A method and apparatus for extracting a ploughed field image

Abstract

The application discloses a kind of farmland image extraction method and equipment, method includes: determining the remote sensing image of to-be-measured area, obtain the basic information of remote sensing image, basic information at least includes time information, spatial information;Determine the multiple identification models pre-stored in database, respectively obtain the matching degree between remote sensing image and multiple identification models according to the basic information of remote sensing image;By edge detection analysis to the remote sensing image of to-be-measured area, the remote sensing image of to-be-measured area is divided into several pending areas;Select the highest identification model of matching degree to identify the land type of several pending areas, generate the image label corresponding to the land type of several pending areas;According to image label, determine whether pending area is cultivated area;If yes, extract the remote sensing image corresponding to pending area as the cultivated land image of to-be-measured area.Make the process of extracting cultivated land image through remote sensing image more detailed, improve accuracy and precision.

Description

Technical Field

[0001] This application relates to the field of farmland protection, specifically to a method and device for extracting farmland images. Background Technology

[0002] Farmland protection refers to the protection of the quantity and quality of arable land through legal, economic, and technological means and measures. A significant reduction in arable land area directly threatens agricultural development. Therefore, we must ensure a certain quantity and quality of arable land. To utilize arable land resources rationally and effectively, we must first utilize remote sensing technology to revise the overall land use plan and conduct a comprehensive survey of the quantity and quality of arable land. However, the accuracy of arable land image extraction using remote sensing technology is currently low, and misjudgments and omissions are easily made. Therefore, there is an urgent need for a method that can accurately, scientifically, and precisely extract arable land images from remote sensing images. Summary of the Invention

[0003] To address the aforementioned problems, this application proposes a method and apparatus for extracting farmland images. The method includes:

[0004] A remote sensing image of the area to be tested is determined, and basic information of the remote sensing image is obtained, including at least temporal and spatial information. Multiple recognition models pre-stored in a database are identified, and the matching degree between the remote sensing image and each of the multiple recognition models is obtained based on the basic information of the remote sensing image. Edge detection analysis is performed on the remote sensing image of the area to be tested to divide the remote sensing image of the area to be tested into several undetermined areas. The recognition model with the highest matching degree is selected to identify the land type of the several undetermined areas, generating image labels corresponding to the land types of the several undetermined areas. Based on the image labels, it is determined whether the undetermined area is arable land. If so, the remote sensing image corresponding to the undetermined area is extracted as the arable land image of the area to be tested.

[0005] In one example, before obtaining the matching degree between the remote sensing image and the various recognition models based on the basic information of the remote sensing image, the method further includes: obtaining weather information within a preset time range of the area to be tested based on the time information of the remote sensing image; if there is rainfall or snowfall within the preset time range of the area to be tested, adjusting the tone of the remote sensing image based on the weather information.

[0006] In one example, selecting the recognition model with the highest matching degree to identify the plurality of undetermined regions and generate image labels for the plurality of undetermined regions specifically includes: obtaining the pixel values ​​and positional relationships of each pixel in the remote sensing image corresponding to the undetermined region, and inputting the pixel values ​​and positional relationships of each pixel into the recognition model; the recognition model generates image labels corresponding to the land type of the undetermined region based on the color recognition results of each pixel, the positional relationships, and the pre-stored correspondences in the recognition model.

[0007] In one example, before determining the remote sensing images of the area to be tested, the method further includes: acquiring multiple remote sensing images from multiple remote sensing data sources about the same area to be tested within different time periods; processing the multiple remote sensing images for heterogeneous data based on the basic information of the multiple remote sensing images, converting them into a unified format; and determining the image labels of the plurality of undetermined areas within the area to be tested based on the multiple remote sensing images of the same area to be tested within different time periods in the unified format.

[0008] In one example, determining the remote sensing image of the area to be measured specifically includes: selecting a shooting time period based on the geographical range of the area to be measured, wherein the shooting time period starts from the day when the average temperature of a consecutive preset number of days exceeds a preset threshold; and capturing remote sensing images of the area to be measured within the shooting time period.

[0009] In one example, after extracting the remote sensing image corresponding to the area to be determined as the cultivated land image of the area to be tested, the method further includes: acquiring historical cultivated land data of the area to be tested; determining the changes in the cultivated land data within the area to be tested; and marking the areas where the cultivated land data has changed in the cultivated land image.

[0010] In one example, after marking the areas where the cultivated land data has changed in the cultivated land image, the method further includes: acquiring remote sensing images corresponding to the areas where the cultivated land data has changed; based on the remote sensing images corresponding to the areas where the cultivated land data has changed, re-determining whether a misjudgment has occurred in the areas where the cultivated land data has changed; confirming that a misjudgment has occurred, storing the cultivated land data corresponding to the misjudgment in a database, and using the cultivated land data corresponding to the misjudgment to train the recognition model.

[0011] In one example, the spatial information includes the resolution of the remote sensing image, the latitude and longitude coordinates of the data coverage area of ​​the remote sensing image, and the altitude of the data coverage area.

[0012] In one example, the method further includes: confirming that the remote sensing data coverage area of ​​the area to be tested exceeds a preset geographical range; obtaining the matching degree between the area to be tested and the multiple recognition models based on the basic information of the remote sensing image corresponding to the area to be tested; and using the recognition model with the highest matching degree to identify the area to be tested to obtain the image label of the area to be tested.

[0013] This application also proposes a farmland image extraction device, comprising:

[0014] At least one processor; and a memory communicatively connected to said at least one processor; wherein,

[0015] The memory stores instructions executable by the at least one processor. These instructions, when executed by the at least one processor, enable the processor to perform the following: determine a remote sensing image of a region to be tested; acquire basic information of the remote sensing image, including at least temporal and spatial information; determine multiple pre-stored recognition models in a database; and, based on the basic information of the remote sensing image, obtain the matching degree between the remote sensing image and each of the multiple recognition models; perform edge detection analysis on the remote sensing image of the region to be tested to divide the remote sensing image of the region to be tested into several undetermined regions; select the recognition model with the highest matching degree to identify the land type of the several undetermined regions, generating image labels corresponding to the land types of the several undetermined regions; determine whether the undetermined region is arable land based on the image labels; if so, extract the remote sensing image corresponding to the undetermined region as the arable land image of the region to be tested.

[0016] The method proposed in this application can divide the overall remote sensing image into several undetermined regions through edge detection analysis. Then, based on the basic information of the remote sensing image, a suitable recognition model is matched to identify the undetermined regions and obtain the image label of each undetermined region. This allows it to determine whether there is farmland in each undetermined region, making the process of extracting farmland images from remote sensing images more detailed and improving accuracy and precision. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0018] Figure 1 This is a schematic diagram of a method for extracting farmland images in an embodiment of this application;

[0019] Figure 2 This is a schematic diagram of a farmland image extraction device in an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0022] like Figure 1 As shown in the figure, this application provides a method for extracting farmland images, including:

[0023] S101: Determine the remote sensing image of the area to be measured, and obtain the basic information of the remote sensing image, which includes at least time information, spatial information, and platform information.

[0024] Before extracting farmland images, it is first necessary to determine the remote sensing image of the area to be measured, i.e., the remote sensing image to be extracted, and to obtain various basic information about the remote sensing image. Since remote sensing data comes in various formats, the basic information of the remote sensing image should at least include the time information, spatial information, and platform information of the remote sensing image.

[0025] Since different shooting times can affect the extraction of farmland images, the time information must include the shooting time of the remote sensing image, and may also include the data generation time of the remote sensing image. In order to clearly express the spatial information of the remote sensing image, the spatial information of the remote sensing image should at least include the resolution of the remote sensing image, the latitude and longitude coordinates of the data coverage area of ​​the remote sensing image, and the altitude of the data coverage area of ​​the remote sensing image.

[0026] S102: Determine the multiple recognition models pre-stored in the database, and obtain the matching degree between the remote sensing image and the multiple recognition models based on the basic information of the remote sensing image.

[0027] To analyze whether arable land exists in a designated area, a recognition model is needed to identify the corresponding remote sensing image. However, due to the varying basic topography of the area, different types of recognition models are required. Therefore, it is necessary to first determine the matching degree between the remote sensing image and various recognition models based on the basic information of the remote sensing image. This requires utilizing the temporal and spatial information from the basic information of the remote sensing image, as different types of recognition models correspond to different times and locations.

[0028] S103: By performing edge detection analysis on the remote sensing image of the area to be tested, the remote sensing image of the area to be tested is divided into several undetermined areas.

[0029] To refine the extracted farmland images, the area to be measured needs to be divided into several undefined regions using edge detection analysis. Edge detection is a fundamental problem in image processing and computer vision, aiming to identify points in a digital image where brightness changes are significant. Significant changes in image attributes typically reflect important events and shifts in those attributes. These include discontinuities in depth, surface orientation, material properties, and scene lighting. Since remote sensing images may contain multiple farmland areas of varying shapes and sizes, the shapes of the undefined regions after segmentation may also be irregular.

[0030] Taking mountainous terrain as an example, due to its uneven terrain, scattered settlements, and often obstructed transportation by mountains and rivers, vegetation distribution is also more complex. Therefore, the area suitable for agriculture in mountainous areas is limited, and arable land is often scattered in small areas. Because individual plots of arable land are small, their boundaries are winding and lack a fixed orientation, and the boundaries between adjacent plots are often not straight, these characteristics pose significant challenges to the extraction of arable land images. Therefore, to reduce interpretation errors caused by these factors while maintaining existing image quality, it is necessary to highlight arable land boundaries as much as possible within the image, which means using edge detection technology to divide the remote sensing image into several undetermined regions. It should be noted that these undetermined regions may or may not contain arable land; they are simply identified through edge detection analysis of the undetermined regions.

[0031] S104: Select the recognition model with the highest matching degree to recognize the plurality of undetermined regions and generate image labels for the plurality of undetermined regions.

[0032] After testing the matching degree between the basic information of the remote sensing image and various preset recognition models, the recognition model with the highest matching degree is selected as the recognition model for that remote sensing image. This model is then used to identify several undetermined regions, generating image labels for each region. It should be noted that there are many types of image labels. For example, if the basic terrain of the undetermined region is sloped and it is determined that there is no arable land in the region, then the image label for that region is "slope, no arable land." Of course, other image labels can also be included, such as the presence of shadows in the remote sensing image of the region. It is important to note that although image labels are diverse, there must be an image label related to arable land, including: "arable land is determined to exist in the undetermined region," "arable land is determined to not exist in the undetermined region," "arable land is highly likely to exist in a certain area of ​​the undetermined region," and "further image processing is required," etc.

[0033] S105: Determine whether the area to be determined is a farmland area based on the image label;

[0034] S106: If so, extract the remote sensing image corresponding to the area to be determined as the farmland image of the area to be tested.

[0035] After obtaining the image labels for the areas to be identified, the presence of farmland in the areas is determined based on the information about farmland contained in the image labels. If the image is low-resolution or contains shadows, further processing is performed, and the image labels for the areas to be identified are re-determined using a recognition model. This determines whether the area is farmland. Finally, several remote sensing images containing farmland in the areas to be identified are extracted as farmland images. For example, if the overall image tone is dark, and the image labels indicate that the remote sensing image of the area to be identified is dark, brightness adjustment and stretching can be performed on the remote sensing image to make the tone brighter and the contrast between farmland and its boundaries more obvious.

[0036] In one embodiment, when a satellite or sensor acquires a remote sensing image, the weather in the area to be measured may vary within a short time range around that point in time, affecting the remote sensing image and thus impacting subsequent farmland identification.

[0037] Therefore, to avoid the impact of varying weather conditions, before calculating the matching degree between the remote sensing image and each preset recognition model, weather information within a preset time range can be obtained from the time information of the remote sensing image. Taking a day as an example, if there is widespread rainfall or snowfall on the day the remote sensing image is generated, color correction is performed on the remote sensing image based on the weather information to prevent the influence of different weather conditions on the remote sensing image.

[0038] In one embodiment, when selecting the recognition model with the highest matching degree to identify the region to be identified and generate the image label of the region to be identified, the location data of all pixels and the color recognition results of all pixels in the remote sensing image corresponding to the region to be identified are first obtained, and the location data and color recognition results are input into the recognition model. The recognition model then generates the image label of the land type corresponding to the region to be identified based on the pre-stored color recognition results, location relationship, and correspondence between geographical types.

[0039] In one embodiment, since remote sensing images may be captured during cloudy or rainy weather, the images may have issues such as low hue, dark colors, or shadows that affect farmland identification.

[0040] Based on this, multiple remote sensing images of the same area under test can be obtained from multiple remote sensing data sources over different time periods. Here, remote sensing data sources refer to images captured by different satellites and sensors. Due to the differences in data sources and capture times, to eliminate issues such as excessively low tones and dark colors in the remote sensing images, it is necessary to convert multiple remote sensing images into a unified format based on their basic information, such as standardizing resolution and data type. Furthermore, remote sensing data from different systems may exhibit heterogeneity. In this case, further processing of the heterogeneous remote sensing data is required. After converting multiple remote sensing images into the same format, image labels for the target areas within the test area can be determined based on these unified format images.

[0041] In one embodiment, since the color of farmland varies at different times, the period with lower interpretation darkness can be selected when capturing remote sensing images. Based on this, spring images corresponding to the area under test can be chosen as the time for capturing remote sensing images. During this period, farmland is being renovated, and its color tone differs significantly from woodland and bare land, making it the optimal time for farmland interpretation. Since individual perceptions of spring may differ, the start of spring can be defined as the period after winter when the average daily temperature is above a preset temperature for a consecutive preset number of days.

[0042] In one embodiment, since arable land conditions typically do not change, historical arable land data is also very helpful for the next round of arable land surveys. Based on this, the current arable land can be compared with the historical arable land data of the area to be surveyed in advance to determine the changes in arable land in the area to be surveyed, and the changes in arable land data can be marked in the arable land image to provide basic data for arable land protection.

[0043] Furthermore, after comparing with historical farmland data, to avoid misjudgment, remote sensing images of potential farmland changes are obtained for the undetermined areas. These images are then used to re-identify areas where farmland data has changed. Alternatively, drones can be dispatched for on-site surveys to prevent misjudgments. Even further, if misjudgment occurs when extracting farmland images from remote sensing images, the corresponding misjudged data and the remote sensing images can be uploaded to a database. The misjudged data can then be used to train various recognition models, reducing the likelihood of misjudgments.

[0044] like Figure 2 As shown in the illustration, this application also provides a farmland image extraction device, comprising:

[0045] At least one processor; and,

[0046] A memory communicatively connected to the at least one processor; wherein,

[0047] The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to:

[0048] A remote sensing image of the area to be tested is determined, and basic information of the remote sensing image is obtained, including at least temporal and spatial information. Multiple recognition models pre-stored in a database are identified, and the matching degree between the remote sensing image and each of the multiple recognition models is obtained based on the basic information of the remote sensing image. Edge detection analysis is performed on the remote sensing image of the area to be tested to divide the remote sensing image of the area to be tested into several undetermined areas. The recognition model with the highest matching degree is selected to identify the land type of the several undetermined areas, generating image labels corresponding to the land types of the several undetermined areas. Based on the image labels, it is determined whether the undetermined area is arable land. If so, the remote sensing image corresponding to the undetermined area is extracted as the arable land image of the area to be tested.

[0049] This application embodiment also provides a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured as follows:

[0050] A remote sensing image of the area to be tested is determined, and basic information of the remote sensing image is obtained, including at least temporal and spatial information. Multiple recognition models pre-stored in a database are identified, and the matching degree between the remote sensing image and each of the multiple recognition models is obtained based on the basic information of the remote sensing image. Edge detection analysis is performed on the remote sensing image of the area to be tested to divide the remote sensing image of the area to be tested into several undetermined areas. The recognition model with the highest matching degree is selected to identify the land type of the several undetermined areas, generating image labels corresponding to the land types of the several undetermined areas. Based on the image labels, it is determined whether the undetermined area is arable land. If so, the remote sensing image corresponding to the undetermined area is extracted as the arable land image of the area to be tested.

[0051] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the description of the method embodiments.

[0052] The devices and media provided in this application are one-to-one with the methods. Therefore, the devices and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0053] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0054] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0055] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0056] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0057] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0058] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0059] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0060] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0061] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for extracting farmland images, characterized in that, include: Determine the remote sensing image of the area to be measured, and obtain the basic information of the remote sensing image, which includes at least time information and spatial information; A number of pre-stored recognition models are identified in the database, and the matching degree between the remote sensing image and the number of recognition models is obtained based on the basic information of the remote sensing image. By performing edge detection analysis on the remote sensing image of the area to be tested, the remote sensing image of the area to be tested is divided into several undetermined areas; The recognition model with the highest matching degree is selected to identify the land type of the several undetermined areas, and image labels corresponding to the land type of the several undetermined areas are generated. Based on the image labels, determine whether the area to be determined is arable land. If so, extract the remote sensing image corresponding to the area to be determined as the farmland image of the area to be tested; Determine the remote sensing image of the area to be measured, specifically including: Based on the geographical range of the area to be tested, a shooting time period is selected, with the shooting time period starting from the day when the average temperature of a consecutive preset number of days exceeds a preset threshold. During the specified shooting time period, remote sensing images of the area to be measured are captured. After extracting the remote sensing image corresponding to the area to be determined as the cultivated land image of the area to be measured, the method further includes: Obtain historical cultivated land data for the area to be tested; Determine the changes in the cultivated land data within the area to be measured, and mark the areas where the cultivated land data has changed in the cultivated land image; After marking the areas where the cultivated land data has changed in the cultivated land image, the method further includes: Obtain remote sensing images corresponding to areas where the cultivated land data has changed; based on the remote sensing images corresponding to areas where the cultivated land data has changed, re-determine whether a misjudgment has occurred in the areas where the cultivated land data has changed. If a misjudgment is confirmed, the farmland data corresponding to the misjudgment is stored in the database, and the recognition model is trained using the farmland data corresponding to the misjudgment.

2. The method according to claim 1, characterized in that, Before obtaining the matching degree between the remote sensing image and the various recognition models based on the basic information of the remote sensing image, the method further includes: Based on the time information of the remote sensing image, obtain the weather information of the area to be measured within a preset time range; If there is rain or snow in the area to be measured within a preset time range, the color tone of the remote sensing image is adjusted according to the weather information.

3. The method according to claim 1, characterized in that, The step of selecting the recognition model with the highest matching degree to identify the plurality of undetermined regions and generating image labels for the plurality of undetermined regions specifically includes: Obtain the pixel values ​​and positional relationships of each pixel in the remote sensing image corresponding to the region to be determined, and input the pixel values ​​and positional relationships of each pixel into the recognition model; The recognition model generates image labels corresponding to the land type of the area to be determined by using the color recognition results of each pixel, the positional relationship, and the pre-stored correspondence in the recognition model.

4. The method according to claim 1, characterized in that, Before determining the remote sensing image of the area to be measured, the method further includes: Multiple remote sensing images of the same area under test within different time periods are acquired from multiple remote sensing data sources. Based on the basic information of the multiple remote sensing images, the multiple remote sensing images are reprocessed into heterogeneous data and converted into a unified format. Based on the unified format of multiple remote sensing images of the same area under test within different time periods, the image labels of several undetermined areas within the area under test are determined.

5. The method according to claim 1, characterized in that, The spatial information includes the resolution of the remote sensing image, the latitude and longitude coordinates of the data coverage area of ​​the remote sensing image, and the altitude of the data coverage area.

6. The method according to claim 5, characterized in that, The method further includes: Confirm that the remote sensing image coverage area of ​​the area to be tested exceeds a preset geographical range; Based on the basic information of the remote sensing image corresponding to the region to be determined, the matching degree between the region to be determined and the various recognition models is obtained; The region to be identified is identified using the recognition model with the highest matching degree, and the image label of the region to be identified is obtained.

7. A farmland image extraction device, characterized in that, include: At least one processor; And, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the steps of any one of claims 1-6.

Images