Method, device, computer equipment and storage medium for detecting buildings in farmland

Abstract

The present disclosure relates to the field of building monitoring technology in farmland areas, and discloses a method, device, computer equipment and storage medium for detecting buildings in farmland. The method comprises: obtaining multiple remote sensing images of target farmland within a preset historical time period, wherein the preset historical time period includes multiple phenological periods of vegetation in the target farmland; determining historical ground object information of the target farmland based on the multiple remote sensing images; wherein the historical ground object information comprises: historical ground object types corresponding to each block of the target farmland in each remote sensing image; determining whether there are buildings in the target farmland based on the historical ground object information of the target farmland. The efficiency and accuracy of building detection in farmland are improved, and the monitoring cost is reduced.

Description

Technical Field

[0001] The present disclosure relates to the technical field of building detection in farmland, and in particular to a method, device, computer equipment and storage medium for building detection in farmland. Background Art

[0002] In the current context of global food security and environmental sustainability, the rational use and protection of basic farmland has become particularly important. Basic farmland is the foundation of agricultural production, and its quality directly affects the yield of crops. Therefore, scientific and accurate monitoring of basic farmland will not only help optimize agricultural management, but also improve food production capacity.

[0003] One of the challenges facing the protection of basic farmland is the challenge of converting basic farmland into construction land. Therefore, in order to prevent basic farmland from being converted into construction land, the inspection of buildings within basic farmland is an important means to implement the basic farmland protection system and ensure that the red line is not crossed.

[0004] However, the current detection of buildings in farmland often faces the following difficulties:

[0005] Adopting the method of ground survey and manual observation: basic farmland monitoring involves a large area. At the same time, there is a lack of automated extraction tools for the extraction of buildings in basic farmland. Most of the work is obtained through manual delineation, which is not only time-consuming and laborious, but also difficult to obtain data quickly.

[0006] The method of using a certain machine: generally, neural networks, object-oriented, support vector machines and other methods are used to extract the type of land objects from remote sensing data collected in the same period. However, the temporal resolution of remote sensing data in the same period often cannot meet the monitoring needs, and it is difficult to use phenological period information. At the same time, if the above machine learning model wants to have better accuracy, it requires a large amount of relatively accurate training data, and accurate training data requires manual annotation of a large number of remote sensing images, which has high labor costs and is difficult to annotate a large amount of training data. Summary of the invention

[0007] In order to solve the above technical problems, the present disclosure provides a method, device, computer equipment and storage medium for detecting buildings in farmland.

[0008] In one aspect, the present disclosure provides a method for detecting buildings in a farmland, the method comprising:

[0009] Acquire multiple remote sensing images of the target farmland within a preset historical time period, wherein the preset historical time period includes multiple phenological periods of vegetation in the target farmland;

[0010] Based on the multiple remote sensing images, determine the historical feature information of the target farmland; wherein the historical feature information includes: the historical feature type corresponding to each block of the target farmland in each of the remote sensing images;

[0011] Based on the historical feature information of the target farmland, it is determined whether there is a building in the target farmland.

[0012] In the above scheme, determining the historical feature information of the target farmland based on the multiple remote sensing images includes:

[0013] Determine the type of ground objects at each pixel of each remote sensing image;

[0014] Based on the ground object type of each pixel point of all the remote sensing images, the historical ground object information of the target farmland is determined.

[0015] In the above solution, the determining of the ground object type of each pixel point of each remote sensing image includes:

[0016] Determining a normalized vegetation index, a normalized water index, and a normalized building index for each pixel of each remote sensing image;

[0017] Based on the correspondence between the preset normalized vegetation index, the normalized water index and the normalized building index and the ground object type, the ground object type of the pixel point is determined.

[0018] In the above solution, the determining of the ground object type of each pixel point of each remote sensing image includes:

[0019] Determining a normalized vegetation index for each pixel of each of the remote sensing images, and determining vegetation pixels and non-vegetation pixels of each of the remote sensing images based on the normalized vegetation index;

[0020] Determine a normalized water index for each of the non-vegetation pixels in each of the remote sensing images, and determine water pixels and non-water pixels in each of the remote sensing images based on the normalized water index;

[0021] A normalized building index is determined for each non-water body pixel of each remote sensing image, and building pixels and non-building pixels of each remote sensing image are determined based on the normalized building index.

[0022] In the above scheme, the method further includes one or a combination of the following:

[0023] A correspondence between one or a combination of the normalized vegetation index, the normalized water index and the normalized building index of a pixel point and the ground feature type of the pixel point is preset.

[0024] In the above scheme, the determining of the historical ground object information of the target farmland based on the ground object type of each pixel point of all the remote sensing images includes:

[0025] Based on the ground object type of each pixel point of all the remote sensing images, each of the remote sensing images is divided into plots according to a preset division rule; wherein the preset division rule includes: the historical ground object type of the pixel points of the same plot is the same;

[0026] Based on the plot division result of the remote sensing image, the historical feature type corresponding to each plot is determined.

[0027] In the above solution, determining whether there is a building in the target farmland based on the historical feature information of the target farmland includes:

[0028] When the historical feature information of the target farmland does not include a building, determining that there is no building in the target farmland;

[0029] If the historical feature information of at least one of the plots includes buildings and at least one non-building type other than buildings, it is determined whether there are buildings in the target farmland based on the number of buildings in the historical feature types of the plot; and / or it is determined whether there are buildings in the target farmland based on the time sequence of all historical feature types of the plot.

[0030] In another aspect, the present disclosure provides a device for detecting buildings in a farmland, the device comprising:

[0031] A remote sensing data acquisition module, used to acquire multiple remote sensing images of the target farmland within a preset historical time period, wherein the preset historical time period includes multiple phenological periods of vegetation in the target farmland;

[0032] A ground feature information determination module, used to determine the historical ground feature information of the target farmland based on the multiple remote sensing images; wherein the historical ground feature information includes: the historical ground feature type corresponding to each block of the target farmland in each of the remote sensing images;

[0033] The detection result output module is used to determine whether there are buildings in the target farmland based on the historical ground feature information of the target farmland.

[0034] In yet another aspect, the present disclosure provides an electronic device, comprising a processor and a memory;

[0035] The processor is used to execute the steps of the method described above by calling the program or instruction stored in the memory.

[0036] In still another aspect, the present disclosure provides a determination machine readable storage medium, wherein the determination machine readable storage medium stores a program or instruction, wherein the program or instruction enables the determination machine to execute the steps of the above-mentioned method.

[0037] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:

[0038] In the scheme implemented by the above-mentioned target farmland building detection method, device, computer equipment and storage medium, multiple remote sensing images of multiple phenological periods of vegetation in the target farmland are obtained, and then based on the multiple remote sensing images, the historical feature information of the target farmland is determined, and finally, based on the historical feature information of the target farmland, it is determined whether there are buildings in the target farmland. In the present invention, in order to address the problems of insufficient temporal resolution and insufficient training data of remote sensing images of target farmland, remote sensing images covering multiple phenological periods of vegetation in the target farmland are obtained, thereby improving the temporal resolution of remote sensing images of the target farmland, and then the historical feature types contained in the target farmland at different times are obtained based on the remote sensing images of the target farmland at different times obtained at different times, and finally, based on the historical feature types of the target farmland at different times, it is determined whether there are buildings in the target farmland. Therefore, the phenological period information of vegetation is fully referenced. Since the remote sensing images of the target farmland in different phenological periods of vegetation are very different, the type of land objects in the target farmland can be determined by comprehensively referring to the remote sensing images of the target farmland at different times. This can effectively avoid misjudgment caused by insufficient farmland information contained in the remote sensing images, thereby improving the accuracy of judging whether there are buildings in the target farmland. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0041] Figure 1 Schematic diagram of an application environment of a method for detecting buildings in farmland according to an embodiment of the present invention;

[0042] Figure 2 is a schematic diagram of a flow chart of a method for detecting buildings in farmland in one embodiment of the present invention;

[0043] Figure 3 yes Figure 1 A schematic flow chart of a specific implementation of step S22;

[0044] Figure 4 yes Figure 3 A schematic flow chart of a specific implementation of step S221;

[0045] Figure 5 yes Figure 1 Another specific implementation flow diagram of step S221;

[0046] Figure 6 is a schematic diagram of a structure of a device for detecting buildings in farmland according to an embodiment of the present invention;

[0047] Figure 7 It is a structural diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments and features of the embodiments of the present disclosure can be combined with each other without conflict.

[0049] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments of the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0050] The method for detecting buildings in farmland provided by the embodiment of the present invention can be applied in the following aspects: Figure 1In the application environment, the client communicates with the server through the network. The server can receive a confirmation request from the user on whether there is a building in the target farmland through the client, and obtain multiple remote sensing images of the target farmland in a preset historical time period according to the request, and determine the historical feature information of the target farmland based on the multiple remote sensing images; wherein the historical feature information includes: the historical feature types corresponding to each block of the target farmland in each remote sensing image; based on the historical feature information of the target farmland, determine whether there is a building in the target farmland. In the present invention, for the problem of detecting buildings in farmland, firstly, remote sensing images of the target farmland in different periods are obtained, and the different periods cover the phenological period of the vegetation in the target farmland. Therefore, the remote sensing images contain the phenological information of the vegetation. Based on the multiple remote sensing images, the feature information (such as the feature type) of the same block of the target farmland in different periods can be obtained, so that the feature information of the same block in different periods can be cross-validated, and the comprehensive judgment of whether the block has a building can be made, thereby avoiding the misjudgment of whether the block has a building, and then improving the accuracy of the detection of whether there is a building in the farmland. The client may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices. The server may be implemented by an independent server or a server cluster consisting of multiple servers. The present invention is described in detail below through specific embodiments.

[0051] See also Figure 2 As shown, Figure 2 A flow chart of a method for detecting buildings in farmland provided by an embodiment of the present invention includes the following steps:

[0052] S21. Acquire a plurality of remote sensing images of the target farmland within a preset historical time period, wherein the preset historical time period includes a plurality of phenological periods of vegetation of the target farmland.

[0053] It can be understood that the target farmland is the basic farmland to be detected whether there are buildings in it. The target farmland is monitored by remote sensing image acquisition equipment, that is, the remote sensing image of the target farmland is acquired. After the remote sensing image acquisition equipment acquires the remote sensing image of the target farmland, the acquired remote sensing image is preprocessed. The preprocessing may include cropping the remote sensing image, cropping out part of the image outside the target farmland, so that the remote sensing image only contains the target farmland, thereby avoiding the image of the detection result of whether there are buildings in the target farmland by the ground objects outside the target farmland area in the remote sensing image, and also reducing the computational cost of subsequent image processing.

[0054] Here, the remote sensing image acquisition device may acquire remote sensing images of the target farmland at a fixed acquisition cycle, or may acquire remote sensing images of the target farmland according to the acquisition time preset by the user, for example, acquiring remote sensing images of the target farmland at a specified phenological period of the vegetation of the target farmland. The historical time period may be a growth cycle of the vegetation or a time period preset by the user, such as a natural year. The remote sensing images of the target farmland may be remote sensing images of the target farmland at different phenological periods, or may be within a historical time period preset by the user.

[0055] Specifically, in one scenario, the preset historical time period may be a natural year, and the natural year may be the previous year of the current time. For example, if the current year is 2024, the preset historical time period may be 2023. Of course, it may also be a natural year before 2023 (such as 2022). However, the closer the time is, the more accurate the monitoring result of whether there are buildings in the target farmland is. Therefore, the preset historical time period is preferably the most recent year. The remote sensing image acquisition device may acquire remote sensing images of the target farmland in a preset time period. The multiple remote sensing images may be acquired in a natural month. At least one most representative remote sensing image (such as the one with the highest resolution) is selected from the multiple remote sensing images acquired in each natural month as the remote sensing image of each natural month, that is, the multiple remote sensing images acquired include remote sensing images acquired in each of the 12 natural months of the natural year.

[0056] Specifically, in another scenario, taking the case where the vegetation in the target farmland is wheat crops, the phenological periods of wheat include: wintering period, greening period, rising and jointing period, heading period, flowering period, grain filling period, milky period, waxy period, etc. You can choose remote sensing images of the target farmland collected when the wheat is in multiple different phenological periods. It is best to choose phenological periods in which wheat presents different growth forms, for example, including wintering areas, rising and jointing periods, and heading periods. Of course, the phenological period of wheat can be determined based on past experience, and multiple remote sensing images can be collected in each phenological period. Then, one of the multiple remote sensing images corresponding to each phenological period is used as the representative remote sensing image corresponding to each phenological period.

[0057] Of course, the acquisition cycle of the remote sensing image of the target farmland, that is, the time granularity of remote sensing image acquisition, can be determined according to the actual situation. In principle, the finer the time granularity, the higher the accuracy of the final detection results of buildings in the farmland. However, the computational cost will also be higher. Therefore, it is necessary to balance the time granularity and computational cost.

[0058] In this way, the historical time period covers multiple phenological periods of vegetation because the remote sensing images of the target farmland are very different at different growth stages of the vegetation, and thus the land feature types of the farmland judged based on the remote sensing images may also be different. Therefore, the land feature information (land feature type) of the farmland can be judged based on the remote sensing images of the target farmland at different times, and then the land feature information of the target farmland obtained based on all the remote sensing images can be cross-judged, thereby increasing the accuracy of building detection in the target farmland.

[0059] Of course, further, in some embodiments, the preprocessing of remote sensing images may also include: atmospheric correction, orthorectification, image fusion, etc., wherein atmospheric correction means that the total radiance of the ground target finally measured by the sensor is not a reflection of the true reflectivity of the surface, and includes radiation errors caused by atmospheric absorption, especially scattering. Atmospheric correction is the process of eliminating these radiation errors caused by atmospheric influences and inverting the true surface reflectivity of the ground object. Orthophoto production is generally achieved by selecting some ground control points on the film, and using the digital elevation model (DEM) data within the range of the film that has been acquired, to simultaneously perform tilt correction and projection error correction on the image, and resample the image into an orthophoto. Image fusion; remote sensing image fusion is to calculate and process multi-source remote sensing data that are redundant or complementary in space, time, and spectrum according to certain rules (or algorithms) to obtain more accurate and richer information than any single data, and generate synthetic image data with new spatial, spectral, and temporal characteristics.

[0060] S22. Determine historical feature information of the target farmland based on multiple remote sensing images; wherein the historical feature information includes: historical feature types corresponding to each block of the target farmland in each remote sensing image.

[0061] It is understandable that within a preset historical time period, the type of land object of the target farmland may change over time, so that the type of land object corresponding to the same plot of the target farmland in the remote sensing images collected by each remote sensing image acquisition device may also be different. Determine the type of land object corresponding to each plot in each remote sensing image, and then integrate the land object types of the same plot in multiple remote sensing images to obtain the historical land object type of the plot. Finally, integrate the historical land object types of all plots to obtain the historical land object information of the target farmland. The land object type here can be a specific fixed object type or a classification to which the fixed object belongs. For example, through remote sensing images, it is identified that there is vegetation on it, and other plots other than the vegetation plot cannot be specifically judged. At this time, the land object type of the remote sensing image can be vegetation and non-vegetation. Specifically, the land object type corresponding to each plot of the target farmland can be vegetation, water, bare land, buildings, etc.

[0062] In this embodiment, since the plant crops in the target farmland have their own growth rules, they show different forms in different growth periods, which may cause misjudgment of the type of farmland objects. For example, in the sowing period of wheat crops, wheat has not grown out of the ground. At this time, the object type of the target farmland obtained according to the remote sensing image is most likely non-vegetation or bare land, that is, it is most likely not vegetation. Therefore, in this embodiment, in order to avoid the above-mentioned misjudgment problem, the historical object information of the target farmland is obtained by using the remote sensing images of the target farmland at different times, so that the object types of different plots at different times can be obtained, and then these object types are integrated for cross-comparison to determine the object type of the plot. In this way, the acquisition timeline of the remote sensing image of the target farmland is lengthened, and the remote sensing images of the target farmland when presenting different object forms can be obtained, which makes full use of the object information of the target farmland at different growth periods of crops and increases the accuracy of the building detection results in the target farmland.

[0063] In some embodiments, Figure 3 As shown, the above S22 may include:

[0064] S221, determining the ground object type of each pixel point of each remote sensing image;

[0065] S222. Determine the historical ground feature information of the target farmland based on the ground feature type of each pixel point of all remote sensing images.

[0066] Among them, the land object type of each pixel point of each remote sensing image is the land object type corresponding to the target farmland of each pixel point on the remote sensing image. The adjacent pixels of the same land object type on each remote sensing image can be divided into a pixel area, and each pixel area is the remote sensing image of a plot of the target farmland. Therefore, by integrating the pixels of the same land object type on all remote sensing images, the plot division result of the target farmland can be obtained, and the historical land object type corresponding to each plot (that is, the land object type corresponding to the plot on each remote sensing image) can also be obtained.

[0067] In some embodiments, Figure 4 As shown, the above S221 may include:

[0068] S221a1, determining a normalized vegetation index, a normalized water index and a normalized building index for each pixel point of each remote sensing image;

[0069] S221a2, based on the correspondence between the pre-set normalized vegetation index, normalized water index and normalized building index and the ground object type, determine the ground object type of each pixel point of each remote sensing image.

[0070] Among them, the normalized vegetation index is used to determine the possibility that the pixel point of the remote sensing image is vegetation. The larger the normalized vegetation index is, the greater the possibility that the target farmland of the pixel point is vegetation. On the contrary, the smaller it is. The normalized water body index is used to determine the possibility that the pixel point of the remote sensing image is a water body. The larger the normalized water body index is, the greater the possibility that the target farmland of the pixel point is a water body. On the contrary, the smaller it is. The normalized building index is used to determine the possibility that the pixel point of the remote sensing image is a building. The larger the normalized building index is, the greater the possibility that the target farmland of the pixel point is a building. On the contrary, the smaller it is.

[0071] It is understandable that the common types of land features in basic farmland are: water area, bare land, forest land, buildings (construction land), crop land and grassland, among which the plots of forest land, crop land and grassland are all vegetation plots, the plots of water area, bare land and buildings are all non-vegetation plots, the plots of water bodies are water body plots, and the plots of bare land and buildings are non-water body plots. The vegetation plots and non-vegetation plots in the remote sensing image can be determined according to the normalized vegetation index, the water body plots and non-water body plots in the remote sensing image can be determined according to the normalized water body index, and the building plots and non-building plots in the remote sensing image can be determined according to the normalized building index.

[0072] In this embodiment, the correspondence between the normalized vegetation index, the normalized water index and the normalized building index of the pixel point and the ground object type of the pixel point is preset. Specifically, first, the remote sensing indexes of multiple template plots are selected, and the multiple template plots correspond to different ground object types, such as crop plots, building plots, forest plots, water body plots, grassland plots, bare land, etc. Then, the remote sensing index of the remote sensing images of multiple template plots in different phenological periods is determined, where the remote sensing index includes the normalized vegetation index, the normalized water body index and the normalized building index; finally, the remote sensing index is statistically analyzed to obtain the value range and mean of each remote sensing index in different phenological periods of multiple template plots, and the threshold of the corresponding remote sensing index for different types of land objects is determined based on the value range and mean. For example, if the normalized vegetation index is greater than 0.2, it is vegetation, otherwise it is non-vegetation; if the normalized water body index is greater than 0.3, it is a water body, otherwise it is a non-water body; if the normalized building index is greater than 0.6, it is a building, otherwise it is a non-building.

[0073] In some embodiments, Figure 5 As shown, the above S221 may include:

[0074] S221b1, determining a normalized vegetation index for each pixel point of each remote sensing image, and determining vegetation pixels and non-vegetation pixels of each remote sensing image based on the normalized vegetation index;

[0075] S221b2, determining a normalized water index for each non-vegetation pixel of each remote sensing image, and determining water pixel points and non-water pixel points of each remote sensing image based on the normalized water index;

[0076] S221b3, determining a normalized building index for each non-water body pixel of each remote sensing image, and determining building pixels and non-building pixels of each remote sensing image based on the normalized building index.

[0077] Among them, the vegetation pixels and non-vegetation pixels in the pixels are determined according to the normalized vegetation index; the non-vegetation in the farmland generally includes buildings, bare land and water bodies. Since the normalized water body index has an extremely high recognition rate for water bodies, the range of water bodies confirmed by the normalized water body index can be directly confirmed. Therefore, the water bodies are eliminated through the normalized water body index. Finally, the buildings and bare land are determined according to the normalized building index.

[0078] In this embodiment, for pixels that have been determined to be vegetation, other remote sensing indices are not calculated, reducing the computational cost of pixel remote sensing index calculation. In this embodiment, a decision tree classification model can be selected as a ground object type extraction model for determining the ground object type of a remote sensing image.

[0079] In some embodiments, the above S222 may include:

[0080] S2221, based on the ground object type of each pixel point of all remote sensing images, dividing each remote sensing image into plots according to a preset division rule; wherein the preset division rule includes: the historical ground object type of the pixel points of the same plot is the same;

[0081] S2222. Determine the type of historical features corresponding to each plot based on the plot division results of the remote sensing image.

[0082] It is understandable that the pixel points whose land object type on a remote sensing image is vegetation are divided into the first plot, however, the land object types of the pixel points contained in the same position as the first plot on other remote sensing images may not all be the same. Therefore, the plot division on the remote sensing image should comprehensively refer to the land object types of the pixel points on all remote sensing images, rather than being determined based on the plot division on one of the remote sensing images. Specifically, the land object types of each pixel point of the remote sensing image in each remote sensing image are sorted according to the acquisition time of the remote sensing image to obtain the historical land object type time sequence corresponding to the pixel point; based on the historical land object type sequence corresponding to the pixel point, each remote sensing image is divided into plots, and after the division, the historical land object type time sequence of the pixel points in the same plot on each remote sensing image is the same.

[0083] In this embodiment, the type of land object in the remote sensing image is determined based on the remote sensing index of the remote sensing image, which reduces the difficulty of identifying the type of land object, fully utilizes the remote sensing data, realizes the automatic extraction of buildings in basic farmland, reduces the manual burden, and improves the efficiency and accuracy of building detection in farmland.

[0084] In one embodiment, first, the reflectance value of the near infrared band, the reflectance value of the red light band, the reflectance of the short-wave infrared band, and the reflectance of the green wave band of the remote sensing image of the target farmland are obtained according to the remote sensing image. Then, the above remote sensing data are brought into the normalized vegetation index formula, the normalized water index formula, and the normalized building index formula to determine the normalized vegetation index, the normalized water index, and the normalized building index.

[0085] Among them, the normalized vegetation index NDVI = (NIR-R) / (NIR+R), where NIR is the reflectance value of the near-infrared band and R is the reflectance value of the red light band;

[0086] Normalized difference building index NDBI = NDBI = (SWIR-NIR) / (SWIR+NIR), where SWIR is the reflectivity of the short-wave infrared band and NIR is the reflectivity of the near-infrared band;

[0087] Normalized water index NDWI = (p(Green)-p(NIR)) / (p(Green)+p(NIR)), where Green is the reflectance of the green wave band and NIR is the reflectance of the near infrared band.

[0088] S23. Determine whether there are any buildings in the target farmland based on the historical land feature information of the target farmland.

[0089] Among them, whether there is ground feature information in the target farmland is determined by the ground feature type (historical ground feature type) corresponding to the target farmland in each remote sensing image. When there is no building in the ground feature type of the target farmland in all remote sensing images, it is determined that there is no building in the target farmland. When there is a building in the historical ground feature type of the target farmland in at least one remote sensing image, it is necessary to refer to the historical ground feature type of the land parcel whose ground feature type is a building to determine whether the land parcel has a building.

[0090] In some embodiments, the above S23 may include:

[0091] S231. When the historical feature information of the target farmland does not include buildings, determine that there are no buildings in the target farmland.

[0092] S232. When the historical feature information of at least one plot of land includes buildings and at least one non-building type other than buildings, determine whether there are buildings in the target farmland based on the number of buildings in the historical feature types of the plot; and / or determine whether there are buildings in the target farmland based on the time series of all historical feature types of the plot.

[0093] It is understandable that when all historical feature types of at least one plot are buildings, it can be determined that the feature type of the plot is a building, that is, there are buildings in the target farmland. When there are buildings and non-buildings in the historical feature types of at least one plot, the specific historical feature type needs to be analyzed. Here, according to the ratio of the number of buildings and non-buildings in the historical feature types, for example, a remote sensing image of each natural month of a natural year of the target farmland is obtained, so that each plot of the target farmland has 12 remote sensing images, that is, 12 historical feature types, if 11 historical feature types are buildings, it is determined that the plot is a building. If only one historical feature type is a building, it is determined that the plot is not a building. If there are 3 historical feature types that are buildings and the other historical feature types are buildings, it can be determined whether the plot is a building based on the time sequence of all historical feature types of the plot. For example, there is a building plot in the remote sensing image that is relatively early in time, but the feature type of the building plot in the remote sensing image that is relatively late in time is vegetation, then it can be basically determined that the plot is vegetation.

[0094] It can be seen that in the above scheme, in order to solve the problem of insufficient temporal resolution of remote sensing images of target farmland and insufficient training data, remote sensing images covering multiple phenological periods of vegetation in target farmland are obtained, thereby improving the temporal resolution of remote sensing images of target farmland. Then, the historical land object types contained in the target farmland at different times are obtained according to the remote sensing images of the target farmland at different times obtained at different times. Finally, according to the historical land object types of the target farmland at different times, it is judged whether there are buildings in the target farmland. Thus, the phenological period information of vegetation is fully referenced. Since the remote sensing images of target farmland at different phenological periods of vegetation are very different, therefore, by comprehensively referring to the remote sensing images of target farmland at different times, the land object type of target farmland is determined, which can effectively avoid the misjudgment caused by the insufficient farmland information contained in the remote sensing images, thereby improving the accuracy of judging whether there are buildings in the target farmland.

[0095] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0096] In one embodiment, a device for detecting buildings in farmland is provided, and the device for detecting buildings in farmland corresponds to the method for detecting buildings in farmland in the above embodiment. Figure 6 As shown, the farmland building detection device includes a remote sensing data acquisition module 601, a ground object information determination module 602 and a detection result output module 603. The detailed description of each functional module is as follows:

[0097] A remote sensing data acquisition module is used to acquire multiple remote sensing images of the target farmland within a preset historical time period, wherein the preset historical time period includes multiple phenological periods of vegetation in the target farmland;

[0098] A ground feature information determination module is used to determine the historical ground feature information of the target farmland based on multiple remote sensing images; wherein the historical ground feature information includes: the historical ground feature types corresponding to each block of the target farmland in each remote sensing image;

[0099] The detection result output module is used to determine whether there are buildings in the target farmland based on the historical ground feature information of the target farmland.

[0100] In one embodiment, the ground feature information determination module 602 is specifically configured to:

[0101] Determine the type of ground objects at each pixel of each remote sensing image;

[0102] Based on the ground object type of each pixel point in all remote sensing images, the historical ground object information of the target farmland is determined.

[0103] In one embodiment, the ground feature information determination module 602 is specifically configured to:

[0104] Determine the normalized vegetation index, normalized water index and normalized building index of each pixel point of each remote sensing image;

[0105] Based on the correspondence between the preset normalized vegetation index, normalized water index and normalized building index and the ground object type, the ground object type of each pixel point in each remote sensing image is determined.

[0106] In one embodiment, the ground feature information determination module 602 is specifically configured to:

[0107] Determine the normalized vegetation index of each pixel point of each remote sensing image, and determine the vegetation pixel points and non-vegetation pixel points of each remote sensing image based on the normalized vegetation index;

[0108] Determine a normalized water index for each non-vegetation pixel of each remote sensing image, and determine water pixel points and non-water pixel points of each remote sensing image based on the normalized water index;

[0109] The normalized building index of each non-water body pixel of each remote sensing image is determined, and the building pixel points and non-building pixel points of each remote sensing image are determined based on the normalized building index.

[0110] In one embodiment, the above-mentioned building detection device in farmland also includes a mapping relationship table module 104, which is used to pre-set the correspondence between the normalized vegetation index, normalized water index and normalized building index of the pixel point or a combination of the three, and the ground feature type of the pixel point.

[0111] In one embodiment, the ground feature information determination module 602 is specifically configured to:

[0112] Based on the ground object type of each pixel point of all remote sensing images, each remote sensing image is divided into plots according to a preset division rule; wherein the preset division rule includes: the historical ground object type of the pixel points of the same plot is the same;

[0113] Based on the plot division results of remote sensing images, the historical landform type corresponding to each plot is determined.

[0114] In one embodiment, the detection result output module 603 is specifically used to:

[0115] If the historical feature information of the target farmland does not include buildings, it is determined that there are no buildings in the target farmland;

[0116] If the historical feature information of at least one plot of land includes buildings and at least one non-building type other than buildings, determine whether there are buildings in the target farmland based on the number of buildings in the historical feature types of the plot; and / or determine whether there are buildings in the target farmland based on the time sequence of all historical feature types of the plot.

[0117] The present invention provides a device for detecting buildings in farmland, which first obtains multiple remote sensing images of a target farmland within a preset historical time period, then determines the historical ground object information of the target farmland based on the multiple remote sensing images, and finally determines whether there are buildings in the target farmland based on the historical ground object information of the target farmland. The ground object type of the target farmland is determined by comprehensively referring to the remote sensing images of the target farmland at different times, thereby improving the accuracy of judging whether there are buildings in the target farmland.

[0118] For the specific definition of the farmland building detection device, please refer to the definition of the farmland building detection method above, which will not be repeated here. Each module in the above-mentioned farmland building detection device can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0119] Figure 7 The structure diagram of an electronic device provided by an embodiment of the present invention is shown. The electronic device provided by the embodiment of the present disclosure may specifically be a server. The electronic device may execute the material display method provided by the above embodiment.

[0120] It should be noted that Figure 7 The electronic device 700 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0121] The electronic device 700 conventionally includes a processor 710 and a computer program product or computer-readable medium in the form of a memory 720. The memory 720 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk or a ROM. The memory 720 has a storage space 721 for executable instructions (or program codes) 7211 for executing any method steps in the above-mentioned method for detecting buildings in farmland. For example, the storage space 721 for executable instructions may include individual executable instructions 7211 for respectively implementing various steps in the above-mentioned method for detecting buildings in farmland. These executable instructions may be read from or written to one or more computer program products. These computer program products include program code carriers such as a hard disk, a compact disk (CD), a memory card or a floppy disk. Such a computer program product is typically a portable or fixed storage unit. The storage unit may have a storage device having a storage medium ... Figure 4 The memory 720 in the electronic device 700 is similarly arranged as a storage segment or storage space. The executable instructions may be compressed, for example, in an appropriate form. Generally, the storage unit includes executable instructions for executing the steps of the method for detecting buildings in a farmland according to the present invention, that is, codes that can be read by a processor such as the processor 710, and when these codes are executed by the electronic device 700, the electronic device 700 executes the various steps of the method for detecting buildings in a farmland described above.

[0122] Of course, to simplify, Figure 4Only some of the components related to the present invention in the electronic device 700 are shown, and components such as a bus, an input / output interface, an input device, an output device, etc. are omitted. In addition, according to specific application conditions, the electronic device 700 may also include any other appropriate components.

[0123] In addition, the embodiment of the present disclosure further provides a server, which is used to execute the local material creation method provided in the above embodiment. The architecture of the server is similar to the architecture of the above electronic device, and will not be repeated here.

[0124] An embodiment of the present invention further provides a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the material display method or the local material creation method provided in each embodiment of the present invention.

[0125] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0126] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0127] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant descriptions on the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0128] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0129] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting buildings in farmland, characterized in that: The method comprises: Acquire multiple remote sensing images of the target farmland within a preset historical time period, wherein the preset historical time period includes multiple phenological periods of vegetation in the target farmland; Based on the multiple remote sensing images, determine the historical feature information of the target farmland; wherein the historical feature information includes: the historical feature type corresponding to each block of the target farmland in each of the remote sensing images; Based on the historical feature information of the target farmland, determining whether there are buildings in the target farmland; The determining the historical feature information of the target farmland based on the plurality of remote sensing images includes: Determine the type of ground objects at each pixel of each remote sensing image; Determining the historical ground object information of the target farmland based on the ground object type of each pixel point of all the remote sensing images; Determining the ground object type of each pixel point of each remote sensing image includes: Determining a normalized vegetation index, a normalized water index, and a normalized building index for each pixel of each remote sensing image; Based on the correspondence between the preset normalized vegetation index, the normalized water index and the normalized building index and the ground object type, the ground object type of the pixel point is determined.

2. The method for detecting buildings in farmland according to claim 1, characterized in that: Determining the ground object type of each pixel point of each remote sensing image includes: Determining a normalized vegetation index for each pixel of each of the remote sensing images, and determining vegetation pixels and non-vegetation pixels of each of the remote sensing images based on the normalized vegetation index; Determine a normalized water index for each of the non-vegetation pixels in each of the remote sensing images, and determine water pixels and non-water pixels in each of the remote sensing images based on the normalized water index; A normalized building index is determined for each non-water body pixel of each remote sensing image, and building pixels and non-building pixels of each remote sensing image are determined based on the normalized building index.

3. The method for detecting buildings in farmland according to claim 1 or 2, characterized in that: The method further comprises one or a combination of the following: A correspondence between one or a combination of the normalized vegetation index, the normalized water index and the normalized building index of a pixel point and the ground feature type of the pixel point is preset.

4. The method for detecting buildings in farmland according to claim 1, characterized in that: The determining of the historical ground object information of the target farmland based on the ground object type of each pixel point of all the remote sensing images includes: Based on the ground object type of each pixel point of all the remote sensing images, each of the remote sensing images is divided into plots according to a preset division rule; wherein the preset division rule includes: the historical ground object type of the pixel points of the same plot is the same; Based on the plot division result of the remote sensing image, the historical feature type corresponding to each plot is determined.

5. The method for detecting buildings in farmland according to claim 1, characterized in that: The determining whether there is a building in the target farmland based on the historical feature information of the target farmland includes: When the historical feature information of the target farmland does not include a building, determining that there is no building in the target farmland; If the historical feature information of at least one of the plots includes buildings and at least one non-building type other than buildings, it is determined whether there are buildings in the target farmland based on the number of buildings in the historical feature types of the plot; and / or it is determined whether there are buildings in the target farmland based on the time sequence of all historical feature types of the plot.

6. A device for detecting buildings in farmland, characterized in that: The device comprises: A remote sensing data acquisition module, used to acquire multiple remote sensing images of the target farmland within a preset historical time period, wherein the preset historical time period includes multiple phenological periods of vegetation in the target farmland; A ground feature information determination module, used to determine the historical ground feature information of the target farmland based on the multiple remote sensing images; wherein the historical ground feature information includes: the historical ground feature type corresponding to each block of the target farmland in each of the remote sensing images; A detection result output module, used for determining whether there are buildings in the target farmland based on the historical ground feature information of the target farmland; The ground feature information determination module is specifically used for: Determine the type of ground objects at each pixel of each remote sensing image; Determining the historical ground object information of the target farmland based on the ground object type of each pixel point of all the remote sensing images; The ground feature information determination module is specifically used for: Determining a normalized vegetation index, a normalized water index, and a normalized building index for each pixel of each remote sensing image; Based on the correspondence between the preset normalized vegetation index, the normalized water index and the normalized building index and the ground object type, the ground object type of the pixel point is determined.

7. A computer device comprising a memory and a processor, wherein the memory stores machine-readable instructions, and when the machine-readable instructions are executed by the processor, the processor executes the steps of the method for detecting buildings in farmland as claimed in any one of claims 1 to 5.

8. A storage medium storing machine-readable instructions, wherein when the machine-readable instructions are executed by one or more processors, the one or more processors execute the steps of the method for detecting buildings in a farmland as claimed in any one of claims 1 to 5.

Images