Image control point processing method and electronic equipment

Abstract

The invention relates to the field of image processing, and provides an image control point processing method and electronic equipment, and the method comprises the steps: obtaining at least one to-be-detected control point in a to-be-detected image, setting a plurality of epipolar lines corresponding to the to-be-detected control point according to the relative relation between a plurality of reference images and the to-be-detected image, the to-be-inspected control points are automatically inspected as the stable control points or the unstable control points by means of the multiple epipolar lines, manual intervention is not needed, experience dependence and subjective deviation of workers are remarkably reduced, and the efficiency, accuracy and reliability of processing the image control points can be considered easily.

Description

[0001] This application is a divisional application of the invention patent application filed on December 4, 2025, with application number 202511816360.3 and entitled "Image Control Point Processing Method, Electronic Device and Storage Medium". Technical Field

[0002] This invention relates to the field of image processing, and more specifically to image control point processing methods and electronic devices. Background Technology

[0003] In the field of image processing, especially in applications such as remote sensing mapping, UAV aerial photography and 3D reconstruction, stable and reliable image control points are the core elements for achieving high-precision positioning and registration. Points with significant geometric features in the image are usually selected, such as the corners of houses, road inflections, or intersections of linear markers, to establish spatial correspondences and attitude corrections between images.

[0004] In existing technologies, automated algorithms are used to identify feature points in images, which are then manually selected to obtain image control points. However, existing technologies have significant limitations: on the one hand, manual extraction relies heavily on the operator's experience and subjective judgment, resulting in low extraction efficiency and poor consistency; on the other hand, while algorithms based on a single image can automatically identify corner points, they neglect stability verification under multiple viewpoints, easily introducing noise or unstable feature points, such as pseudo-corner points affected by changes in lighting, occlusion, or image distortion. Summary of the Invention

[0005] The present invention aims to at least partially solve the technical problems in the related art. To achieve the above objectives, the present invention provides an image control point processing method and an electronic device.

[0006] Firstly, an image control point processing method is provided, comprising: At least one control point to be tested is obtained from the detection of the image to be tested in a real scene; Based on the relative relationship between multiple reference images and the image to be detected, multiple epipolar lines corresponding to the control point to be tested are set respectively. Each epipolar line corresponding to any control point to be tested is used to test the constraint attributes related to the corresponding reference image. The constraint attribute of any epipolar line is a strong constraint or a weak constraint opposite to a strong constraint. A strong constraint reflects that the epipolar line contains a constraint line segment suitable for being within the corresponding reference image and the corresponding reference image is constrained by the constraint line segment. The maximum similarity between the reference image and the image to be detected is not less than a preset similarity threshold. After traversing multiple epipolar lines corresponding to the same control point to be tested, if the number of corresponding strong constraint epipolar lines is greater than or equal to the second quantity threshold, then the corresponding control point to be tested is determined to be a stable control point; otherwise, the corresponding control point to be tested is determined not to be a stable control point.

[0007] Using the image control point processing method described in the first aspect, when the relative relationships between multiple reference images and the image to be detected are known, multiple epipolar lines corresponding to the same control point to be tested can be set. By means of multiple epipolar lines, the control point to be tested can be automatically tested as a stable control point or an unstable control point without human intervention. This significantly reduces the dependence on human experience and subjective bias, and helps to balance the efficiency, accuracy and reliability of image control point processing.

[0008] Secondly, an image control point processing method is provided, comprising: Get the current image; The corresponding control point to be tested is obtained by optimizing the position of any initial control point in the current image. Based on the relative relationship between multiple reference images and the current image, multiple epipolar lines are set corresponding to the control points to be tested. Define the target region to which any of the control points to be inspected belongs in the current image, and set a count variable with an initial value of zero for the target region; According to each of the aforementioned epipolar lines, if there is a region of the same name in the corresponding reference image that is suitable to match the corresponding target region, then the count variable is kept unchanged; otherwise, the count variable is incremented by 1 and compared with a preset second quantity threshold when the increment is completed. If the count variable is equal to the second quantity threshold, then the corresponding control point to be tested is determined as a stable control point; If the count variable is less than the second quantity threshold and multiple epipolar lines corresponding to the same target region have been traversed, then the corresponding control point to be tested is determined to be an unstable control point.

[0009] Thirdly, an image control point processing method is provided, which includes: Get the current image; The corresponding control point to be tested is obtained by optimizing the position of any initial control point in the current image. Based on the relative relationship between multiple reference images and the current image, multiple epipolar lines corresponding to the control point to be tested are set respectively. Each epipolar line corresponding to any control point to be tested is used to test the constraint attributes related to the corresponding reference image. The constraint attribute of any epipolar line is a strong constraint or a weak constraint opposite to a strong constraint. A strong constraint reflects that the epipolar line contains a constraint line segment suitable for being within the corresponding reference image and the maximum similarity between the corresponding reference image and the current image obtained by being constrained by the constraint line segment is not less than a preset similarity threshold. After traversing multiple epipolar lines corresponding to the same control point to be tested, if the number of corresponding strong constraint epipolar lines is greater than or equal to the second quantity threshold, then the corresponding control point to be tested is determined to be a stable control point; otherwise, the corresponding control point to be tested is determined not to be a stable control point.

[0010] Using the image control point processing method described in the second or third aspect, after optimizing the point positions, the position of any control point to be tested is better than the position of the corresponding initial control point in the current image. Given that the relative relationship between each reference image and the current image is known, the corresponding epipolar lines can be set more accurately. By using multiple epipolar lines, the control points to be tested can be automatically tested as stable or unstable control points without human intervention. This significantly reduces the reliance on human experience and subjective bias, and helps to improve the accuracy, reliability and automation of image control point processing.

[0011] Fourthly, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the image control point processing method as described above.

[0012] The aforementioned electronic device has the same beneficial effects as the image control point processing method described above, and will not be repeated here. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating an image control point processing method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of S1 in an embodiment of the present invention; Figure 3 This is a schematic diagram of the first grayscale image according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the second grayscale image according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the initial mask image according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the expanded mask image according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the current image in an embodiment of the present invention; Figure 8 For corresponding Figure 7 A schematic diagram of a local area in the diagram; Figure 9 and Figure 10 These are schematic diagrams illustrating a control point to be tested near a corresponding initial control point, according to an embodiment of the present invention. Figure 11 This is a schematic diagram of another image control point processing method according to an embodiment of the present invention; Figure 12 This is a schematic diagram of S221 to S223 of an embodiment of the present invention. Detailed Implementation

[0014] Embodiments of the present invention will now be described in detail with reference to the accompanying drawings. When referring to the drawings, unless otherwise indicated, the same reference numerals in different drawings denote the same or similar elements. It should be noted that the embodiments described in the following exemplary embodiments do not represent all embodiments of the present invention. They are merely examples of apparatuses and methods consistent with some aspects of the present invention disclosed in the claims, and the scope of the present invention is not limited thereto. Features in the various embodiments of the present invention can be combined with each other without contradiction.

[0015] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0016] See Figure 1 An embodiment of the image control point processing method of the present invention includes S110 to S130.

[0017] S110, Obtain the current image, in which at least one initial control point exists.

[0018] Optionally, see Figure 2 S110 includes S111 to S115.

[0019] S111, the first grayscale image of the real scene is processed into a second grayscale image through a preset line segment detection model.

[0020] For example, at least one of various line segment detection models such as HoughP-line, EDlines, LSD, and DeepLSD can be used. As long as the line segment detection model meets the requirements of efficiency and accuracy, the embodiments of the present invention are not limited.

[0021] S112, Generate an initial mask image that matches the second grayscale image.

[0022] S113, Dilate the initial mask image to form a dilated mask image.

[0023] For example, in a scenario of a certain urban area, Figure 3 The first grayscale image is shown. Figure 4 The corresponding second grayscale image is shown. Figure 5 The corresponding initial mask image is shown. Figure 6 The corresponding inflated mask image is shown.

[0024] S114, Obtain at least one initial control point based on the inflated mask image.

[0025] S114 includes: extracting multiple distinct candidate control points from the dilated mask image, each candidate control point being the intersection point of multiple line segments; arranging the multiple candidate control points in descending order according to their corresponding grayscale change data to form a candidate sequence, wherein the total number of control points in the candidate sequence is greater than a first quantity threshold; and sorting the first... Each of the candidate control points was determined as an initial control point, wherein... Less than or equal to the first quantity threshold; or, compared with the first in the candidate sequence. The grayscale change data matched with each candidate control point is used as the reference data. Each candidate control point whose grayscale change data is greater than or equal to the reference data is selected as the initial control point. Here, each grayscale change data can refer to the gradient of a pixel.

[0026] Optionally, the first quantity threshold is equal to ,in, Indicates rounding up. This indicates the number of connection points in the aerial triangulation data corresponding to the current image. This represents a suitable proportionality coefficient.

[0027] For example, It can belong to [2000, 2500]. It can belong to [0.001, 0.005], for example: if For 2000 and If the value is 0.001, then the first quantity threshold is 2; if For 2325 and If the value is 0.0021, then the first quantity threshold is 5; if For 2500 and If the value is 0.005, then the first quantity threshold is 13.

[0028] To prevent the first quantity threshold from being too large or too small, the number of connection points... and the proportionality coefficient These values ​​can be set in the computer program according to actual needs, and the embodiments of the present invention do not limit the specific values.

[0029] Compared to the initial mask image, the line segments in the dilated mask image are clearer and more defined, which is more conducive to extracting the intersection points of the line segments as candidate control points.

[0030] S115, Configure the second grayscale image as the current image according to each initial control point, for example: see Figure 7 and Figure 8 Each green dot in the current image represents the corresponding initial control point.

[0031] S120, optimize the positions of each initial control point in the current image to obtain the corresponding control points to be tested.

[0032] Optionally, S120 includes: calculating the corresponding first sub-pixel position based on the first pixel position of each initial control point in the current image, and using each first sub-pixel position as the position of the corresponding control point to be tested.

[0033] Optionally, S120 includes: determining the detection area to which each initial control point belongs in the current image through a preset pixel-level corner detection model, detecting the second pixel position with significant corner features from each detection area, and taking each second pixel position as the position of the corresponding control point to be tested.

[0034] For example, the size of each region to be detected can be 12×12, and the center position of each region to be detected is the same as the position of the corresponding initial control point; see [link to documentation]. Figure 9 and Figure 10 The red marker indicates the location of the corresponding initial control point, and the blue marker indicates the location of the corresponding second pixel. Compared to the location of the initial control point, the second pixel location is closer to the corner, thus making the location of the control point to be tested more accurate.

[0035] Optionally, S120 includes: determining the detection region to which each initial control point belongs in the current image, detecting the second pixel position with significant corner features for each detection region, calculating the corresponding second sub-pixel position based on each second pixel position, and using each second sub-pixel position as the position of the corresponding control point to be tested.

[0036] S130, based on the relative relationship between multiple reference images and the current image, set multiple epipolar lines corresponding to the control points to be tested, and test whether the control points to be tested that are commonly corresponding to the multiple epipolar lines belong to stable control points.

[0037] Optionally, in S130, the process of checking whether the control point to be tested corresponding to multiple epipolar lines belongs to a stable control point includes: checking the constraint attributes related to the corresponding reference image for each epipolar line corresponding to any control point to be tested, wherein the constraint attribute of each epipolar line is a strong constraint or a weak constraint; after traversing multiple epipolar lines corresponding to the same control point to be tested, if the number of corresponding strong constraint epipolar lines is greater than or equal to a second quantity threshold, then the corresponding control point to be tested is determined to belong to a stable control point; otherwise, the corresponding control point to be tested is determined not to belong to a stable control point.

[0038] Optionally, the ratio between the second quantity threshold and the number of all reference images can be expressed as: , The threshold is greater than 0.5 and less than 1. For example, 5 epipolar lines correspond to 5 reference images, and the second quantity threshold can be 3.

[0039] Optionally, the constraint attributes of each epipolar line corresponding to any control point to be tested, in relation to the corresponding reference image, include: if any epipolar line contains a constraint line segment suitable for inclusion within the corresponding reference image, then the search region to which the constraint line segment belongs is defined in the corresponding reference image, and the maximum similarity between the search region and the target region to which the corresponding control point to be tested belongs is calculated, wherein the target region belongs to the current image; if the maximum similarity corresponding to any search region is greater than or equal to a preset similarity threshold, then the corresponding epipolar line is determined to be a strongly constrained epipolar line; if any epipolar line does not contain a constraint line segment or any maximum similarity is less than the similarity threshold, then the corresponding epipolar line is determined not to be a strongly constrained epipolar line.

[0040] Any reference image may pass through the corresponding epipolar line or may not have an intersection with the corresponding epipolar line. Therefore, it is necessary to detect whether any epipolar line contains constraint line segments distributed within the corresponding reference image. Only when the aforementioned constraint line segments exist in the epipolar line is it necessary to define the search area. Furthermore, using similarity as a condition to determine whether an epipolar line belongs to a strongly constrained epipolar line helps to balance the accuracy and efficiency of verifying the relationship between the epipolar line and the reference image and the control points to be tested.

[0041] Using the image control point processing method described above, after optimizing the point positions, the position of any control point to be tested is better than the position of the corresponding initial control point in the current image. Given that the relative relationship between each reference image and the current image is known, the corresponding epipolar lines can be set more accurately. By using multiple epipolar lines, the control points to be tested can be automatically tested as stable or unstable control points without human intervention. This significantly reduces the reliance on human experience and subjective bias, and helps to improve the accuracy, reliability and automation of image control point processing.

[0042] See Figure 11 Another embodiment of the image control point processing method of the present invention includes: S210 and S220.

[0043] S210, at least one control point to be tested is obtained from the image to be tested in a real scene.

[0044] For example, the image to be detected can be the first grayscale image mentioned above. The pixel positions of each corner point can be obtained from the first grayscale image by a preset pixel-level corner detection model. Alternatively, the sub-pixel positions of the corner points can be calculated by a preset sub-pixel-level corner detection model. According to the aforementioned pixel positions or sub-pixel positions, the corresponding control points to be tested can be marked in the first grayscale image.

[0045] For example, at least one of several pixel-level corner detection models such as Harris, Shi-Tomasi, and Susan can be used. For instance, the Harris corner detection model can be called through the cornerHarris function in the OpenCV vision library, and the Shi-Tomasi corner detection model can be called through the goodFeaturesToTrack function in the OpenCV vision library.

[0046] For example, a subpixel-level corner detection model based on interpolation or / and a subpixel-level corner detection model based on fitting can be used, for example, by calling the subpixel-level corner detection model through the cornerSubPix function in the OpenCV vision library.

[0047] S220: Based on the relative relationship between multiple reference images and the image to be tested, set multiple epipolar lines corresponding to the same control point to be tested, and test whether the control point to be tested that is commonly associated with the multiple epipolar lines is a stable control point.

[0048] Optionally, see Figure 12 In S220, the test is conducted based on multiple nuclear lines to determine whether the control point to be tested is a stable control point, including S221 to S223.

[0049] S221 represents the constraint attributes related to the epipolar line inspection and the corresponding reference image for any control point to be inspected.

[0050] S221 includes: detecting whether each epipolar line corresponding to any control point to be tested contains a constraint line segment suitable for inclusion within the corresponding reference image; if so, defining the search region to which the constraint line segment belongs in the corresponding reference image, and calculating the maximum similarity between the search region and the target region to which the corresponding control point to be tested belongs; if not, determining that the corresponding epipolar line does not belong to a strongly constrained epipolar line; detecting whether the maximum similarity corresponding to any search region is greater than or equal to a preset similarity threshold; if so, determining that the corresponding epipolar line belongs to a strongly constrained epipolar line; if not, determining that the corresponding epipolar line does not belong to a strongly constrained epipolar line.

[0051] S222: Determine whether multiple epipolar lines corresponding to the same control point to be inspected have been traversed. If so, execute S223; otherwise, continue executing S221.

[0052] S223, check whether the number of corresponding strong constraint epipolar lines is greater than or equal to the second quantity threshold. If yes, determine that the corresponding control point to be tested belongs to a stable control point. If no, determine that the corresponding control point to be tested does not belong to a stable control point (i.e., an unstable control point). Then, stable control points can be selected and / or unstable control points can be removed.

[0053] For example, the number of all epipolar lines corresponding to the same control point to be tested and having strong constraint attributes is counted to obtain the corresponding number of strong constraint epipolar lines.

[0054] Optionally, the method of verifying whether a control point corresponding to multiple epipolar lines is a stable control point includes: Define the target region in the current image to which any control point to be tested belongs, and set a count variable with an initial value of zero for the target region. For example, the count variable can be represented as count, and the initial value of count is zero. If a region with the same name that matches the corresponding target region exists in the reference image corresponding to each nuclear line, then the count variable remains unchanged; otherwise, the count variable is incremented by 1 and compared with a preset second quantity threshold when the increment is complete. For example: count = count + 1. If the count variable is equal to the second quantity threshold, the corresponding control point to be inspected is determined as a stable control point so that the corresponding inspection task is ended immediately to save inspection time. If the count variable is less than the second quantity threshold and multiple epipolar lines corresponding to the same target region have been traversed, then the corresponding control point to be tested is determined to be an unstable control point.

[0055] It should be noted that the specific method of S220 is the same as that of S130, and will not be repeated here.

[0056] Another embodiment of the present invention provides an electronic device including a memory and at least one processor adapted to be coupled to the aforementioned memory. The aforementioned memory is adapted to store a computer program, and the aforementioned at least one processor is adapted to execute the aforementioned computer program to implement the image control point processing method mentioned above. For example, the electronic device may be a server or a terminal device, wherein each processor may be connected to the memory via a universal serial bus.

[0057] Another embodiment of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the image control point processing method mentioned above.

[0058] The electronic device and computer-readable storage medium described in the embodiments of the present invention can be found in the detailed description of the above-mentioned image control point processing method and its beneficial effects, which will not be repeated here.

[0059] Generally, the computer instructions used to implement the method of the present invention can be carried by any combination of one or more computer-readable storage media. Any storage medium can be temporary or non-temporary, and can include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any combination thereof.

[0060] Computer-readable storage media can be any tangible medium containing a stored program, and more specific examples (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the foregoing.

[0061] This program can be used, or combined with, an instruction execution system, apparatus, or device, and is written in one or more programming languages ​​or a combination thereof to perform the operations of this invention. The programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—as well as conventional procedural programming languages—such as the "C" language or similar programming languages. In particular, Python, suitable for neural network computation, and platform frameworks such as TensorFlow and PyTorch can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or connected to an external computer (e.g., via the Internet according to an Internet service provider).

[0062] Although embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. An image control point processing method, characterized in that, include: At least one control point to be tested is obtained from the detection of the image to be tested in a real scene; Based on the relative relationship between multiple reference images and the image to be detected, multiple epipolar lines corresponding to the control point to be tested are set respectively. Each epipolar line corresponding to any control point to be tested is used to test the constraint attributes related to the corresponding reference image. The constraint attribute of any epipolar line is a strong constraint or a weak constraint opposite to a strong constraint. A strong constraint reflects that the epipolar line contains a constraint line segment suitable for being within the corresponding reference image and the corresponding reference image is constrained by the constraint line segment. The maximum similarity between the reference image and the image to be detected is not less than a preset similarity threshold. After traversing multiple epipolar lines corresponding to the same control point to be tested, if the number of corresponding strong constraint epipolar lines is greater than or equal to the second quantity threshold, then the corresponding control point to be tested is determined to be a stable control point; otherwise, the corresponding control point to be tested is determined not to be a stable control point.

2. The image control point processing method according to claim 1, characterized in that, The constraint attributes associated with the epipolar line inspection and the corresponding reference image for any of the control points to be inspected include: If any of the epipolar lines contains a constraint line segment suitable for inclusion within the corresponding reference image, then the search region to which the constraint line segment belongs is defined in the corresponding reference image, and the maximum similarity between the search region and the target region to which the corresponding control point to be tested belongs is calculated; otherwise, it is determined that the corresponding epipolar line does not belong to a strongly constrained epipolar line. If the maximum similarity corresponding to any of the search regions is greater than or equal to a preset similarity threshold, then the corresponding epipolar line is determined to be a strongly constrained epipolar line; otherwise, the corresponding epipolar line is determined not to be a strongly constrained epipolar line.

3. The image control point processing method according to claim 1 or 2, characterized in that, The image to be detected is set as a grayscale image, and the location of the control point to be inspected is the pixel position or sub-pixel position of the corner point; the ratio between the second quantity threshold and the number of all the reference images is greater than 0.5 and less than 1.

4. An image control point processing method, characterized in that, include: Get the current image; The corresponding control point to be tested is obtained by optimizing the position of any initial control point in the current image. Based on the relative relationship between multiple reference images and the current image, multiple epipolar lines are set corresponding to the control points to be tested. Define the target region to which any of the control points to be inspected belongs in the current image, and set a count variable with an initial value of zero for the target region; According to each of the aforementioned epipolar lines, if there is a region of the same name in the corresponding reference image that is suitable to match the corresponding target region, then the count variable is kept unchanged; otherwise, the count variable is incremented by 1 and compared with a preset second quantity threshold when the increment is completed. If the count variable is equal to the second quantity threshold, then the corresponding control point to be tested is determined as a stable control point; If the count variable is less than the second quantity threshold and multiple epipolar lines corresponding to the same target region have been traversed, then the corresponding control point to be tested is determined to be an unstable control point.

5. The image control point processing method according to claim 4, characterized in that, Obtaining the current image includes: The first grayscale image of the real scene is processed into a second grayscale image using a preset line segment detection model. Generate an initial mask image that matches the second grayscale image; The initial mask image is dilated to form a dilated mask image; At least one initial control point is obtained based on the inflated mask image; The second grayscale image is configured as the current image according to each of the initial control points.

6. The image control point processing method according to claim 5, characterized in that, Obtaining at least one initial control point based on the inflated mask image includes: Extract multiple distinct candidate control points from the inflated mask image; Multiple candidate control points are arranged into a candidate sequence in descending order according to their corresponding grayscale change data. The total number of control points in the candidate sequence is greater than a first quantity threshold, which is equal to... ,in, Indicates rounding up. This indicates the number of connection points in the aerial triangulation data corresponding to the current image. Represents a suitable proportionality coefficient The first in the candidate sequence Each of the candidate control points is determined as the initial control point, wherein... Less than or equal to the first quantity threshold; or, will be compared with the first in the candidate sequence The grayscale change data that matches the candidate control points is used as the reference data, and each candidate control point whose grayscale change data is greater than or equal to the reference data is selected as the initial control point.

7. The image control point processing method according to claim 4, characterized in that, Regarding the optimized position of any initial control point in the current image, the corresponding control point to be tested includes: The corresponding first sub-pixel position is calculated based on the first pixel position of each initial control point, and each first sub-pixel position is used as the position of the corresponding control point to be tested; or... In the current image, the detection region to which each of the initial control points belongs is determined. The second pixel position with significant corner features is detected from each of the detection regions. The corresponding second sub-pixel position is calculated based on each second pixel position. Each second sub-pixel position is taken as the position of the corresponding control point to be tested.

8. The image control point processing method according to any one of claims 4 to 7, characterized in that, The ratio between the second quantity threshold and the number of all the reference images is greater than 0.5 and less than 1.

9. An image control point processing method, characterized in that, include: Get the current image; The corresponding control point to be tested is obtained by optimizing the position of any initial control point in the current image. Based on the relative relationship between multiple reference images and the current image, multiple epipolar lines corresponding to the control point to be tested are set respectively. Each epipolar line corresponding to any control point to be tested is used to test the constraint attributes related to the corresponding reference image. The constraint attribute of any epipolar line is a strong constraint or a weak constraint opposite to a strong constraint. A strong constraint reflects that the epipolar line contains a constraint line segment suitable for being within the corresponding reference image and the maximum similarity between the corresponding reference image and the current image obtained by being constrained by the constraint line segment is not less than a preset similarity threshold. After traversing multiple epipolar lines corresponding to the same control point to be tested, if the number of corresponding strong constraint epipolar lines is greater than or equal to the second quantity threshold, then the corresponding control point to be tested is determined to be a stable control point; otherwise, the corresponding control point to be tested is determined not to be a stable control point.

10. An electronic device, characterized in that, The system includes a processor, a memory, and a computer program adapted to be stored on the memory, wherein the processor is configured to implement the image control point processing method as described in any one of claims 1 to 9 when executing the computer program.

Images