Land utilization rate calculation method, device, electronic equipment and storage medium

By identifying and analyzing the characteristic data of above-ground attachments in the monitoring image, combined with the use information, the problem of low land utilization calculation accuracy in the prior art is solved, and a more accurate and efficient land utilization evaluation is achieved.

CN118314457BActive Publication Date: 2024-11-15广州市基础地理信息中心(广州市规划和自然资源自动化中心) +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410413147.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-08
Publication Date
2024-11-15
Estimated Expiration
2044-04-08

AI Technical Summary

Technical Problem

In the prior art, the calculation method of land utilization is difficult to reflect the actual state of the land through the area ratio only, the calculation results are relatively accurate and difficult to accurately reflect the development purpose.

Method used

By obtaining the monitoring image of the target land, identifying the above-ground attachments and obtaining their characteristic data, including the number and density of objects, and using the CNN neural network and similarity matching model to identify and classify the above-ground attachments and uses to determine the above-ground attachments and uses. The matching of information can then calculate the land use rate.

Benefits of technology

It improves the accuracy and efficiency of land use calculation, simplifies the calculation process, and can more accurately reflect the development purpose and status of the land.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118314457B_ABST
    Figure CN118314457B_ABST
Patent Text Reader

Abstract

The present invention discloses a method, device, electronic device and storage medium for calculating land utilization rate, the method comprising: after determining the target land and the use information of the target land, obtaining the monitoring image of the target land according to the use information; obtaining the ground attachments of the target land from the monitoring image and when it is determined that the ground attachments are the same as the objects included in the use information, obtaining the characteristic data of the ground attachments in the monitoring image, the characteristic data comprising: the number of objects and the density of objects; using the characteristic data to calculate the land utilization rate according to the use rules included in the use information. The present invention can not only simplify the calculation process and shorten the processing time of the calculation to improve the calculation efficiency; but also combine the number of objects and density for comprehensive calculation, which can effectively improve the calculation accuracy.
Need to check novelty before this filing date? Find Prior Art

Description

Background Art

[0002] Land utilization rate is an indicator that reflects the degree of land utilization. It is the ratio of developed land area to total land area. Land utilization rate is of great significance in national land space planning management, natural resource management, environmental protection and other aspects.

[0003] Land development and utilization generally involves planting or construction on the land. For this purpose, the commonly used calculation method is: manually detect the total land area, detect the real-time development area of ​​the land according to the land type, and calculate the ratio of the real-time development area to the total area to obtain the land utilization rate.

[0004] However, the currently commonly used calculation method has the following technical problems: it is difficult to reflect the actual status of the land only through the area ratio, the calculation result deviates from the development purpose of the land, and the calculation accuracy is low. Summary of the invention

[0005] The present invention provides a land utilization rate calculation method, device, electronic device and storage medium, and the method can solve one or more of the above-mentioned technical problems.

[0006] A first aspect of an embodiment of the present invention provides a method for calculating land utilization rate, the method comprising:

[0007] After determining the target land and the usage information of the target land, obtaining a monitoring image of the target land according to the usage information;

[0008] When obtaining ground attachments of the target land from the monitoring image and determining that the ground attachments are the same as the objects included in the usage information, obtaining feature data of the ground attachments in the monitoring image, the feature data including: the number of objects and the density of objects;

[0009] The characteristic data is used to calculate the land utilization rate according to the usage rules included in the usage information.

[0010] In a possible implementation manner of the first aspect, determining that the ground attachment is the same as the object included in the usage information includes:

[0011] Calling a preset recognition model to recognize the ground attachment to obtain recognition information, wherein the recognition information includes: object chromaticity value and object outline;

[0012] According to the object contour, searching for a corresponding purpose object from the purpose information, and obtaining a chromaticity value corresponding to the purpose object to obtain a purpose chromaticity value;

[0013] If the object chromaticity value is the same as the usage chromaticity value, it is determined that the ground attachment is the same as the object included in the usage information;

[0014] If the object chromaticity value is different from the purpose chromaticity value or the corresponding purpose object cannot be found from the purpose information according to the object contour, it is determined that the ground attachment is different from the object included in the purpose information.

[0015] In a possible implementation manner of the first aspect, acquiring the ground attachments of the target land from the monitoring image includes:

[0016] If the monitoring image is one, preprocessing the monitoring image to obtain a processed image;

[0017] A preset CNN neural network is called to identify the processed image to obtain ground attachments, wherein the preset CNN neural network includes a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a first fully connected layer, a second fully connected layer and an output layer connected in sequence.

[0018] In a possible implementation manner of the first aspect, acquiring the ground attachments of the target land from the surveillance image includes:

[0019] If there are multiple surveillance images, each surveillance image is subjected to definition adjustment processing to obtain multiple adjusted images;

[0020] Extracting a view feature from each of the monitoring images to obtain a plurality of two-dimensional view features, wherein the ground attachment is a three-dimensional object;

[0021] After the plurality of two-dimensional view features are classified and combined to obtain the combined features, a classification decision process is performed on the combined features using a preset similarity matching model, and the ground attachments are determined according to the classification categories.

[0022] In a possible implementation manner of the first aspect, acquiring a monitoring image of the target land according to the usage information includes:

[0023] If the land use type included in the use information is agricultural land, determining whether the horizontal direction of the camera of the real-time shooting device facing the target land is blocked;

[0024] If the camera of the real-time shooting device is facing the horizontal direction of the target land without any obstruction, the camera of the real-time shooting device is called to shoot at the center of the target land to obtain a monitoring image;

[0025] If the camera of the real-time shooting device is blocked in the horizontal direction toward the target land, obtain the horizontal distance between the camera of the real-time shooting device and the center of the target land, and control the camera to rise and shoot toward the center of the target land according to the distance to obtain a monitoring image.

[0026] In a possible implementation manner of the first aspect, after the step of calculating the land utilization rate by using the characteristic data according to the usage rule included in the usage information, the method further includes:

[0027] If the land utilization rate is lower than the preset utilization rate, the current time node is obtained, and the historical density is searched according to the current time node;

[0028] If the difference between the object density and the historical density is greater than a preset value, land planning suggestion information is generated and sent to the management terminal for reference by the management personnel.

[0029] A second aspect of an embodiment of the present invention provides a land utilization rate calculation device, the device comprising:

[0030] An image acquisition module is used to acquire a monitoring image of the target land according to the usage information after determining the target land and the usage information of the target land;

[0031] A feature acquisition module, configured to acquire ground attachments of the target land from the monitoring image and, when it is determined that the ground attachments are the same as the objects included in the usage information, acquire feature data of the ground attachments in the monitoring image, wherein the feature data includes: the number of objects and the density of objects;

[0032] The utilization rate calculation module is used to calculate the land utilization rate according to the usage rules included in the usage information by using the characteristic data.

[0033] In a possible implementation manner of the second aspect, determining that the ground attachment is the same as the object included in the usage information includes:

[0034] Calling a preset recognition model to recognize the ground attachment to obtain recognition information, wherein the recognition information includes: object chromaticity value and object outline;

[0035] According to the object contour, searching for a corresponding purpose object from the purpose information, and obtaining a chromaticity value corresponding to the purpose object to obtain a purpose chromaticity value;

[0036] If the object chromaticity value is the same as the usage chromaticity value, it is determined that the ground attachment is the same as the object included in the usage information;

[0037] If the object chromaticity value is different from the purpose chromaticity value or the corresponding purpose object cannot be found from the purpose information according to the object contour, it is determined that the ground attachment is different from the object included in the purpose information.

[0038] In a possible implementation manner of the second aspect, acquiring the ground attachments of the target land from the monitoring image includes:

[0039] If the monitoring image is one, preprocessing the monitoring image to obtain a processed image;

[0040] A preset CNN neural network is called to identify the processed image to obtain ground attachments, wherein the preset CNN neural network includes a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a first fully connected layer, a second fully connected layer and an output layer connected in sequence.

[0041] In a possible implementation manner of the second aspect, acquiring the ground attachments of the target land from the monitoring image includes:

[0042] If there are multiple surveillance images, each surveillance image is subjected to definition adjustment processing to obtain multiple adjusted images;

[0043] Extracting a view feature from each of the monitoring images to obtain a plurality of two-dimensional view features, wherein the ground attachment is a three-dimensional object;

[0044] After the plurality of two-dimensional view features are classified and combined to obtain the combined features, a classification decision process is performed on the combined features using a preset similarity matching model, and the ground attachments are determined according to the classification categories.

[0045] In a possible implementation manner of the second aspect, acquiring a monitoring image of the target land according to the usage information includes:

[0046] If the land use type included in the use information is agricultural land, determining whether the horizontal direction of the camera of the real-time shooting device facing the target land is blocked;

[0047] If the camera of the real-time shooting device is facing the horizontal direction of the target land without any obstruction, the camera of the real-time shooting device is called to shoot at the center of the target land to obtain a monitoring image;

[0048] If the camera of the real-time shooting device is blocked in the horizontal direction toward the target land, obtain the horizontal distance between the camera of the real-time shooting device and the center of the target land, and control the camera to rise and shoot toward the center of the target land according to the distance to obtain a monitoring image.

[0049] In a possible implementation manner of the second aspect, the apparatus further includes:

[0050] A density acquisition module is used to acquire the current time node and search for the historical density according to the current time node if the land utilization rate is lower than the preset utilization rate;

[0051] The suggestion sending module is used to generate and send land planning suggestion information to the management terminal for reference by the management personnel if the difference between the object density and the historical density is greater than a preset value.

[0052] A third aspect of an embodiment of the present invention provides a land utilization rate calculation system, the system comprising: a cloud system and a plurality of real-time shooting devices, the cloud system being applicable to the land utilization rate calculation method as described above;

[0053] The cloud system is respectively connected to each of the real-time shooting devices for communication.

[0054] Compared with the prior art, the land utilization rate calculation method, device, electronic device and storage medium provided by the embodiments of the present invention have the following beneficial effects: after determining the target land and the use information of the target land, the present invention can obtain the monitoring image of the target land according to the use information; obtain the ground attachments of the target land from the monitoring image and when it is determined that the ground attachments are the same as the objects included in the use information, obtain the feature data of the ground attachments in the monitoring image; and calculate the land utilization rate according to the use rules included in the use information using the feature data. The present invention can not only simplify the calculation process and shorten the calculation processing time to improve the calculation efficiency by performing recognition and calculation through real-time detected images, but also effectively improve the calculation accuracy by combining the number and density of objects for comprehensive calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flow chart of a land utilization rate calculation method provided by an embodiment of the present invention;

[0056] Figure 2 It is a structural schematic diagram of a land utilization rate calculation device provided by an embodiment of the present invention;

[0057] Figure 3 It is a structural schematic diagram of a land utilization rate calculation system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] In order to solve the above problems, a land utilization rate calculation method provided in the embodiment of the present application will be introduced and explained in detail through the following specific embodiments.

[0060] Reference Figure 1 , which shows a flow chart of a land utilization rate calculation method provided by an embodiment of the present invention.

[0061] In one embodiment, the land utilization rate calculation method is applicable to a cloud platform, a cloud system or a backend management system, which can perform rapid calculations based on the use information and monitoring images of the target land, simplify the calculation process, and improve the calculation accuracy.

[0062] In one embodiment, the land utilization rate calculation method may be to calculate the land utilization rate of agricultural land. For example, if corn is planted on farmland, the land utilization rate may be calculated based on data such as the planting density and the planting time. For another example, if wheat is planted on farmland, the land utilization rate may be calculated based on data such as the planting density, the planting time, and the height of the wheat.

[0063] As an example, the land utilization rate calculation method may include:

[0064] S11. After determining the target land and the usage information of the target land, acquiring a monitoring image of the target land according to the usage information.

[0065] Specifically, the target land may be agricultural land for which the user needs to calculate the land utilization rate. For example, a piece of agricultural land in xx village, xx town, xx city. After determining the target land, the use information of the target land may be obtained. The use information may be the use information reported by the land to the relevant departments, and may include: the type of plants to be planted, the size of the land area, the planting time, the land authority and other information. For example, the target land is agricultural land in the countryside, used to grow peanuts, and its use information may include: the land is at xx meters on xx road, xx village, xx town, xx city, the land is used to grow peanuts, the land area is 1.5 mu, the planting start time is xx / xx / xx, the estimated planting area is 1 mu, and the ownership of the land is Zhang San, etc.

[0066] After determining the use information of the target land, the real-time shooting equipment (for example, a camera or a remote sensing tower) installed on the side of the target land can be called to shoot the target land according to the land location of the use information, so as to obtain the monitoring image of the target land.

[0067] In one embodiment, the real-time shooting device can communicate with the cloud system, and the real-time shooting device can be a tower-type ground remote sensing platform. The general height range is about 30 to 300 meters. The real-time shooting device can be loaded and unloaded with various bands and types of remote sensing sensors, cameras and other detection equipment to capture images and obtain various remote sensing data and images.

[0068] In one implementation, the real-time shooting device may be set on the side of the target land. For example, if the target land is agricultural land, the real-time shooting device may be set beside the field of the agricultural land, and the surrounding fields may be detected by the real-time shooting device.

[0069] During specific operations, the cloud system can extract land location information from the usage information, and then determine the attached real-time shooting device based on the land location information of the usage information, and then control the camera of the real-time shooting device to shoot towards the target land, thereby obtaining a monitoring image.

[0070] In one operation mode, the actual position of each real-time shooting device is different, and the shooting direction of the real-time shooting device may also be blocked, resulting in low clarity of the image captured by the real-time shooting device or difficulty in capturing the entire target land, which in turn causes deviations in subsequent calculations and reduces the accuracy of the calculations. For example, the real-time shooting device is set in a field path next to two farmlands, and can capture farmlands on both sides at the same time. The plants planted in the farmland on the left are shorter, so the real-time shooting device can capture the entire picture of the farmland; while the plants planted in the farmland on the right are taller, and some of the plants block the real-time shooting device camera from shooting, making it impossible for the real-time shooting device to capture the entire picture of the farmland.

[0071] In order to avoid the above situation, as an example, step S11 may include the following sub-steps:

[0072] S111. If the land use type included in the usage information is agricultural land, determine whether there is any obstruction in the horizontal direction of the camera of the real-time shooting device toward the target land.

[0073] S112: If the camera of the real-time shooting device is facing the horizontal direction of the target land without any obstruction, the camera of the real-time shooting device is called to shoot toward the center of the target land to obtain a monitoring image.

[0074] S113. If the camera of the real-time shooting device is blocked in the horizontal direction toward the target land, obtain the distance between the camera of the real-time shooting device and the center of the target land in the horizontal direction, and control the camera to rise and shoot toward the center of the target land according to the distance to obtain a monitoring image.

[0075] Specifically, it may be determined first whether the land use type included in the use information is agricultural land. If the land use type included in the use information is not agricultural land, an alarm message or an error prompt message may be fed back to the user.

[0076] If the land use type included in the use information is agricultural land, it can be determined whether the horizontal direction of the camera of the real-time shooting device toward the target land is blocked. Specifically, it can be determined whether there is any blockage in a preset distance in front of the camera.

[0077] If there is no obstruction within the preset distance in front of the camera, the real-time shooting device camera can be directly called to shoot at the center of the target land to obtain a monitoring image.

[0078] On the contrary, if there is an obstruction within a preset distance in front of the camera, it is determined that there is an obstruction in the horizontal direction of the camera of the real-time shooting device toward the target land.

[0079] At this time, in order to reduce the impact of occlusion, the horizontal distance between the camera and the center of the target land can be obtained, and then the camera can be controlled to rise according to the distance. After rising, the camera can be controlled to shoot toward the center of the target land to obtain a monitoring image.

[0080] For example, assuming that the horizontal distance between the camera and the center of the target land is 5 meters, and the current height of the camera is 3 meters, the camera can be controlled to rise 5 meters by calling the internal lifting device of the real-time shooting device, so that the current height of the camera is 8 meters, and then the center of the target land is shot to obtain a monitoring image.

[0081] Optionally, if the adjusted height is greater than the maximum height of the camera, the camera may be adjusted to the maximum height.

[0082] S12. When obtaining ground attachments of the target land from the monitoring image and determining that the ground attachments are the same as the objects included in the usage information, obtain feature data of the ground attachments in the monitoring image, wherein the feature data includes: the number of objects and the density of objects.

[0083] In one embodiment, the ground attachments of the target land can be obtained from the monitoring image. The ground attachments can be objects planted or placed on the land, and it can be determined whether the ground attachments are the same as the objects included in the use information. If it is determined that the ground attachments are the same as the objects included in the use information, it is determined that the currently detected land meets the requirements, and feature data can be extracted from the monitoring image. The feature data includes: the number of objects and the density of objects.

[0084] In one embodiment, if it is determined that the attachments on the ground are different from the objects included in the use information, the land may be used illegally, the land utilization rate of the land may be determined to be 0, and an alarm message is sent to the background terminal.

[0085] In an optional embodiment, the captured surveillance image may be one or more images. In order to perform real-time recognition on one surveillance image, as an example, obtaining the ground attachments of the target land from the surveillance image may include the following sub-steps:

[0086] S21: If the monitoring image is one, pre-process the monitoring image to obtain a processed image.

[0087] S22. Calling a preset CNN neural network to identify the processed image to obtain ground attachments, wherein the preset CNN neural network includes a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a first fully connected layer, a second fully connected layer and an output layer connected in sequence.

[0088] In one implementation, if the surveillance image is one, the surveillance image may be preprocessed to obtain a processed image. The preprocessing may include: removing null values, removing boundaries, standardizing, and removing noise. Optionally, the preprocessing may be performed in a conventional operation mode. In one embodiment, the preprocessing may be adjusted according to actual needs. The null values ​​and image boundaries are removed, and the minimum boundary left is used as the range for subsequent research.

[0089] In order to eliminate phase interference and remove noise, time domain median filtering is required. The principle is that the reflectivity of the image is located in the abnormal maximum value area in each band, while the image is mostly located in the abnormal minimum value area, and taking the median has a good effect.

[0090] After completing the preprocessing, the preset CNN neural network can be called to identify the processed image to obtain the ground attachments. The preset CNN neural network includes a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a first fully connected layer, a second fully connected layer and an output layer connected in sequence.

[0091] During recognition, the convolutional layer and pooling layer are actually performing feature extraction operations, and the extracted features are then input into the fully connected layer for recognition.

[0092] Specifically, in the first convolutional layer of CNN, a local area (filter) of the image is selected to scan the entire image. All nodes enclosed by the local area will be multiplied and accumulated with the filter, and then connected to a node in the next layer. Assuming that the image to be scanned is a grayscale image (that is, it has only one color channel), all filters are also two-dimensional matrices.

[0093] It should be noted that if the image to be scanned by the convolution layer is a color image (that is, the image has three channels, RGB). If a filter convolves an image (regardless of whether the image is single-channel or multi-channel), it can only obtain information about one surface, and multiple filters can obtain multiple surfaces (this is a figurative understanding). The process of filter scanning the image (convolution) can be regarded as the process of extracting features. One convolution extracts features of a class of images (for example, multiple filters extract pixel features of the red channel, green channel, and blue channel of the image respectively).

[0094] The main function of the pooling layer is to compress the image. If the pooling layer is sandwiched between consecutive convolutional layers, it is used to compress the data convolved by the previous convolutional layer (or it can also be understood as filtering) to reduce overfitting.

[0095] For a 3*3 window, select the largest number as the value of the corresponding element of the output matrix. For example, if the largest number in the first 3*3 window of the input matrix is ​​5, then the first element of the output matrix is ​​5, and so on. Please note that the pixels scanned by the window each time will not be repeated to avoid repeated filtering of pixels.

[0096] The fully connected layer uses the output of the previous convolutional layer and pooling layer as feature values ​​to train the network, and uses the back-propagation algorithm to correct the parameters and thresholds of the fully connected layer and the weights of the previous convolutional layer filter.

[0097] The convolutional neural network (CNN) of the present invention can use the pixel values ​​of the image after a series of processing (convolution, pooling) as the output value to feed the model training. The advantage of this approach is that it can retain the multi-dimensional information of the image and improve the accuracy of model classification and recognition.

[0098] In an optional embodiment, the captured surveillance images may be one or more. In order to perform real-time recognition on the multiple surveillance images, as an example, the step of obtaining the ground attachments of the target land from the surveillance images may include the following sub-steps:

[0099] S31. If there are multiple surveillance images, perform definition adjustment processing on each surveillance image to obtain multiple adjusted images.

[0100] S32. Extracting view features from each of the monitoring images to obtain a plurality of two-dimensional view features, wherein the ground attachment is a three-dimensional object.

[0101] S33, after classifying and combining the multiple two-dimensional view features to obtain combined features, use a preset similarity matching model to perform classification decision processing on the combined features, and determine the ground attachments according to the classification category.

[0102] In one embodiment, the real-time shooting device may control the camera to shoot the target land multiple times to obtain multiple monitoring images. When there are multiple monitoring images, in order to comprehensively identify the multiple monitoring images, the clarity adjustment process may be performed on each monitoring image to obtain multiple adjusted images. Specifically, a super-resolution reconstruction algorithm may be used to adjust the clarity of the monitoring images.

[0103] After the adjustment is completed, each surveillance image can be adjusted and extracted to obtain the surveillance image view features.

[0104] Specifically, a preset convolutional network model can be used to adjust and extract each surveillance image to obtain multiple two-dimensional view features, wherein the ground attachment is a three-dimensional object, and the multiple surveillance images are images of the ground attachment at multiple angles. Specifically, the multiple surveillance images are all images of the ground attachment.

[0105] In this embodiment, the ground attachment is an object whose type is recognized by the real-time shooting device in the current scene.

[0106] In this embodiment, the three-dimensional ground attachments can be understood as any object that actually exists, that is, the three-dimensional ground attachments are based on the actual parameters of the three directions of the three-dimensional coordinate system. Specifically, for the agricultural land scene, it can be cultivated agricultural products, including peanuts, corn, rice or tomatoes.

[0107] In this embodiment, the preset convolutional network model may include a feature extraction network, and the feature extraction network may use DenseNet, ResNet or other convolutional network structures.

[0108] As an optional implementation, multiple monitoring images can be obtained by using real-time shooting equipment on several sides of the farmland.

[0109] By implementing this embodiment, real-time shooting equipment can be controlled around the target land to obtain multiple monitoring images, thereby avoiding errors in obtaining multiple images, thereby improving the accuracy of image acquisition, and further improving the accuracy of object recognition.

[0110] As an optional implementation, after the step of acquiring multiple monitoring images of the target land, the following steps may also be included:

[0111] Verify multiple surveillance images to determine whether each surveillance image in the multiple surveillance images matches the preset ground attachment. If they match, execute subsequent steps; if they do not match, discard the surveillance image.

[0112] It should be noted that there may be multiple preset ground attachments, including planted plants, crops, crop-related equipment or tools, etc.

[0113] In this embodiment, another verification result of multiple surveillance images can be achieved, thereby improving the accuracy of the multiple surveillance images and improving the accuracy of object recognition.

[0114] In this embodiment, the different angles of the multiple surveillance images are based on spatial angles.

[0115] In this embodiment, each surveillance image may include a two-dimensional view feature, wherein multiple two-dimensional view features are for multiple surveillance images as a whole. Therefore, a surveillance image may include or not include a two-dimensional view feature, or may include multiple two-dimensional image features, which is not limited in this embodiment.

[0116] In this embodiment, the preset convolutional network model also needs to obtain the category score corresponding to each of the multiple two-dimensional view features during the training process. The category score can be recognized and verified by the machine or operator, thereby improving the accuracy of the convolutional network model in extracting the two-dimensional view features, thereby improving the accuracy of the convolutional network model during use.

[0117] In this embodiment, the category score refers to a specific value corresponding to each two-dimensional view feature, and the specific value is used to match the corresponding object category. It can be seen that the category score is used to match the object category.

[0118] In this embodiment, the category score may be a specific score value (actual value), a percentage value, or other types of data, which is not limited in this embodiment.

[0119] In this embodiment, the specific score value can be directly used for category matching, and the percentage value is used to indicate how likely the feature is to be an object of a specified category. In this embodiment, the convolutional network model is an artificial intelligence model.

[0120] Next, multiple two-dimensional view features are combined and processed according to a preset aggregation network model to obtain a combination feature and a classification result corresponding to the combination feature. In this embodiment, the aggregation network model is an artificial intelligence model.

[0121] In this embodiment, the combined feature refers to a registered three-dimensional feature set.

[0122] In this embodiment, the classification result may be expressed in the form of a numerical value, in the form of a percentage, or in the form of a category name of an object, and this is not limited in this embodiment.

[0123] In this embodiment, the above-mentioned numerical value means the numerical value used to match the category; the percentage means the matching ratio with a certain type of object; the above-mentioned category name is the name of the category, such as table, chair or cup, etc.

[0124] Finally, multiple combined features are classified and decided based on the preset similarity matching model to obtain the positive and negative example score results.

[0125] In this embodiment, the similarity matching model is an artificial intelligence model.

[0126] As an optional implementation, when a classification decision is performed on a plurality of combined features according to a preset similarity matching model, some two-dimensional view features in the plurality of combined features may be obtained;

[0127] According to the preset similarity matching model, some two-dimensional view features are classified and decided to obtain positive and negative example score results.

[0128] By implementing this embodiment, classification decision processing can be performed based on some two-dimensional view features in the multiple combined features, thereby avoiding the round-robin of all data.

[0129] In this embodiment, the partial two-dimensional view feature may refer to a portion of the two-dimensional view features in a plurality of combined features.

[0130] In this embodiment, the two-dimensional view feature is the current minimum feature unit and will not be further subdivided, so a part of the two-dimensional view feature does not have an understanding deviation of a part of the two-dimensional view feature.

[0131] In this embodiment, the positive and negative example score results may be positive example scores or negative example scores, and may be specifically divided into true positive example scores, false positive example scores, true negative example scores, and false negative example scores.

[0132] In this embodiment, the positive and negative example score results are used to indicate whether the combined feature is consistent with multiple two-dimensional view features, that is, to determine whether they are features of the same object. Specifically, it is to determine whether the combined feature is a feature of a ground attachment, or whether it satisfies the identification feature of a ground attachment.

[0133] When the positive and negative example score result is a positive example score, the category of the ground attachment is determined based on the classification result.

[0134] In this embodiment, when the positive and negative example scores are positive example scores, it is used to indicate that the combined feature is a feature of a ground attachment, that is, after completing this step of verification, it can be shown that the combined feature has no deviation problem, so the category of the object can be determined based on the classification result.

[0135] Through the above method, multiple images of ground attachments (three-dimensional objects) based on multiple angles can be preferentially obtained, and multiple two-dimensional view features can be obtained by extracting features from multiple images using a convolutional network model, so that the two-dimensional feature view can be extracted by an artificial intelligence model. Compared with the previous extraction method, the feature extraction using an artificial intelligence model can effectively improve the accuracy of feature acquisition; after obtaining the above two-dimensional view features, the features are combined in the aggregation network model to obtain combined features and classification results corresponding to the combined features. In this step, the features can be combined by the artificial intelligence model to obtain effective combined features, so that the subsequent steps can make classification decisions based on the combined features, and then improve the accuracy of recognition by comparing more comprehensive comparison objects, and avoid the problem of low universality caused by a single feature; finally, the two-dimensional view features and the combined features are comprehensively classified and decided according to the preset similarity matching model, and the object category corresponding to the classification result is determined when the two are judged to be the same or similar, thereby achieving the effect of improving the recognition accuracy through verification, and the stability can also be improved by using the positive and negative example scores. It can be seen that the above implementation process can confirm the type of object by combining multiple monitoring images with an artificial intelligence model, thereby achieving the technical effect of improving the accuracy and universality of identifying ground attachments.

[0136] In an optional embodiment, the determining that the ground attachment is the same as the object included in the usage information may include the following sub-steps:

[0137] S121. Calling a preset recognition model to recognize the ground attachment to obtain recognition information, wherein the recognition information includes: object chromaticity value and object outline.

[0138] S122. Searching for a corresponding purpose object from the purpose information according to the object contour, and acquiring a chromaticity value corresponding to the purpose object to obtain a purpose chromaticity value.

[0139] S123: If the object chromaticity value is the same as the usage chromaticity value, it is determined that the ground attachment is the same as the object included in the usage information.

[0140] S124: If the object chromaticity value is different from the purpose chromaticity value or the corresponding purpose object cannot be found from the purpose information according to the object contour, it is determined that the ground attachment is different from the object included in the purpose information.

[0141] After obtaining the ground attachments, in order to further determine whether the ground attachments meet the requirements, a preset recognition model may be called to recognize the ground attachments to obtain recognition information, wherein the recognition information includes: object chromaticity value and object contour.

[0142] In one embodiment, the preset recognition model may use DenseNet, ResNet or other convolutional network structures.

[0143] Next, an object with the same outline as the object can be searched from the usage information according to the object outline to obtain the corresponding usage object, and at the same time, a chromaticity value corresponding to the usage object can be obtained to obtain the usage chromaticity value.

[0144] If the object chromaticity value is the same as the purpose chromaticity value, it is determined that the ground attachment is the same as the object included in the purpose information. Conversely, if the object chromaticity value is different from the purpose chromaticity value, or if the corresponding purpose object cannot be found from the purpose information according to the object outline, it is determined that the ground attachment is different from the object included in the purpose information.

[0145] In actual application scenarios, the captured image may have chromaticity value deviation, or, due to brightness or angle, there may be chromaticity value deviation.

[0146] In a preferred embodiment, the chromaticity value interval corresponding to the purpose object can be obtained to obtain the purpose chromaticity interval.

[0147] If the object chromaticity value is within the usage chromaticity interval, it is determined that the ground attachment is the same as the object included in the usage information. Conversely, if the object chromaticity value and the usage chromaticity value are not within the usage chromaticity interval, it is determined that the ground attachment is different from the object included in the usage information.

[0148] S13: Calculate the land utilization rate using the characteristic data according to the usage rules included in the usage information.

[0149] In one embodiment, the corresponding calculation formula or conversion formula can be searched in the database according to the usage rules contained in the usage information, and then the feature data is converted into a corresponding text vector, and the text vector is substituted into the calculation formula or conversion formula to calculate the land utilization rate.

[0150] In one embodiment, the calculation formula of the land utilization rate may be as follows:

[0151]

[0152] C is the final calculated land utilization rate; A and B are calculation constants or calculation weight values, which can be adjusted according to actual needs; P i is the number of objects, P k is the density of the object.

[0153] In one of the embodiments, the calculated land utilization rate can also represent the unit yield of the planted crops. If the number of planted crops is large and the density is high, the land utilization rate is high. Conversely, if the number of planted crops is small and the density is low, the land utilization rate is low.

[0154] In one embodiment, the calculated land utilization rate may be too small to meet the requirements due to image deviation, or the land utilization rate may be too small to meet the requirements due to illegal planting by the user. As an example, the method may also include:

[0155] S14. If the land utilization rate is lower than a preset utilization rate, obtain the current time node, and search for historical density according to the current time node.

[0156] S15. If the difference between the object density and the historical density is greater than a preset value, land planning suggestion information is generated and sent to the management terminal for reference by the management personnel.

[0157] Specifically, if the land utilization rate is lower than the preset utilization rate, the current time node is obtained, and the historical density is searched according to the current time node. The historical density may be the density of the planted crops.

[0158] Then, the difference between the object density and the historical density can be calculated. If the difference between the two densities is greater than the preset value, it means that the land utilization rate may be lower than the preset utilization rate due to the user's planting. At this time, land planning suggestion information can be generated and sent to the management terminal for reference by the management personnel. The management terminal can communicate with the cloud system, and specifically can be a terminal used by the management personnel.

[0159] In this embodiment, the embodiment of the present invention provides a method for calculating land utilization, and its beneficial effects are: after determining the target land and the use information of the target land, the present invention can obtain the monitoring image of the target land according to the use information; when the ground attachments of the target land are obtained from the monitoring image and it is determined that the ground attachments are the same as the objects included in the use information, the feature data of the ground attachments in the monitoring image is obtained; and the land utilization rate is calculated according to the use rules included in the use information using the feature data. The present invention recognizes and calculates through real-time detected images, which can not only simplify the calculation process and shorten the calculation processing time to improve the calculation efficiency, but also combines the number of objects and density for comprehensive calculation, which can effectively improve the calculation accuracy.

[0160] The present invention also provides a land utilization rate calculation device, see Figure 2 , showing a schematic structural diagram of a land utilization rate calculation device provided by an embodiment of the present invention.

[0161] Wherein, as an example, the land utilization rate calculation device may include:

[0162] An image acquisition module 201 is used to acquire a monitoring image of the target land according to the usage information after determining the target land and the usage information of the target land;

[0163] The feature acquisition module 202 is used to acquire the ground attachments of the target land from the monitoring image and, when it is determined that the ground attachments are the same as the objects included in the usage information, acquire feature data of the ground attachments in the monitoring image, wherein the feature data includes: the number of objects and the density of objects;

[0164] The utilization rate calculation module 203 is used to calculate the land utilization rate according to the usage rules included in the usage information by using the characteristic data.

[0165] Optionally, the determining that the ground attachment is the same as an object included in the usage information includes:

[0166] Calling a preset recognition model to recognize the ground attachment to obtain recognition information, wherein the recognition information includes: object chromaticity value and object outline;

[0167] According to the object contour, searching for a corresponding purpose object from the purpose information, and obtaining a chromaticity value corresponding to the purpose object to obtain a purpose chromaticity value;

[0168] If the object chromaticity value is the same as the usage chromaticity value, it is determined that the ground attachment is the same as the object included in the usage information;

[0169] If the object chromaticity value is different from the purpose chromaticity value or the corresponding purpose object cannot be found from the purpose information according to the object contour, it is determined that the ground attachment is different from the object included in the purpose information.

[0170] Optionally, acquiring ground attachments of the target land from the monitoring image includes:

[0171] If the monitoring image is one, preprocessing the monitoring image to obtain a processed image;

[0172] A preset CNN neural network is called to identify the processed image to obtain ground attachments, wherein the preset CNN neural network includes a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a first fully connected layer, a second fully connected layer and an output layer connected in sequence.

[0173] Optionally, acquiring ground attachments of the target land from the monitoring image includes:

[0174] If there are multiple surveillance images, each surveillance image is subjected to definition adjustment processing to obtain multiple adjusted images;

[0175] Extracting a view feature from each of the monitoring images to obtain a plurality of two-dimensional view features, wherein the ground attachment is a three-dimensional object;

[0176] After the plurality of two-dimensional view features are classified and combined to obtain the combined features, a classification decision process is performed on the combined features using a preset similarity matching model, and the ground attachments are determined according to the classification categories.

[0177] Optionally, acquiring a monitoring image of the target land according to the usage information includes:

[0178] If the land use type included in the use information is agricultural land, determining whether the horizontal direction of the camera of the real-time shooting device facing the target land is blocked;

[0179] If the camera of the real-time shooting device is facing the horizontal direction of the target land without any obstruction, the camera of the real-time shooting device is called to shoot at the center of the target land to obtain a monitoring image;

[0180] If the camera of the real-time shooting device is blocked in the horizontal direction toward the target land, obtain the horizontal distance between the camera of the real-time shooting device and the center of the target land, and control the camera to rise and shoot toward the center of the target land according to the distance to obtain a monitoring image.

[0181] Optionally, the device further comprises:

[0182] A density acquisition module is used to acquire the current time node and search for the historical density according to the current time node if the land utilization rate is lower than the preset utilization rate;

[0183] The suggestion sending module is used to generate and send land planning suggestion information to the management terminal for reference by the management personnel if the difference between the object density and the historical density is greater than a preset value.

[0184] The present invention also provides a land utilization rate calculation system. Figure 3 , showing a structural schematic diagram of a land utilization rate calculation system provided by an embodiment of the present invention.

[0185] As an example, the land utilization rate calculation system may include: a cloud system and a plurality of real-time shooting devices, and the cloud system is applicable to the land utilization rate calculation method as described in the above embodiment;

[0186] The cloud system is respectively connected to each of the real-time shooting devices for communication.

[0187] Those skilled in the art can clearly understand that, for the sake of convenience and brevity of description, the specific working process of the device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0188] Furthermore, an embodiment of the present application also provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the land utilization rate calculation method as described in the above embodiment is implemented.

[0189] Furthermore, an embodiment of the present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer-executable program, and the computer-executable program is used to enable a computer to execute the land utilization rate calculation method as described in the above embodiment.

[0190] Those skilled in the art will appreciate that the embodiments of the present application may also provide computer program products. Therefore, the present application may adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application may adopt the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program codes.

[0191] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), apparatuses, and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0192] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0193] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1The steps for the functions specified in one or more boxes.

[0194] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for calculating land utilization rate, characterized in that: The method comprises: After determining the target land and the usage information of the target land, obtaining a monitoring image of the target land according to the usage information; When obtaining ground attachments of the target land from the monitoring image and determining that the ground attachments are the same as the objects included in the usage information, obtaining feature data of the ground attachments in the monitoring image, the feature data including: the number of objects and the density of objects; Calculating the land use rate using the characteristic data according to the use rules included in the use information; The calculation formula of the land utilization rate is as follows: C is the land utilization rate; A and B are the calculation weights; P i is the number of objects, P k is the density of the object.

2. The land utilization rate calculation method according to claim 1, characterized in that: The determining that the ground attachment is the same as the object included in the usage information includes: Calling a preset recognition model to recognize the ground attachment to obtain recognition information, wherein the recognition information includes: object chromaticity value and object outline; According to the object contour, searching for a corresponding purpose object from the purpose information, and obtaining a chromaticity value corresponding to the purpose object to obtain a purpose chromaticity value; If the object chromaticity value is the same as the usage chromaticity value, it is determined that the ground attachment is the same as the object included in the usage information; If the object chromaticity value is different from the purpose chromaticity value or the corresponding purpose object cannot be found from the purpose information according to the object contour, it is determined that the ground attachment is different from the object included in the purpose information.

3. The land utilization rate calculation method according to claim 1, characterized in that: The acquiring the ground attachments of the target land from the monitoring image includes: If the monitoring image is one, preprocessing the monitoring image to obtain a processed image; A preset CNN neural network is called to identify the processed image to obtain ground attachments, wherein the preset CNN neural network includes a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a first fully connected layer, a second fully connected layer and an output layer connected in sequence.

4. The land utilization rate calculation method according to claim 1, characterized in that: The acquiring the ground attachments of the target land from the monitoring image includes: If there are multiple surveillance images, each surveillance image is subjected to definition adjustment processing to obtain multiple adjusted images; Extracting a view feature from each of the monitoring images to obtain a plurality of two-dimensional view features, wherein the ground attachment is a three-dimensional object; After the plurality of two-dimensional view features are classified and combined to obtain the combined features, a classification decision process is performed on the combined features using a preset similarity matching model, and the ground attachments are determined according to the classification categories.

5. The land utilization rate calculation method according to claim 1, characterized in that: The step of acquiring a monitoring image of the target land according to the usage information includes: If the land use type included in the use information is agricultural land, determining whether the horizontal direction of the camera of the real-time shooting device facing the target land is blocked; If the camera of the real-time shooting device is facing the horizontal direction of the target land without any obstruction, the camera of the real-time shooting device is called to shoot at the center of the target land to obtain a monitoring image; If the camera of the real-time shooting device is blocked in the horizontal direction toward the target land, obtain the horizontal distance between the camera of the real-time shooting device and the center of the target land, and control the camera to rise and shoot toward the center of the target land according to the distance to obtain a monitoring image.

6. The method for calculating land utilization rate according to claim 1, characterized in that: After the step of calculating the land utilization rate using the characteristic data according to the usage rules included in the usage information, the method further includes: If the land utilization rate is lower than the preset utilization rate, the current time node is obtained, and the historical density is searched according to the current time node; If the difference between the object density and the historical density is greater than a preset value, land planning suggestion information is generated and sent to the management terminal for reference by the management personnel.

7. A land utilization rate calculation device, characterized in that: The device comprises: An image acquisition module is used to acquire a monitoring image of the target land according to the usage information after determining the target land and the usage information of the target land; A feature acquisition module, configured to acquire ground attachments of the target land from the monitoring image and, when it is determined that the ground attachments are the same as the objects included in the usage information, acquire feature data of the ground attachments in the monitoring image, wherein the feature data includes: the number of objects and the density of objects; A utilization rate calculation module, used to calculate the land utilization rate according to the usage rules included in the usage information by using the characteristic data; The calculation formula of the land utilization rate is as follows: C is the land utilization rate; A and B are the calculation weights; P i is the number of objects, P k is the density of the object.

8. A land utilization rate calculation system, characterized in that: The system comprises: a cloud system and a plurality of real-time shooting devices, wherein the cloud system is applicable to the land utilization rate calculation method according to any one of claims 1 to 6; The cloud system is respectively connected to each of the real-time shooting devices for communication.

9. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the land utilization rate calculation method as described in any one of claims 1 to 6 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer-executable program, and the computer-executable program is used to enable a computer to execute the land utilization rate calculation method according to any one of claims 1 to 6.