Geographic coordinate determination method and device based on image recognition, terminal and storage medium

Abstract

The invention relates to the technical field of geographic coordinate processing, and particularly discloses a geographic coordinate determination method and device based on image recognition, a terminal and a storage medium, and the method comprises the steps: anchoring a target region through an image recognition technology, building a corresponding relation between a pixel coordinate and a geographic coordinate through a known position in a host region, therefore, the positioning of the target area and the determination of the geographic coordinates are realized.

Description

Technical Field

[0001] This application relates to the field of geographic coordinate processing technology, and more specifically, to a geographic coordinate determination method, apparatus, terminal and storage medium based on image recognition. Background Technology

[0002] Global Navigation Satellite Systems (GNSS), such as the US Global Positioning System (GPS), use signals from multiple satellites to measure distances and determine location. They have become the mainstream technology for obtaining absolute geographic coordinates in modern society and are widely used in fields such as traffic navigation, logistics tracking, and surveying and exploration.

[0003] However, GNSS technology heavily relies on a clear, unobstructed line-of-sight path between the terminal device and navigation satellites. In practical applications, this prerequisite is often not met, leading to challenges in numerous scenarios.

[0004] In urban canyon environments, towering buildings can severely obstruct and reflect satellite signals. Similarly, in other areas with limited satellite signals, such as dense forests or deep canyons, satellite signals can become extremely unstable due to vegetation or terrain obstruction, making it difficult to provide continuous and reliable positioning services.

[0005] Therefore, an optimized technical solution is expected. Summary of the Invention

[0006] The technical problem to be solved by this application is to provide a method, device, terminal and storage medium for determining geographic coordinates based on image recognition, which solves the problem of unreliable and inaccurate positioning caused by limited satellite signals in the prior art.

[0007] The technical problem to be solved in this application is achieved by the following technical solution:

[0008] In a first aspect, this application provides a geographic coordinate determination method based on image recognition, comprising:

[0009] Define a host region whose size is larger than the target region;

[0010] Acquire an image of the host region and obtain at least two reference points with known geographic coordinates for the host region, including longitude and latitude coordinates;

[0011] Image processing is performed on the host region image to obtain an optimized host region image;

[0012] Extracting pixel coordinates of the target region from the host region optimized image;

[0013] Establish a coordinate mapping relationship based on the geographic coordinates and pixel coordinates of the reference point;

[0014] Based on the coordinate mapping formula and the pixel coordinates of the target area, the geographic coordinates of the target area are determined.

[0015] Furthermore, a host region larger than the target region is defined, including:

[0016] Determine the center point of the target area and its geographic coordinates;

[0017] The bounding box of the host region is calculated based on the longitude extension value, latitude extension value, and geographic coordinates of the center point to anchor the host region.

[0018] Furthermore, an image of the host region is acquired, and at least two reference points with known geographic coordinates for the host region are acquired, including longitude and latitude coordinates. The image of the host region is acquired from a high-precision map.

[0019] Furthermore, image processing is performed on the host region image to obtain a host region optimized image, including:

[0020] Convert the host region image to a grayscale image using color conversion;

[0021] A denoised image is obtained by removing noise from a grayscale image using Gaussian blur denoising.

[0022] Edge detection is used to delineate local region boundaries within the host region to obtain an optimized host region image.

[0023] Furthermore, edge detection is used to delineate local region boundaries within the host region to obtain an optimized host region image, including:

[0024] Canny edge detection is used to delineate local region boundaries within the host region to obtain an optimized host region image.

[0025] Furthermore, the pixel coordinates of the target region are extracted from the host region optimized image, including:

[0026] The optimized host region image is input into a U-Net-based recognition model to obtain a target prediction map;

[0027] Each pixel in the target prediction map is compared with a confidence threshold to determine the mask value of each pixel, thus obtaining a mask map composed of the mask values ​​of each pixel.

[0028] The function in the image processing library is called to search for the region boundaries of the mask image and extract the pixel coordinates corresponding to the region boundaries to obtain the pixel coordinates of the target region.

[0029] Furthermore, based on the geographic coordinates and pixel coordinates of the reference point, a coordinate mapping relationship is established, including:

[0030] Based on the geographic coordinates (lon1, lat1) and pixel coordinates (x1, y1) of reference point P1, and the geographic coordinates (lon2, lat2) and pixel coordinates (x2, y2) of reference point P2, the following coordinate mapping relationship is established:

[0031]

[0032]

[0033] Where x and y are the pixel x-coordinate and pixel y-coordinate of the geographic coordinate point to be determined, respectively; lon1 and lon2 are the longitudes of reference points P1 and P2, respectively; lat1 and lat2 are the latitudes of reference points P1 and P2, respectively; x1 and x2 are the pixel x-coordinates of reference points P1 and P2, respectively; y1 and y2 are the pixel y-coordinates of reference points P1 and P2, respectively; and Longitude(x) and Latitude(y) are the longitudes and latitudes of the geographic coordinate point to be determined, respectively.

[0034] Secondly, this application also provides a geographic coordinate determination device based on image recognition, which is used to execute the method provided by the first aspect of this application or any possible implementation of the first aspect of this application.

[0035] Thirdly, this application also provides a terminal, including a processor and a memory, wherein a computer program is stored in the memory, and when the processor calls the computer program in the memory, it executes the method provided by the first aspect of this application or any possible implementation of the first aspect of this application.

[0036] Fourthly, this application also provides a storage medium storing a computer program, which, when executed, implements the method provided by the first aspect of this application or any possible implementation of the first aspect of this application.

[0037] In summary, this application includes the following beneficial technical effects:

[0038] This method employs image recognition technology to anchor the target area, establishing a correspondence between pixel coordinates and geographic coordinates based on known locations within the host region, thereby achieving the localization of the target area and the determination of its geographic coordinates. This approach leverages existing image data resources and is suitable for auxiliary or alternative localization in harsh environments. Furthermore, the widespread availability of data sources such as street view images, drone aerial images, and satellite remote sensing images makes image recognition-based geographic coordinate determination methods highly scalable and promising for future applications. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating a method for determining geographic coordinates based on image recognition.

[0040] Figure 2 This is a schematic diagram of the terminal structure. Detailed Implementation

[0041] To facilitate a clear understanding of the technical means, creative features, objectives, and effects of this application, the following description, in conjunction with specific illustrations, further elaborates on this application.

[0042] This application discloses a geographic coordinate determination method based on image recognition. It first determines a host region with a range larger than the target region, and acquires an image of the host region. It also acquires at least two reference points with known geographic coordinates for the host region, including longitude and latitude coordinates, to provide data support for further identification of the target region and establishment of coordinate mapping relationships.

[0043] Next, image processing is performed on the host region image to obtain the host region optimized image, filtering out interference and noise information in the original image to provide more accurate data input for target region recognition;

[0044] Subsequently, pixel coordinates of the target region are extracted from the optimized image of the host region to anchor the target region using image recognition technology. Based on the geographic coordinates and pixel coordinates of the reference point, a coordinate mapping relationship is established to further determine the geographic coordinates of the target region.

[0045] This approach leverages existing image data resources and is suitable for auxiliary or alternative positioning in harsh environments. Furthermore, the widespread availability of data sources such as street view images, drone aerial images, and satellite remote sensing images makes image recognition-based geographic coordinate determination methods highly scalable and promising for future applications.

[0046] like Figure 1 As shown, the geographic coordinate determination method based on image recognition provided in this application specifically includes the following steps:

[0047] S1. Determine a host region whose range is larger than the target region.

[0048] It should be understood that the image recognition-based geographic coordinate determination method proposed in this application is linked to a specific region. In the technical solution of this application, this specific region is designated as the target region. Thus, the image recognition-based geographic coordinate determination method can be highly targeted to this target region, thereby achieving high-precision geographic coordinate estimation and determination.

[0049] In one embodiment, approximate geographic coordinates (latitude and longitude coordinates) provided by the user are obtained to acquire initial location information for the target area. These latitude and longitude coordinates can come from devices with weak GPS signals or be obtained through manual marking using a handheld map application.

[0050] Next, using these latitude and longitude coordinates as the center point of the calculation, a certain range is extended outward to define the host region. Specifically, the bounding box of the host region is calculated using preset longitude and latitude extension values, where the preset latitude and longitude extension values ​​are pre-set according to the precision of the application scenario. For example, for large natural terrain, 0.05-0.1 degrees is set as the preset extension value.

[0051] The calculation process for the bounding box is as follows: minimum longitude Lon min =Lon target -d lon Minimum latitude (Lat) min =Lat target -d lat Maximum longitude Lon max =Lon target +d lon Maximum latitude Lat max =Lat target +d lat Then [Lon] min Lat min Lon max Lat max This constitutes the geographical range of the host region. Among them, Lon target and Lat target These are the longitude and latitude coordinates of the center point, respectively, d lon and d lat These are the longitude extension value and the latitude extension value, respectively.

[0052] In addition, during actual operation, the delineation results are visualized after the bounding box is determined, so that users can confirm them and ensure that the host area can encompass the target area.

[0053] By defining the host region, the target region gains sufficient contextual information, ensuring its accurate delineation within the processing area and providing a crucial prerequisite for selecting reference points to determine unknown geographic coordinates within the target region. Furthermore, since the host region covers a wider area of ​​the target region, even with potentially significant deviations in initial positioning information, the target can still fall within the host region, giving the entire method greater fault tolerance.

[0054] S2. Obtain an image of the host region and obtain at least two reference points with known geographic coordinates for the host region, including longitude and latitude coordinates.

[0055] Once the host region is determined, (high-definition) images of the host region are acquired from local storage, image acquisition devices, or through manual input.

[0056] In one example, the image of the host region is obtained from a high-precision map. This is readily available data known to those skilled in the art, and since how it is processed is not the focus of this application, it will not be described in detail here.

[0057] It should be noted that this process requires obtaining the geographic coordinates of at least two points in the host region as reference points to unlock scale and rotation information, thereby establishing the most basic coordinate mapping relationship. That is, when acquiring an image of the host region in a high-precision map, at least two points are selected as reference points. The pixel coordinates (x1, y1) and geographic coordinates (lon1, lat1) of reference point P1, and the pixel coordinates (x2, y2) and geographic coordinates (lon2, lat2) of reference point P2 are known information. Here, x1 and x2 are the x-coordinates of reference points P1 and P2, respectively; y1 and y2 are the y-coordinates of reference points P1 and P2, respectively; lon1 and lon2 are the longitudes of reference points P1 and P2, respectively; and lat1 and lat2 are the latitudes of reference points P1 and P2, respectively.

[0058] While two points represent the lower mathematical limit for establishing the most basic mapping relationship, in practical applications, three or more reference points are typically used to improve accuracy and robustness. This allows for handling more complex deformations, increasing the accuracy of coordinate estimation and the tolerance space for implementation, resulting in more accurate and reliable overall results. For example, three non-collinear points can establish an affine transformation, which can handle image stretching and shearing, making calculations more precise.

[0059] S3. Perform image processing on the host region image to obtain an optimized host region image.

[0060] Among them, the host region optimized image is obtained by noise reduction, feature enhancement and other methods from the image of the host region, which can provide accurate data input for the recognition of the target region.

[0061] In one embodiment, an image of the host region is loaded using an image processing library (such as OpenCV for Python). The host region image is then converted to grayscale through color conversion, thus eliminating interference from color information. This makes it easier to capture edge and contour information in the host region image while reducing computational complexity. Next, Gaussian blur denoising is used to remove noise such as salt-and-pepper noise or Gaussian noise present in satellite images, preventing noise from being misidentified as edges during subsequent detection. Then, edge detection, such as Canny edge detection, is used to find local boundaries in the host region, such as earthquake faults, river boundaries, and road edges.

[0062] S4. Extract the pixel coordinates of the target region from the optimized image of the host region.

[0063] In fact, the main task of step S4 is image recognition, that is, to identify the specific location of the target area in the optimized image of the host area in order to anchor the area whose geographic coordinates need to be estimated.

[0064] In one embodiment, a deep learning model such as U-Net or DeepLab is trained to perform image semantic segmentation on the host region optimized image, thereby identifying the target region.

[0065] Specifically, the host region optimized image is first input into the U-Net-based recognition model to obtain the target prediction map; then, each pixel in the target prediction map is compared with the confidence threshold to determine the mask value of each pixel and obtain a mask map composed of the mask values ​​of each pixel; subsequently, functions in the image processing library are called to search for the region boundaries of the mask map and extract the pixel coordinates corresponding to the region boundaries to obtain the pixel coordinates of the target region.

[0066] Taking water body identification as an example, when it is necessary to urgently assess an area that has been flooded due to a severe rainstorm, a satellite image covering the area (the image of the host region) is acquired. The host region is then optimized through image processing, and a pre-trained U-Net-based identification model is called to output the corresponding water body prediction map.

[0067] During training, the recognition model is fed a predetermined number of satellite images containing various water bodies (rivers, lakes, reservoirs, flood-prone areas, etc.) and their corresponding labels. These labels are pixel-level annotations manually created for the satellite images, using different colors to distinguish between water bodies and non-water bodies. The recognition model continuously learns and attempts to generate predictive images that closely approximate the manually annotated images.

[0068] In the water body prediction map, each pixel represents the probability that it belongs to a water body. Each pixel in the water body prediction map is iterated through and compared with a pre-set confidence threshold. If the probability is greater than the threshold, the pixel is considered a water body and its mask value is set to 1; otherwise, it is considered not a water body and its mask value is set to 0.

[0069] All mask values ​​are combined to form a binary mask image. Then, functions in the image processing library are called to process the binary mask image to automatically find the region boundaries and extract the pixel coordinates of all pixels that constitute these boundaries.

[0070] In this way, the pixel coordinates of the target region in the host region image can be accurately obtained (anchored by combining the pixel coordinates of the boundary pixels). It is worth noting that the host region image, the optimized host region image, and the water prediction map have the same size.

[0071] S5. Establish coordinate mapping relationships based on the geographic coordinates and pixel coordinates of the reference point.

[0072] As an exemplary embodiment, the quantification formulas involving longitude and latitude in the technical solution of this application are specifically as follows:

[0073]

[0074]

[0075] Where x and y are the pixel x-coordinate and pixel y-coordinate of the geographic coordinate point to be determined, respectively; lon1 and lon2 are the longitudes of reference points P1 and P2, respectively; lat1 and lat2 are the latitudes of reference points P1 and P2, respectively; x1 and x2 are the pixel x-coordinates of reference points P1 and P2, respectively; y1 and y2 are the pixel y-coordinates of reference points P1 and P2, respectively; and Longitude(x) and Latitude(y) are the longitudes and latitudes of the geographic coordinate point to be determined, respectively.

[0076] By establishing a coordinate mapping relationship, the geographic coordinates of the geographic point to be determined can be obtained by inputting the pixel x-coordinate and pixel y-coordinate of the geographic point to be determined into the relationship.

[0077] S6. Determine the geographic coordinates of the target area based on the coordinate mapping relationship and the pixel coordinates of the target area.

[0078] In other words, the pixel coordinates of the target area are used as input data, and the geographic coordinates of the target area are calculated through coordinate mapping formulas, thereby realizing geographic coordinate determination based on image recognition.

[0079] The image recognition-based geographic coordinate determination method provided in this application can be applied to processing equipment. It provides a specific processing scheme and improves it in detail. This results in a geographic coordinate estimation model with universality, convenient configuration of input parameters, and high processing efficiency, which can meet the high-quality estimation needs of geographic coordinates in fields such as exploration and mapping, and provide accurate and effective data support for engineering activities.

[0080] The image recognition-based geographic coordinate estimation model processing method mentioned in this application can be executed by an image recognition-based geographic coordinate estimation model processing device, or by different types of processing devices such as servers, physical hosts, or user equipment (UE) that integrate the image recognition-based geographic coordinate estimation model processing device. The image recognition-based geographic coordinate estimation model processing device can be implemented in hardware or software. The UE can specifically be a smartphone, tablet, laptop, desktop computer, or personal digital assistant (PDA) or other terminal device. The processing devices can be configured in a device cluster.

[0081] It is understandable that in practical applications, considering that the technical solution of this application is usually based on high-precision maps, the processing device that performs the geographic coordinate estimation model based on image recognition only needs to have the corresponding image and data processing capabilities, and its specific device type and deployment form are quite flexible.

[0082] If subsequent model applications are involved, adjustments can be made based on specific application scenarios. For example, the processing equipment could be related equipment at the engineering site, or it could be related equipment that provides remote data support services to the engineering site through remote services such as web services.

[0083] As an example, the processing equipment may include a first processing equipment that performs model training in the background and a second processing equipment that performs model application in the field. The first processing equipment can distribute the model obtained from the initial training or after continuous updates in actual application to the second processing equipment for model deployment and specific model application.

[0084] This application also discloses a geographic coordinate determination device based on image recognition, which operates as described above in the geographic coordinate determination method based on image recognition.

[0085] like Figure 2As shown, this application also provides a terminal from a hardware structure perspective. Specifically, the terminal of this application may include a processor 301, a memory 302, and an input / output device 303. The processor 301 is used to execute the computer program stored in the memory 302 to implement each step of the image recognition-based geographic coordinate determination method; or, when the processor 301 executes the computer program stored in the memory 302 to implement each step of the image recognition-based geographic coordinate determination method, the memory 302 is used to store the computer program required by the processor 301 to execute the above-mentioned image recognition-based geographic coordinate determination method.

[0086] For example, a computer program may be divided into one or more modules / units, one or more of which are stored in memory 302 and executed by processor 301 to complete this application. One or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in a computer device.

[0087] The terminal may include, but is not limited to, processor 301, memory 302, and input / output device 303. Those skilled in the art will understand that the illustrations are merely examples of the terminal and do not constitute a limitation on the terminal. The terminal may include more or fewer components than illustrated, or combine certain components, or different components. For example, the terminal may also include network access devices, buses, etc., and processor 301, memory 302, input / output device 303, etc., are connected via a bus.

[0088] Processor 301 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal, connecting various parts of the device through various interfaces and lines.

[0089] The memory 302 can be used to store computer programs and / or modules. The processor 301 implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory 302 and by calling data stored in the memory 302. The memory 302 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created according to the use of the terminal, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0090] If the modules / units integrated in the aforementioned terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above.

[0091] Computer programs include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0092] The foregoing has shown and described the basic principles, main features, and advantages of this application. Those skilled in the art should understand that this application is not limited to the above embodiments, and various changes and modifications can be made without departing from the spirit and scope of this application; all such changes and modifications fall within the scope of protection claimed in this application. The scope of protection of this application is defined by the appended claims and their equivalents.

Claims

1. An image recognition-based geographic coordinate determination method, characterized by, The method comprises the following steps: determining a host region with a range greater than a target region; obtaining an image of the host region and obtaining reference points of at least two known geographic coordinates of the host region, the geographic coordinates comprising longitude coordinates and latitude coordinates; performing image processing on the image of the host region to obtain a host region optimized image; extracting pixel coordinates of the target region from the host region optimized image; establishing a coordinate mapping relationship based on the geographic coordinates and pixel coordinates of the reference points; determining the geographic coordinates of the target region based on the coordinate mapping relationship and the pixel coordinates of the target region. 2.The image recognition-based geographic coordinate determination method of claim 1, wherein, The method for determining a host region with a range greater than a target region comprises the following steps: determining a center point of the target region and geographic coordinates of the center point; calculating a bounding box of the host region based on a longitude expansion value, a latitude expansion value, and the geographic coordinates of the center point to anchor the host region. 3.The image recognition-based geographic coordinate determination method of claim 1, wherein, The method for obtaining an image of the host region and obtaining reference points of at least two known geographic coordinates of the host region comprises the following steps: the image of the host region is obtained from a high-definition map. 4.The image recognition-based geographic coordinate determination method of claim 1, wherein, The method for performing image processing on the image of the host region to obtain a host region optimized image comprises the following steps: converting the image of the host region into a grayscale image through color conversion; removing noise in the grayscale image through Gaussian blur denoising to obtain a denoised image; using edge detection to delineate the local region boundaries in the host region to obtain the host region optimized image.

5. The image recognition based geo-coordinate determination method of claim 4, wherein, The method for using edge detection to delineate the local region boundaries in the host region to obtain the host region optimized image comprises the following steps: using Canny edge detection to delineate the local region boundaries in the host region to obtain the host region optimized image. 6.The image recognition based geographic coordinate determination method of claim 1, wherein, The method for extracting pixel coordinates of the target region from the host region optimized image comprises the following steps: inputting the host region optimized image into a U-Net-based recognition model to obtain a target prediction map; comparing each pixel point in the target prediction map with a confidence threshold to determine a mask value of each pixel point to obtain a mask map composed of mask values of each pixel point; calling a function in an image processing library to search for region boundaries of the mask map and extract pixel coordinates corresponding to the region boundaries to obtain the pixel coordinates of the target region. 7.The image recognition-based geographic coordinate determination method of claim 1, wherein, The method for establishing a coordinate mapping relationship based on the geographic coordinates and pixel coordinates of the reference points comprises the following steps: based on the geographic coordinates (lon1, lat1) and pixel coordinates (x1, y1) of reference point P1 and the geographic coordinates (lon2, lat2) and pixel coordinates (x2, y2) of reference point P2, the following coordinate mapping relationship is established: Wherein, x and y are respectively the pixel horizontal coordinate and the pixel vertical coordinate of the to-be-determined geographic coordinate point, lon1 and lon2 are respectively the longitude of the reference points P1 and P2, lat1 and lat2 are respectively the latitude of the reference points P1 and P2, x1 and x2 are respectively the pixel horizontal coordinate of the reference points P1 and P2, y1 and y2 are respectively the pixel vertical coordinate of the reference points P1 and P2, and Longitude(x) and Latitude(y) are respectively the longitude and the latitude of the to-be-determined geographic coordinate point.

8. An image recognition-based geographical coordinate determination apparatus, characterized by comprising: The method of any one of claims 1-7.

9. A terminal, characterized by comprising: A processor and a memory are included, the memory has a computer program stored therein, and the processor executes the computer program in the memory to perform the method of any one of claims 1-7.

10. A storage medium, characterized by A computer program is stored in the storage medium, and the computer program is executed to implement the method of any one of claims 1-7.

Images