a target prediction image is generated, including: determining a target image size, and determining mapping pixel points of each grid area according to the target image size; based on the mapping pixel points of each grid area and the target prediction data, determining a predicted pixel value of each pixel point in the target prediction image, so as to realize the generation of the target prediction image; wherein the predicted pixel value of a pixel point is determined based on the prediction information in each grid area of ​​at least one grid area where the mapping pixel point is a corresponding pixel point; Based on the target observation data, generating a target observation image, comprising: determining an observed pixel value of each pixel in the target observation image based on the mapped pixel points of each grid area and the target observation data, so as to realize generating the target observation image; wherein the observed pixel value of a pixel is determined based on the observation information in each grid area of ​​at least one grid area where the mapped pixel point is a corresponding pixel point; and the image size of the target prediction image and the target observation image is the same; Performing feature extraction on the target prediction image to obtain a prediction feature extraction result of the target prediction image, and performing feature extraction on the target observation image to obtain an observation feature extraction result of the target observation image; Determining target similarity judgment information between the target forecast data and the target observation data based on the forecast feature extraction result and the observation feature extraction result; Based on the target similarity judgment information, a target air pollution simulation evaluation result is determined, and the target air pollution simulation evaluation result supports indicating a simulation effect of the target numerical model on the target pollution object.

2. The method according to claim 1, characterized in that The step of extracting features from the target prediction image to obtain a prediction feature extraction result of the target prediction image includes: Performing spot detection on the target prediction image to obtain spot features of each prediction spot in at least one prediction spot; Determining a prediction feature extraction result of the target prediction image based on the spot features of each prediction spot; The step of extracting features from the target observation image to obtain an observation feature extraction result of the target observation image includes: Performing spot detection on the target observation image to obtain spot features of each observation spot in at least one observation spot; Based on the spot features of the respective observation spots, an observation feature extraction result of the target observation image is determined.

3. The method according to claim 2, characterized in that The step of determining the prediction feature extraction result of the target prediction image based on the spot features of each prediction spot includes: Determining a spot indication image of the target prediction image based on the spot features of the respective prediction spots and the target prediction image; Performing key point detection on the spot indication image of the target prediction image to obtain key point features of each prediction key point in at least one prediction key point in the spot indication image of the target prediction image, and determining a prediction feature extraction result of the target prediction image based on the key point features of each prediction key point in the spot indication image of the target prediction image; The step of determining the observation feature extraction result of the target observation image based on the spot features of each observation spot includes: Determining a spot indication image of the target observation image based on the spot features of the respective observation spots and the target observation image; Key point detection is performed on the spot indication image of the target observation image to obtain key point features of each observation key point in at least one observation key point in the spot indication image of the target observation image, and based on the key point features of each observation key point in the spot indication image of the target observation image, an observation feature extraction result of the target observation image is determined.

4. The method according to any one of claims 1 to 3, characterized in that: The determining, based on the forecast feature extraction result and the observation feature extraction result, target similarity judgment information between the target forecast data and the target observation data comprises: Performing feature matching on the forecast feature extraction result and the observation feature extraction result to obtain a feature matching result; Based on the feature matching result, target similarity judgment information between the target forecast data and the target observation data is determined.

5. The method according to claim 4, characterized in that The forecast feature extraction result includes key point features of each forecast key point in the forecast key point set, and the observation feature extraction result includes key point features of each observation key point in the observation key point set; the feature matching of the forecast feature extraction result and the observation feature extraction result to obtain the feature matching result includes: Performing feature matching on the key point features of each forecast key point in the forecast key point set and the key point features of each observation key point in the observation key point set to obtain a feature matching result; The determining, based on the feature matching result, target similarity judgment information between the target forecast data and the target observation data comprises: Based on the feature matching result, at least one matching point in the forecast key point set is determined, and the number of matching points in the at least one matching point is used to calculate the target similarity judgment information between the target forecast data and the target observation data.

6. The method according to claim 5, characterized in that The performing feature matching on the key point features of each forecast key point in the forecast key point set and the key point features of each observation key point in the observation key point set to obtain a feature matching result includes: Dividing the key point features of each prediction key point in the prediction key point set into at least one prediction feature subset, and dividing the key point features of each observation key point in the observation key point set into at least one observation feature subset, wherein one feature subset includes the key point features of each key point in at least one key point; Based on the at least one observation feature subset, respectively perform an approximate nearest neighbor search on each forecast feature subset in the at least one forecast feature subset to obtain the nearest neighbor observation feature subset of each forecast feature subset, wherein the nearest neighbor observation feature subset of a forecast feature subset is an observation feature subset in the at least one observation feature subset; and use the nearest neighbor observation feature subset of each forecast feature subset as a feature matching result; The determining, based on the feature matching result, at least one matching point in the set of prediction key points comprises: For any forecast key point in the forecast key point set, if a key point feature of a matching observation key point corresponding to any forecast key point exists in the nearest neighbor observation feature subset of the forecast feature subset to which the any forecast key point belongs, then the any forecast key point is taken as a matching point, and the matching observation key point corresponding to the any forecast key point refers to an observation key point whose key point feature has a distance with the key point feature of the any forecast key point less than a preset key point feature threshold; and / or, If the distance between the forecast feature subset to which any forecast key point belongs and the nearest observed feature subset of the forecast feature subset to which any forecast key point belongs is less than a preset feature subset distance threshold, then any forecast key point is taken as a matching point to determine at least one matching point in the forecast key point set.

7. An atmospheric pollution simulation and assessment device based on image processing, characterized in that: The device comprises: an acquisition unit, configured to acquire target forecast data and target observation data, wherein the target forecast data includes forecast information of the target pollutant in each grid area in the target area, the target forecast data is forecasted by a target numerical model, and the target observation data includes observation information of the target pollutant in each grid area; A processing unit, configured to generate a target prediction image based on the target prediction data, comprising: determining a target image size, and determining mapping pixels of each grid area according to the target image size; determining a predicted pixel value of each pixel in the target prediction image based on the mapping pixels of each grid area and the target prediction data, so as to realize the generation of the target prediction image; wherein the predicted pixel value of a pixel is determined based on the prediction information in each grid area of ​​at least one grid area where the mapping pixel is a corresponding pixel; The processing unit is further used to generate a target observation image based on the target observation data, including: determining the observed pixel value of each pixel in the target observation image based on the mapped pixel points of each grid area and the target observation data, so as to realize the generation of the target observation image; wherein the observed pixel value of a pixel is determined based on the observation information in each grid area of ​​at least one grid area where the mapped pixel point is the corresponding pixel point; the image size of the target prediction image and the target observation image is the same; The processing unit is further used to perform feature extraction on the target prediction image to obtain a prediction feature extraction result of the target prediction image, and perform feature extraction on the target observation image to obtain an observation feature extraction result of the target observation image; The processing unit is further used to determine target similarity judgment information between the target forecast data and the target observation data based on the forecast feature extraction result and the observation feature extraction result; The processing unit is further used to determine a target air pollution simulation evaluation result based on the target similarity judgment information, and the target air pollution simulation evaluation result supports indicating a simulation effect of the target numerical model on the target pollution object.

8. An electronic device, characterized in that: include: processor; as well as Memory for storing programs, The program includes instructions, which, when executed by the processor, cause the processor to perform the method according to any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to make a computer execute the method according to any one of claims 1-6.

Images