Landmark extraction method, terminal device and storage medium based on bit correlation
By performing binary processing and bit storage on the remote sensing images in the satellite imaging system, combined with bit matching calculation, the problem of high computing resources consumption in the traditional grayscale correlation method is solved, and efficient landmark extraction on the satellite platform is realized , ensuring computing efficiency and accuracy.
Patent Information
- Application Number
- CN202410258425.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-07
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-03-07
AI Technical Summary
When extracting landmarks, the traditional grayscale correlation method consumes a lot of computing resources and slow calculation speed, making it difficult to run on a satellite-borne embedded platform with limited resources.
The landmark extraction method based on bit-related is adopted, and the remote sensing image and landmark feature template are binarized and converted into bit-by-bit storage, reducing the amount and number of calculation data and improving calculation efficiency.
On the premise of ensuring the accuracy of landmark extraction, the calculation rate is significantly improved, and it can run on a satellite-based embedded platform with limited resources to meet the requirements of real-time image processing.
Smart Images

Figure CN118212428B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of satellite imaging system image processing, and in particular to a landmark extraction method based on bit correlation, a terminal device and a storage medium. Background Art
[0002] In satellite imaging systems, due to factors such as satellite platform position and attitude measurement errors, camera optical axis pointing errors, and optical system distortion, ground features in satellite images are distorted and offset, which affects the effective detection and precise positioning of ground features. Therefore, images in satellite imaging systems need to be accurately geometrically corrected through spatial reference points with known positions. Landmarks are a common type of spatial reference point, so landmark extraction has become an important part of geometric correction of remote sensing images.
[0003] Landmarks are generally areas with highly recognizable coastal contours, such as estuaries and archipelagos. Landmark images are easily disturbed by noise, sunshine, climate and other factors, and the extraction reliability is not stable. Grayscale correlation method is a highly robust landmark extraction method that can meet the needs of high-reliability landmark extraction. This method compares and matches remote sensing images with landmark feature templates face to face, which can improve the accuracy of landmark calculation. However, this method has a large amount of calculation, especially when applied to large-size remote sensing images, it takes too long; in addition, this method consumes a lot of computing resources and is difficult to run on resource-limited satellite embedded platforms. Summary of the invention
[0004] The main technical problem solved by the present invention is that the traditional grayscale correlation method consumes a lot of computing resources and has a slow computing speed when performing landmark extraction.
[0005] According to the first aspect, an embodiment provides a landmark extraction method based on bit correlation, including:
[0006] Obtaining a remote sensing image for landmark extraction and a corresponding landmark feature template;
[0007] Binarization is performed on the first grayscale image of the remote sensing image and the second grayscale image of the landmark feature template respectively;
[0008] Convert the first grayscale image and the second grayscale image after the binarization process into bit storage;
[0009] Matching the first grayscale image stored in the bit position with the second grayscale image;
[0010] If the matching result meets the preset conditions, landmark information is extracted from the remote sensing image.
[0011] According to the second aspect, an embodiment provides a terminal device, including:
[0012] Memory, used to store programs;
[0013] A processor is used to implement the method described in the first aspect by executing the program stored in the memory.
[0014] According to a third aspect, an embodiment provides a computer-readable storage medium, on which a program is stored, and the program can be executed by a processor to implement the method described in the first aspect.
[0015] According to the bit-correlated landmark extraction method, terminal device and storage medium of the above-mentioned embodiment, the grayscale values in the remote sensing image are binarized and converted into a bit array for storage, and then matched with the landmark feature module. The amount of data to be calculated and the total number of calculations are significantly reduced. While ensuring the accuracy of the landmark extraction calculation results, the calculation efficiency is greatly improved and the requirements of real-time image processing can be met. The method of the present application can be run on a satellite-borne embedded platform with limited resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 Schematic diagram of the flow of the landmark extraction method of the embodiment of the present application (I);
[0017] Figure 2 Schematic diagram of the flow of the landmark extraction method of the embodiment of the present application (II);
[0018] Figure 3 A schematic diagram of preprocessing a selected remote sensing image according to an embodiment of the present application;
[0019] Figure 4 A schematic diagram of the division of rectangular regions in a remote sensing image in an embodiment of the present application;
[0020] Figure 5 A schematic diagram of binary storage after image binarization in an embodiment of the present application;
[0021] Figure 6 A schematic diagram of obtaining a threshold value for image binarization in an embodiment of the present application;
[0022] Figure 7 This is a schematic diagram of the binarization process in the embodiment of the present application;
[0023] Figure 8 is a schematic diagram of a grayscale histogram of a remote sensing image without cloud layers in an embodiment of the present application;
[0024] Fig. 9 is a schematic diagram of a grayscale histogram of a remote sensing image with clouds in an embodiment of the present application;
[0025] Fig.10A schematic diagram of the process of converting bit storage in an embodiment of the present application;
[0026] Fig.11 A schematic diagram of the matching calculation process of the existing correlation method;
[0027] Fig.12 A schematic diagram of the matching calculation process in an embodiment of the present application;
[0028] Fig.13 This is a schematic diagram after bit storage in an embodiment of the present application. DETAILED DESCRIPTION
[0029] The present invention is further described in detail below by specific embodiments in conjunction with the accompanying drawings. Wherein similar elements in different embodiments adopt associated similar element numbers. In the following embodiments, many detailed descriptions are for making the present application better understood. However, those skilled in the art can easily recognize that some features can be omitted in different situations, or can be replaced by other elements, materials, methods. In some cases, some operations related to the present application are not shown or described in the specification, this is to avoid the core part of the present application being overwhelmed by too much description, and for those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations according to the description in the specification and the general technical knowledge in the art.
[0030] In addition, the features, operations or characteristics described in the specification can be combined in any appropriate manner to form various implementations. At the same time, the steps or actions in the method description can also be interchanged or adjusted in a manner that is obvious to those skilled in the art. Therefore, the various sequences in the specification and the drawings are only for the purpose of clearly describing a certain embodiment and are not meant to be a required sequence, unless otherwise specified that a certain sequence must be followed.
[0031] The serial numbers assigned to the components herein, such as "first", "second", etc., are only used to distinguish the objects described and do not have any order or technical meaning. The "connection" and "coupling" mentioned in this application, unless otherwise specified, include direct and indirect connections (couplings).
[0032] Landmarks refer to landmark areas with obvious shape or texture information and known accurate spatial location information. They are important spatial references for geometric correction of remote sensing images. Compared with other landmark extraction methods, the correlation method is more robust and can stably extract landmarks under complex conditions such as poor image quality and cloud interference. It is suitable for fully automated, unmanned landmark extraction. However, the traditional correlation method has the problems of occupying a lot of computing resources and too slow calculation speed. It is difficult to apply in scenarios with few computing resources and high requirements for computing timeliness. In the field of satellite imaging, the terminal equipment carried by the satellite requires strong processor reliability in order to ensure stable operation. Generally, the computing processing power is low. The traditional correlation method occupies more resources and is difficult to apply in the field of satellite imaging.
[0033] The landmark extraction method based on bit correlation provided in the embodiment of the present application performs binarization processing on the remote sensing image and the landmark feature module and converts them into bit storage, performs matching calculation on the remote sensing image stored in the bit and the landmark feature template, and extracts the landmark area that meets the landmark feature template on the remote sensing image. Under the premise of ensuring the accuracy of landmark extraction, the present application uses bits, a unit of measurement that occupies little memory and is easier to calculate, to match and calculate landmarks, significantly improving the calculation rate and greatly reducing the required computing resources, while retaining the advantage of good robustness of the correlation method.
[0034] In the embodiment of the present application, remote sensing images are images acquired by satellite photography. Some images have ocean information and land information, and these images need to be subjected to landmark extraction. Among them, the rectangular area of the remote sensing image is an image cropped from the remote sensing image, and the size corresponds to the landmark feature template, so the rectangular area can also be directly regarded as a remote sensing image.
[0035] In an embodiment of the present application, the landmark feature template (referred to as landmark template) can be a standard landmark image, which can be stored as a color or grayscale image; it can be stored conventionally, or it can be converted into bit storage in advance using the relevant steps in the landmark extraction method provided in the present application.
[0036] like Figure 1 As shown, a landmark extraction method based on bit correlation provided by an embodiment of the present application may include:
[0037] Step 1: Obtain the remote sensing image to be subjected to landmark extraction and the corresponding landmark feature template.
[0038] Step 2: Binarize the first grayscale image of the remote sensing image and the second grayscale image of the landmark feature template. In some embodiments, the landmark feature template may have been binarized, and in this case, the landmark feature template does not need to be processed in step 2.
[0039] Step 3: Convert the binarized first grayscale image and the second grayscale image into bit storage, specifically using binary storage. In some embodiments, the landmark feature template may have been converted into bit storage, and in this case, the landmark feature template / second grayscale image does not need to be converted in step 3.
[0040] Step 4: Match the first grayscale image stored in the bit position with the second grayscale image.
[0041] Step 5: If the matching result meets the preset conditions, the landmark information is extracted from the remote sensing image.
[0042] In some embodiments, Figure 2 As shown, in step 1, obtaining a remote sensing image to be subjected to landmark extraction and a corresponding landmark feature template may include:
[0043] Step 110: Acquire a remote sensing image for landmark extraction from remote sensing images observed by satellite.
[0044] Step 120: Obtain imaging information recorded by the remote sensing image, and obtain a corresponding landmark feature template according to the imaging information.
[0045] According to the imaging parameters or imaging information carried by the remote sensing image, the approximate range of the longitude and latitude of the corresponding landmark of the remote sensing image can be determined, and the landmark feature templates distributed within the longitude and latitude range of the remote sensing image can be selected from the landmark library to support the subsequent landmark extraction. The landmark library is a preset image library that stores the landmark feature templates corresponding to each landmark range.
[0046] A preliminary judgment is made on the landmark area corresponding to the remote sensing image based on the imaging information, and the corresponding landmark feature template data is extracted based on the imaging information. Under normal circumstances, based on the position, posture, camera pointing and other parameters of the remote sensing image at the imaging time, the longitude and latitude range of the landmark corresponding to the image can be roughly determined, and then the landmark feature template that meets the longitude and latitude requirements is extracted from the known landmark library. At the same time, since the longitude and latitude of the landmark feature template are known, the location area of the landmark in the remote sensing image can usually be roughly determined. It should be noted that the ground resolution of the landmark feature template should be consistent with the remote sensing image, otherwise the resolution of the two should be adjusted to be consistent by interpolation. In an embodiment of the present application, the landmark feature template in the landmark library can be a rectangular block of an 8*8 grid (or pixel point), which can be confirmed to be located in a 256*256 image block of the remote sensing image, and this will be discussed later.
[0047] It should be noted that the method of the embodiment of the present application has no requirements on the size of the landmark feature template, the size of the geometric error, and the size of the remote sensing image of the landmark to be extracted, and can be adapted to any of them. The above parameters are only for example description.
[0048] In some embodiments, Figure 2 As shown, before the first grayscale image of the remote sensing image and the second grayscale image of the landmark feature template are binarized respectively, the following steps may be further included:
[0049] Step 200: pre-process the remote sensing image and convert it into a standard orthophoto.
[0050] The preprocessing may include at least one of non-uniform correction, blind pixel elimination, rotation correction and scaling.
[0051] The embodiment of the present application can perform image preprocessing on remote sensing images to facilitate later storage and calculation, while ensuring that the method of the present application can resist noise interference and ensure the accuracy of landmark extraction.
[0052] like Figure 3 As shown, the selected remote sensing image is preprocessed, which may include non-uniform correction, blind pixel elimination, rotation correction, scaling, etc. It is converted into a standard orthophoto. The remote sensing image obtained by satellite observation imaging will be affected by the imaging geometry and the radiation response error of the camera itself, resulting in stretching deformation, bright and dark spots, abnormal grayscale fluctuations, etc. in the remote sensing image, which will have an adverse effect on the subsequent landmark extraction. Therefore, it is necessary to preprocess the remote sensing image for subsequent processing.
[0053] In some embodiments, the remote sensing image may include an ocean image and a land image. Since landmarks are generally areas with highly recognizable coastal contour terrain, such as estuaries and archipelagos, the content recorded by the remote sensing image correspondingly includes an ocean image and a land image.
[0054] Wherein, in step 2, the first grayscale image of the remote sensing image and the second grayscale image of the landmark feature template are binarized respectively, which may include:
[0055] Step 210: Obtain a first grayscale image of the remote sensing image and a second grayscale image of the landmark feature template. If the remote sensing image and the landmark feature template are stored as grayscale images, step 220 can be directly executed.
[0056] Step 220: Determine a grayscale threshold according to the grayscale values of the ocean image in the first grayscale image and the second grayscale image.
[0057] In some embodiments, in step 220, determining the grayscale threshold according to the grayscale values of the first grayscale image and the second grayscale image belonging to the ocean image may include:
[0058] Step 221: obtain the histograms corresponding to the first grayscale image and the second grayscale image, obtain the peak value of the grayscale value belonging to the ocean image in the grayscale image, and use the grayscale value corresponding to the peak value as the grayscale threshold.
[0059] Step 230: binarize the first grayscale image and the second grayscale image according to the grayscale threshold, mark the grids with a value higher than the grayscale threshold as the first value, and mark the remaining grids as the second value.
[0060] The image is binarized for subsequent bit storage and related processing; the grayscale value of each grid in the remote sensing image and the landmark feature template is obtained, and the grayscale threshold is set according to the grayscale distribution of the grayscale values of the grids belonging to the ocean location in each rectangular area of the remote sensing image, and the grids with grayscale values higher than the grayscale threshold are changed to the first value, and the grids not higher than the threshold are changed to the second value. The binarization process of the landmark feature template is similar.
[0061] In some embodiments, the landmark feature template may be an image that has been binarized, so in the above step 2 or step 210-step 230, only the remote sensing image / first grayscale image needs to be binarized. In the subsequent step 3, the landmark feature template may be an image that has been binarized and stored with converted bits.
[0062] In some embodiments, the original remote sensing image data is often saved in the form of a grayscale value array. In this case, the remote sensing image corresponds to the first grayscale image. When it is a color image, the corresponding grayscale image needs to be obtained. The grayscale value can be an integer between 0 and 255. The computer stores these grayscale values in a 16-bit binary form. Each grid grayscale value occupies 2 bytes of memory, so an 8*8 image template occupies 128 bytes of memory. However, the landmark extraction process only needs to use the coastline contour and surrounding information in the image. The complex grayscale information can be simplified into only two categories of sea and land information through binarization. Figure 5 Yes Figure 4 Schematic diagram of the binarization process of the rectangular area Ai,j in the remote sensing image shown. Figure 5 (a) shows the grayscale information of the rectangular area Ai,j. The lower left area is the land area, the upper right area is the ocean area, and the number of each grid is the current grayscale value. Figure 5 (b) in the figure represents the grayscale information stored in binary form by the computer. After the binarization process is completed, Figure 5 (c) shows the land and sea information of the rectangular area Ai,j. The grids in the ocean area are replaced by -1, and the grids in the land area are replaced by 1. Figure 5 (d) in the figure represents the land and sea information stored in binary form by the computer. It can be seen that an 8*8 image needs to store data corresponding to 64 grids.
[0063] Binarization is a key step in the method provided by the embodiment of the present application. This processing step can reduce the influence of factors such as illumination and noise, retain only the basic shape information of the image, and also facilitate subsequent bit compression. The key to this step is to select a good threshold for image binarization. When the image grayscale value is greater than or equal to the threshold, the corresponding image value is set to 1; when it is less than the threshold, the corresponding image value is set to -1. The key to this step is to select the binarization threshold to achieve effective segmentation of the image area, and to effectively preserve the shape information of the image after binarization.
[0064] like Figure 6 As shown in , when selecting the binarization threshold, the image to be binarized is usually subjected to histogram statistics, and the segmentation points of its grayscale changes are analyzed and used as the threshold for image binarization. Figure 7 As shown, the preprocessed remote sensing image (first grayscale image) is binarized to obtain a binarized image.
[0065] In some embodiments, the remote sensing images may include ocean images, land images, and cloud images.
[0066] Wherein, in step 2, before the first grayscale image of the remote sensing image and the second grayscale image of the landmark feature template are binarized respectively, the following steps may also be included:
[0067] Step 201: Obtain a grayscale histogram of a first grayscale image corresponding to each rectangular area in a remote sensing image.
[0068] Step 202: Determine a rectangular area with invalid cloud layers in the first grayscale image according to the grayscale histogram, and remove the rectangular area from the first grayscale image.
[0069] There are clouds in the remote sensing images taken by satellites. The clouds will cover the sea and land information. The edge of the clouds will have a great impact on the subsequent matching. Therefore, in the embodiment of the present application, the rectangular area containing clouds will be removed before the remote sensing image is binarized. Usually, the grayscale of the cloud layer is excluded based on the threshold, because the grayscale histogram of the cloud-free remote sensing image is relatively special, with two obvious peaks. Figure 8 is the grayscale histogram of a cloud-free remote sensing image. The horizontal axis is the grayscale value, and the vertical axis is the number of grids (number of pixels). The peak on the left is the ocean area, and the peak on the right is the land area. Fig. 9 This is the grayscale histogram of a remote sensing image with clouds. In addition to the two peaks of land and ocean, there are also several cloud columnar bars with larger grayscale values, indicating that there are clouds in the remote sensing image, and the grayscale value of the clouds is much higher than that of the ocean and land. Therefore, the corresponding cloud threshold can be set to remove the rectangular area with clouds.
[0070] For example, Figure 8 and Fig. 9 As shown, the maximum grayscale value of the remote sensing image without clouds is about 300, while the maximum grayscale value of the remote sensing image with clouds is about 450. By detecting each rectangular area in the remote sensing image, when there is a grayscale value greater than the cloud threshold (such as 350) and / or the number of grids corresponding to the cloud layer (the number of grids greater than or equal to 350 grayscale values) exceeds the threshold, it is determined that the current rectangular area has a cloud interference problem.
[0071] In some embodiments, in step 3, converting the first grayscale image and the second grayscale image after the binarization process into bit storage may include:
[0072] Step 310: Replace the first value in the first grayscale image and the second grayscale image with a first bit value, such as 1, and replace the second value with a second bit value, such as 0, to obtain a grayscale image converted into bits.
[0073] Step 320: For each row of the grayscale image corresponding to the bit, merge all the bit values in each row, and store a binary value equivalent to each row of grids in binary form. Therefore, bit storage is also understood as converting the bit values into binary storage after merging each row of values.
[0074] like Fig.10 In the first row of (b), the values of the 8 grids in each row are "1", "1", "0", "0", "0", "0", "0", "0", "0", which are combined and stored in binary as "11000000", which is equivalent to storing 192. Fig.10 In the binary image shown in (a), each grid originally needs to be stored in binary format, and 64 values need to be stored. After merging, Fig.10 As shown in (c), only 8 values after merging 8 rows need to be stored. In other words, the embodiment of the present application can reduce the image storage capacity.
[0075] After the binarization, the remote sensing image and landmark feature module are converted into bit storage, that is, the first value (such as number 1) and the second value (such as number -1) of the grid are replaced by the unit binary digits 1 and 0, as shown in Fig.10 As shown, Fig.10 (a) is the land and sea information of the rectangular area Ai,j; Fig.10 (b) in the figure replaces the original 1 and -1 with binary digits 1 and 0 to obtain the rectangular area Asi,j after bit conversion; Fig.10(c) in the figure is the computer storing the rectangular area Asi,j in binary. As a result, the memory occupied by the sea and land information of each grid point is compressed from 2 bytes / 16 bits to 1 bit. Without deleting the stored information, the bit storage compresses the image data volume by 15 times, significantly reducing the consumption of computing resources.
[0076] In some embodiments, the landmark feature template may be an image that has been binarized and bit-converted for storage.
[0077] In the embodiment of the present application, the grayscale value of each pixel of the remote sensing image can be represented by 1-3 bytes, and in special cases, more bytes may be used. However, after the image is binarized, since there are only two values 1 and -1, each pixel can be represented by only one bit, which effectively reduces the required image storage space and simplifies subsequent processing.
[0078] In the embodiment of the present application, the binary digits 1 and 0 are used to replace the 1 and -1 to obtain the remote sensing image / landmark feature template converted into bit storage. Through this method, the required storage space can be compressed to 1 / 8 to 1 / 24 of the original, significantly reducing resource consumption. This method is applicable to spaceborne embedded platforms.
[0079] In some embodiments, Fig.12 As shown, in step 4, matching the first grayscale image stored in the bit with the second grayscale image may include:
[0080] Step 410: Screenshot / crop the first grayscale image of the remote sensing image stored in the bit according to the image size of the landmark feature template to obtain multiple matrix areas to be matched; wherein the image size of the remote sensing image is larger than the image size of the landmark feature template.
[0081] Step 420: perform an identical or equivalent calculation on the second grayscale image of the landmark feature template stored in the bit and the corresponding grid row of the matrix region.
[0082] Step 430: Determine the similarity between the matrix areas of the second grayscale image and the first grayscale image according to the calculation result.
[0083] Step 440: Calculate the familiarity corresponding to all matrix regions as the matching result.
[0084] All rectangular areas in the bit-stored remote sensing image are matched one by one with the same landmark feature template. The matrix area in the remote sensing image is subjected to the same grid XOR calculation (or XOR and negation operation) with the landmark feature template.
[0085] Traditional correlation methods, such as Fig.11As shown, it is necessary to perform an XOR calculation on each grid corresponding to the rectangular area and the landmark template, a total of 64 times. The result matrix Ki,j also consists of 64 grids and needs to store 64 values.
[0086] like Fig.12 As shown, Fig.11 After bit conversion, the first row of matrix region Ai,j shown in (a) is converted from "1", "1", "-1", "-1", "-1", "-1", "-1", "-1" to Fig.12 The first row shown in (a) is 1100 0000 (equivalent to storing 192) and Fig.11 The first row of the landmark feature template shown in (b) is converted to Fig.12 The first row of (b) in the figure is 1110 0000 (equivalent to storage 224); the similarity comparison between the two is performed through the same or calculation (or the exclusive or operation and the negation operation): (11000000 XNOR 11100000) = 11011111 (equivalent to storage 223), which only requires one same or operation and compares the sea and land information of a row of 8 grid points. The remaining 7 rows of information are also compared similarly to obtain Fig.12 The result matrix Ki,j shown in (c) can also be stored in the same way. The first row 11011111 is equivalent to storing 223. Fig.11 The result matrix shown has a storage capacity of 8 values and a smaller storage capacity.
[0087] The rectangular area and the landmark template after the conversion bit storage are both 8 numbers, which is equivalent to matching a row of grids at one time, and only 8 identical or identical calculations are required. The embodiment of the present application not only reduces the storage capacity, but also reduces the total number of matching operations.
[0088] This processing step is the core of the entire method and the key to improving the algorithm processing speed. After the remote sensing image is converted into bit storage, data correlation can be performed through bit operations. Bit operations are low-level operations with very fast calculation speed, which can effectively improve the speed of landmark extraction.
[0089] Bit correlation can be achieved by using the bit operator "XOR". For example, the conventional array correlation is:
[0090] .
[0091] The relevant calculation after biting is:
[0092] .
[0093] The two methods have the same effect when identifying array configurations. Bit correlation does not use conventional multiplication, but rather uses bitwise XOR operations. Because it uses underlying operations, its calculation speed is very fast. Moreover, on a 64-bit processor, 64 pixels can be calculated at a time.
[0094] In some embodiments, in step 5, if the matching result meets the preset conditions, extracting landmark information from the remote sensing image may include:
[0095] Step 510: If there is a matrix region with a similarity greater than a preset threshold, it is determined that the current landmark feature template matches the remote sensing image, and the matrix region with the highest similarity is used as the landmark position information corresponding to the landmark feature template for extraction.
[0096] Select the row and column position where the maximum similarity occurs, that is, the location of the landmark. If the maximum value is less than the preset threshold (such as 90% of the number of pixels in the landmark template), it is considered that the extraction failed. Assuming that the number of landmark rows is m and the number of columns is n, if the correlation peak is less than m*n*0.9, it is considered that the image quality is too poor and the extraction failed. If there is no rectangular area in the remote sensing image that meets the recognition requirement, it is determined that there is no landmark area corresponding to the landmark feature template in the remote sensing image.
[0097] In some embodiments, Figure 2 As shown, the landmark extraction method may further include:
[0098] Step 6: If the matching result does not meet the preset conditions, replace the new landmark feature template and rematch, and repeat one or more steps from step 1 to step 5. Until all the landmark feature templates extracted in step 1 have completed the matching calculation.
[0099] The specific implementation process of matching the remote sensing image stored in the bit position with the landmark feature template and extracting the landmark area that meets the landmark feature template on the remote sensing image includes:
[0100] A) Select the landmark feature template.
[0101] B) Crop a piece of the bit-mapped image to ensure that the size of the cropped image block is exactly the same as the size of the landmark feature template.
[0102] C) Perform the XOR calculation on the cropped image patch and the landmark template, and the count is 1.
[0103] D) Return to step B, translate the position of the cut image block, and cut a new image block. Continue steps C and D until all image areas to be matched are traversed.
[0104] In all result matrices, the matrix areas corresponding to the grids with calculation results of 1 greater than 90% and / or image plane errors less than 20 pixels are selected as landmark candidate sets, and then the rectangular area with the largest similarity (the proportion of the value 1 in the total grids in the result matrix) is selected from the landmark candidate set as the landmark position information corresponding to the landmark feature template for extraction. It should be noted that the maximum similarity is still compared again from the two aspects of the number of grids with calculation results of 1 and the image plane error. When the proportion of the number of grids with results of 1 in all result matrices is less than 90%, that is, no rectangular area is selected in the landmark candidate set, it means that there is no rectangular area matching the landmark feature template in the remote sensing image.
[0105] In the embodiment of the present invention, the image plane error can be calculated in the following way: the real longitude and latitude of the upper left corner of the landmark feature template is converted to the image plane through the line of sight to obtain the theoretical coordinates ( x 1 ,y 1), the matching result of the landmark feature template in the remote sensing image is the actual coordinate ( x 0 ,y 0), the distance between the actual coordinates and the theoretical coordinates is the image plane error Diserr, and 1 pixel is the length unit of the image plane grid coordinate system.
[0106] For example, .
[0107] The schematic diagram after bit storage is as follows Fig.13 As shown, if the original image storage format is used, 8*2 bytes = 16 bytes are needed to store a line, while using the bit image to store a line only requires 1 byte, and the consecutive 8 bits are merged, and 11000000 is equivalent to 192 for storage.
[0108] In some embodiments, two groups of test objects may be used to test the method of the embodiments of the present application on a C language program platform.
[0109] Test 1 is to extract landmarks from the scanning camera image 10 times. The test object is a 256*256 remote sensing image and a 64*64 landmark feature template. Through the existing binary calculation and the bit correlation processing of this embodiment, it can be obtained that the single extraction time takes 1.94s, while the single extraction time of this embodiment only takes 0.19s.
[0110] Test 2 is to extract landmarks from the gaze camera image 10 times. The test object is a 512*512 remote sensing image and a 64*64 landmark feature template. Through the existing binary calculation and the bit correlation of this embodiment, it can be determined that a single extraction time takes 9.31s, while a single extraction of this embodiment only takes 0.82s. The results of the two tests show that the landmark extraction method provided in the embodiment of the present application has a significant advantage in computing speed.
[0111]
[0112] In summary, the method provided by this application has at least the following technical effects:
[0113] 1. Compared with the traditional landmark extraction method based on feature points or contours, the embodiment of the present application performs face-to-face matching of remote sensing images through landmark feature templates, which can meet the needs of high-reliability landmark extraction and ensure the accuracy of landmark extraction.
[0114] 2. The embodiment of the present application performs binarization processing on the grayscale values in the remote sensing image and converts them into a bit array for storage and matching calculation. The amount of data to be calculated and the total number of calculations are significantly reduced. Under the premise of ensuring the accuracy of the landmark extraction calculation results, the calculation efficiency is greatly improved and the requirements of real-time image processing can be met. The method of the present invention can be run on a satellite-borne embedded platform with limited resources.
[0115] The landmark extraction method provided in the present application is implemented by a terminal device, which is a satellite-borne terminal device applied to a satellite imaging system, and the terminal device may include a memory and a processor. For example, the terminal device may be a computer, a server, or other device with computing and data processing capabilities.
[0116] A memory is used to store a program. A processor is used to implement the landmark extraction method described in the above embodiment by executing the program stored in the memory.
[0117] Those skilled in the art will appreciate that all or part of the functions of the various methods in the above-mentioned embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above-mentioned embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, and the storage medium can include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to implement the above-mentioned functions. For example, the program is stored in the memory of the device, and when the program in the memory is executed by the processor, all or part of the above-mentioned functions can be implemented. In addition, when all or part of the functions in the above-mentioned embodiments are implemented by computer programs, the program can also be stored in a storage medium such as a server, another computer, disk, optical disk, flash disk or mobile hard disk, and can be downloaded or copied and saved in the memory of the local device, or the system of the local device is updated, and when the program in the memory is executed by the processor, all or part of the functions in the above-mentioned embodiments can be implemented.
[0118] This document is described with reference to various exemplary embodiments. However, those skilled in the art will recognize that changes and modifications may be made to the exemplary embodiments without departing from the scope of this document. For example, various operating steps and components for performing the operating steps may be implemented in different ways (e.g., one or more steps may be deleted, modified, or combined into other steps) depending on the specific application or considering any number of cost functions associated with the operation of the system.
[0119] Although the principles of this invention have been shown in various embodiments, many modifications of structures, arrangements, proportions, elements, materials and components particularly suitable for specific environments and operational requirements can be used without departing from the principles and scope of this invention. The above modifications and other changes or amendments will be included in the scope of this invention.
[0120] The foregoing specific description has been described with reference to various embodiments. However, those skilled in the art will recognize that various modifications and changes can be made without departing from the scope of the present disclosure. Therefore, the consideration of the present disclosure will be illustrative rather than restrictive, and all these modifications will be included in its scope. Similarly, the advantages, other advantages and solutions to the problems of various embodiments have been described above. However, the benefits, advantages, solutions to the problems and any elements that can produce these, or make them more clear, should not be interpreted as critical, necessary or necessary. The term "include" and any other variants used in this article are all non-exclusive inclusions, so that the process, method, article or device including the list of elements not only includes these elements, but also includes other elements that are not explicitly listed or do not belong to the process, method, system, article or device. In addition, the term "coupled" and any other variants used in this article refer to physical connections, electrical connections, magnetic connections, optical connections, communication connections, functional connections and / or any other connections.
[0121] Those skilled in the art will appreciate that many changes may be made to the details of the above-described embodiments without departing from the basic principles of the invention. Therefore, the scope of the present invention should be determined solely by the claims.
Claims
1. A landmark extraction method based on bit correlation, characterized in that: include: Obtaining a remote sensing image for landmark extraction and a corresponding landmark feature template; Binarizing the first grayscale image of the remote sensing image and the second grayscale image of the landmark feature template respectively; Converting the first grayscale image and the second grayscale image after the binarization process into bits for storage; Matching the first grayscale image and the second grayscale image stored in bits; If the matching result meets the preset conditions, extracting landmark information from the remote sensing image; The remote sensing image includes an ocean image and a land image; wherein the first grayscale image of the remote sensing image and the second grayscale image of the landmark feature template are binarized respectively, including: Acquire a first grayscale image of the remote sensing image and a second grayscale image of the landmark feature template; Determine a grayscale threshold according to the grayscale values belonging to the ocean image in the first grayscale image and the second grayscale image; Binarize the first grayscale image and the second grayscale image according to the grayscale threshold, mark the grids with grayscale values higher than the grayscale threshold as the first value, and mark the remaining grids as the second value; The step of converting the first grayscale image and the second grayscale image after the binary processing into bit storage includes: The first value in the first grayscale image and the second grayscale image is replaced by a first bit value, and the second value is replaced by a second bit value, so as to obtain a grayscale image converted into bit storage; For each row of the grayscale image corresponding to the bit, merge all the bit values in each row and store a binary value equivalent to each row of grids in binary form; The step of matching the first grayscale image and the second grayscale image stored in bits includes: According to the image size of the landmark feature template, a first grayscale image of the remote sensing image stored in the bit is captured to obtain a plurality of matrix regions to be matched; wherein the image size of the remote sensing image is larger than the image size of the landmark feature template; Performing an identical or identical calculation on the second grayscale image of the landmark feature template stored in the bit position and the corresponding grid of the matrix area; Determine the similarity between the matrix area of the second grayscale image and the first grayscale image according to the calculation result; Calculate the corresponding degree of all matrix regions as the matching result.
2. The landmark extraction method according to claim 1, characterized in that: Before the first grayscale image of the remote sensing image and the second grayscale image of the landmark feature template are binarized respectively, the method further includes: Preprocessing the remote sensing image to convert it into a standard orthophoto; The preprocessing includes at least one of non-uniform correction, blind pixel elimination, rotation correction and scaling.
3. The landmark extraction method according to claim 1, characterized in that: Determining a grayscale threshold according to the grayscale values of the first grayscale image and the second grayscale image belonging to the ocean image includes: Obtain a histogram corresponding to the first grayscale image and the second grayscale image, obtain a peak value of the grayscale value belonging to the ocean image in the grayscale image, and use the grayscale value corresponding to the peak value as the grayscale threshold.
4. The landmark extraction method according to claim 1, characterized in that: If the matching result meets the preset conditions, the landmark information is extracted from the remote sensing image, including: If there is a matrix area with a degree of familiarity greater than a preset threshold, it is determined that the current landmark feature template matches the remote sensing image, and the matrix area with the highest degree of familiarity is used as the landmark position information corresponding to the landmark feature template for extraction; And / or, the landmark extraction method further comprises: If the matching result does not meet the preset conditions, replace the new landmark feature template and match again.
5. The landmark extraction method according to claim 1, characterized in that: The remote sensing image has at least one rectangular area including an ocean image, a land image and a cloud image; Before the first grayscale image of the remote sensing image and the second grayscale image of the landmark feature template are binarized respectively, the method further includes: Obtaining a grayscale histogram of a first grayscale image corresponding to each rectangular area in the remote sensing image; A rectangular area with invalid clouds in the first grayscale image is determined according to the grayscale histogram, and the rectangular area is removed from the first grayscale image.
6. A terminal device, characterized in that: include: Memory, used to store programs; A processor, configured to implement the method according to any one of claims 1 to 5 by executing the program stored in the memory.
7. A computer-readable storage medium, characterized in that: The medium stores a program, which can be executed by a processor to implement the method according to any one of claims 1 to 5.
Citation Information
Patent Citations
Efficient construction method of hyperspectral remote sensing data compression and classification model
CN112149712A
Forest land distribution extraction method and device based on remote sensing image texture features
CN112949657A
Method for identifying mixed loading of people and goods at entrance and exit
CN117173689A
