Remote sensing image feature point extraction method and device, electronic equipment and storage medium

Abstract

The invention provides a remote sensing image feature point extraction method and device, electronic equipment and a storage medium. The remote sensing image feature point extraction method comprises the following steps: acquiring initial feature points of a remote sensing image and the confidence of each initial feature point; dividing the remote sensing image into a plurality of grid regions; for each grid region, dividing the grid region into a plurality of sub-regions according to the density of the initial feature points in the grid region; for each sub-region, a first feature point in the sub-region is determined according to the initial feature point in the sub-region, and the first feature point at least comprises the initial feature point with the highest confidence coefficient in the sub-region; and determining a target feature point of the remote sensing image according to all the first feature points. According to the scheme, high-quality uniformly-distributed feature points of the remote sensing image can be obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image processing. Specifically, it relates to a method for extracting feature points of a remote sensing image, an apparatus for extracting feature points of a remote sensing image, an electronic device, a storage medium, and a computer program product. Background Art

[0002] In tasks such as remote sensing image matching, image stitching, target recognition, and 3D reconstruction, it is necessary to extract significant and representative feature points from complex images to provide reliable input data for subsequent matching and calculation. In many remote sensing image processing tasks, the detection and extraction of feature points are key steps. Feature points (such as corner points, edges, or interest points) carry important information of the image and are the basis for subsequent processing and analysis.

[0003] Feature points directly detected may have problems such as redundancy, instability, and low computational efficiency, resulting in low-quality feature points among the detected feature points. And low-quality feature points will affect the accuracy of subsequent matching and calculation and will also occupy additional operating resources. Summary of the Invention

[0004] The present invention is proposed in view of the above problems.

[0005] According to a first aspect of the present invention, there is provided a method for extracting feature points of a remote sensing image. The method includes: obtaining initial feature points of a remote sensing image and the confidence of each initial feature point; dividing the remote sensing image into a plurality of grid regions; for each grid region, dividing the grid region into a plurality of sub-regions according to the density of the initial feature points in the grid region; for each sub-region, determining a first feature point in the sub-region according to the initial feature points in the sub-region, where the first feature point at least includes the initial feature point with the highest confidence in the sub-region; and determining target feature points of the remote sensing image according to all the first feature points.

[0006] Exemplarily, the dividing the remote sensing image into a plurality of grid regions includes: dividing the remote sensing image into a plurality of grid regions according to the initial feature points of the remote sensing image, where the number of initial feature points in each grid region is not less than a first quantity.

[0007] Exemplarily, the determining a first feature point in the sub-region according to the initial feature points in the sub-region includes: for the initial feature points in the sub-region, comparing the confidence between every two initial feature points with a distance less than a first distance threshold and removing the initial feature point with a smaller confidence among the two initial feature points until the distance between any remaining two initial feature points is not less than the first distance threshold to obtain at least one first feature point.

[0008] Exemplarily, after obtaining at least one first feature point with the highest confidence, the method further includes: for each first feature point in each grid region, comparing the confidences between every two second feature points with a distance less than a first distance threshold, and removing the second feature point with a smaller confidence among the two second feature points until the distance between any remaining two second feature points is not less than the first distance threshold, so as to update the at least one first feature point, where each of the two second feature points is a first feature point located in adjacent sub-regions in the grid region.

[0009] Exemplarily, after updating the at least one first feature point with the highest confidence, the method further includes: for the updated first feature points, comparing the confidences between every two third feature points with a distance less than a first distance threshold, and removing the third feature point with a smaller confidence among the two third feature points until the distance between any remaining two third feature points is not less than the first distance threshold, where each of the two third feature points is an updated first feature point located in adjacent grid regions.

[0010] Exemplarily, the sub-region is a square, and the method further includes: determining the first distance threshold according to the side length of the sub-region, where the first distance threshold is not less than 1 / 2 of the side length.

[0011] Exemplarily, dividing the grid region into multiple sub-regions according to the density of the initial feature points in the grid region includes: dividing the grid region into multiple sub-regions on average according to the density of the initial feature points in the grid region and the resolution of the remote sensing image.

[0012] Exemplarily, before determining the target feature points, the method further includes: determining, among the first feature points, each group of first feature points with repeated positions; for each group of first feature points with repeated positions, only retaining one of the first feature points in the group with repeated positions to update the first feature points.

[0013] Exemplarily, determining the target feature points of the remote sensing image according to all the first feature points includes: for each first feature point, determining the normalized confidence of the first feature point according to the confidence of the first feature point, the minimum confidence among all the first feature points, and the maximum confidence among all the first feature points; removing the first feature points with a normalized confidence less than a first threshold to obtain the target feature points of the remote sensing image.

[0014] According to the second aspect of the present invention, there is also provided a device for extracting feature points of a remote sensing image, including:

[0015] An extraction module for obtaining initial feature points of a remote sensing image and the confidence of each initial feature point;

[0016] A grid division module for dividing the remote sensing image into a plurality of grid regions;

[0017] A sub-region division module for dividing each grid region into a plurality of sub-regions according to the density of the initial feature points in the grid region;

[0018] A screening module for determining a first feature point in each sub-region according to the initial feature points in the sub-region, where the first feature point at least includes the initial feature point with the highest confidence in the sub-region;

[0019] A determination module for determining the target feature points of the remote sensing image according to all the first feature points.

[0020] According to the third aspect of the present invention, there is also provided an electronic device, including: a processor and a memory, where computer program instructions are stored in the memory, and when the computer program instructions are run by the processor, they are used to execute the above-mentioned method for extracting feature points of a remote sensing image.

[0021] According to the fourth aspect of the present invention, there is also provided a storage medium, on which program instructions are stored, and when the program instructions are run, they are used to execute the above-mentioned method for extracting feature points of a remote sensing image.

[0022] According to the fifth aspect of the present invention, there is also provided a computer program product, including computer program instructions, and when the computer program instructions are run, they are used to execute the above-mentioned method for extracting feature points of a remote sensing image.

[0023] In the above technical solution, after obtaining the initial feature points of the remote sensing image and the confidence of each initial feature point, the remote sensing image is divided into a plurality of grid regions, and then for each grid region, the grid region is divided into a plurality of sub-regions according to the density of the initial feature points in the grid region. After that, for each sub-region, according to the initial feature points in the sub-region, a first feature point in the sub-region is determined, where the first feature point at least includes the initial feature point with the highest confidence in the sub-region. Finally, according to all the first feature points, the target feature points of the remote sensing image are determined. In this way, the initial feature points of the remote sensing image can be screened to obtain high-quality and evenly distributed feature points of the remote sensing image.

[0024] The above description is only an overview of the technical solution of the present invention. In order to be able to understand the technical means of the present invention more clearly, it can be implemented according to the content of the description. And in order to make the above and other purposes, features and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention are specifically described below. Brief Description of the Drawings

[0025] The above and other objects, features, and advantages of the present invention will become more apparent by describing embodiments of the present invention in more detail in conjunction with the accompanying drawings. The drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. In the drawings, the same reference numerals generally represent the same components or steps.

[0026] Figure 1 A schematic flowchart showing a method for extracting feature points of a remote sensing image according to an embodiment of the present invention;

[0027] Figure 2 A schematic flowchart showing an update of the first feature points according to an embodiment of the present invention;

[0028] Figure 3 A schematic flowchart showing determining target feature points of a remote sensing image according to all the first feature points according to an embodiment of the present invention;

[0029] Figure 4 A schematic diagram showing a method for extracting feature points of a remote sensing image according to an embodiment of the present invention;

[0030] Figure 5 A schematic block diagram showing an apparatus for extracting feature points of a remote sensing image according to an embodiment of the present invention;

[0031] Figure 6 A schematic block diagram showing an electronic device according to an embodiment of the present invention. Detailed Description of the Embodiments

[0032] In order to make the objectives, technical solutions, and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited by the exemplary embodiments described herein. Based on the embodiments of the present invention described herein, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present invention.

[0033] To at least partially solve the above problems, a method for extracting feature points of a remote sensing image is proposed. After obtaining the initial feature points of the remote sensing image and the confidence of each initial feature point, the remote sensing image is divided into multiple grid regions. Then, for each grid region, the grid region is divided into multiple sub-regions according to the density of the initial feature points in the grid region. After that, for each sub-region, according to the initial feature points in the sub-region, the first feature point in the sub-region is determined, where the first feature point at least includes the initial feature point with the highest confidence in the sub-region. Finally, according to all the first feature points, the target feature points of the remote sensing image are determined. In this way, high-quality feature points of the remote sensing image can be obtained.

[0034] Figure 1 FIG. shows a schematic flowchart of a method for extracting feature points of a remote sensing image according to an embodiment of the present invention. As Figure 1 shown, the method for extracting feature points of the remote sensing image may include steps S110 to S140.

[0035] In step S110, the initial feature points of the remote sensing image and the confidence of each initial feature point are obtained.

[0036] The remote sensing image may be any suitable image obtained by using a remote sensing imaging device. Exemplarily, the remote sensing image may be an image in the form of an RGB image, a grayscale image, a binary image, etc. The remote sensing image may be a static image or any video frame in a dynamic video. The remote sensing image may be an image of any suitable size and suitable resolution. The remote sensing image may be a raw image directly collected by a remote sensing sensor or an image after preprocessing operations on the raw image. The preprocessing operations may include all operations for improving the visual effect of the remote sensing image, enhancing its clarity, or highlighting certain features in the image. Exemplarily but not restrictively, the preprocessing operations may include operations such as digitization, geometric transformation, normalization, filtering, etc. on the raw image. On the basis of not affecting subsequent image processing, the remote sensing image may also be a synthesized image.

[0037] Exemplarily, a feature point detection algorithm may be used to detect feature points of the remote sensing image to determine the initial feature points of the remote sensing image and the confidence of each initial feature point. For example, feature detection algorithms such as SIFT, SURF, and ORB may be used to detect feature points of the remote sensing image and the confidence of each feature point to obtain the initial feature points of the remote sensing image and the confidence of each initial feature point. Exemplarily, the extracted feature points may be any one or more types of feature points, such as edge feature points, corner feature points, etc.

[0038] In step S120, the remote sensing image is divided into multiple grid regions.

[0039] According to the size of the preset grid area, the remote sensing image can be evenly divided into multiple grid areas, or the remote sensing area can be divided into grid areas of different sizes. Understandably, the initial feature points in the remote sensing image are usually not evenly distributed, so the size of the preset grid area can be determined according to the distribution of the initial feature points in the remote sensing image.

[0040] Exemplarily, according to the initial feature points of the remote sensing image, the remote sensing image is divided into multiple grid areas, where the number of initial feature points in each grid area is not less than the first quantity.

[0041] On the premise of ensuring that the number of initial feature points in each grid area is not less than the first quantity, the remote sensing image can be divided into multiple grid areas of the same size according to the initial feature points of the remote sensing image. In this way, there can be enough initial feature points in each grid area. For example, the position of each initial feature point in the remote sensing image can be determined first, and then the remote sensing image is divided into multiple grid areas of the same size. When the number of feature points in each grid area is determined according to the position of the initial feature points and the position of the grid area in the remote sensing image. When the number of initial feature points in a grid area is less than the first quantity, the length and / or width of each grid area can be increased until the number of initial feature points in each grid area is not less than the first quantity.

[0042] On the premise of ensuring that the number of initial feature points in each grid area is not less than the first quantity, the remote sensing image can be divided into grid areas with different partial sizes according to the initial feature points of the remote sensing image. In this way, there can be enough initial feature points in each grid area, and the number of initial feature points in each grid area is closer. And the grid areas in the image area with denser initial feature points can be divided smaller, and the grid areas in the image area with sparser initial feature points can be divided larger. For example, the position of each initial feature point in the remote sensing image can be determined first, and then the remote sensing image is divided into multiple grid areas of the same size. When the number of feature points in each grid area is determined according to the position of the initial feature points and the position of the grid area in the remote sensing image. When the number of initial feature points in a grid area is less than the first quantity, the length and width of these grid areas can be increased, the length and width of other grids can be appropriately reduced, and the number of grid areas can be appropriately increased and decreased until the number of initial feature points in each grid area is not less than the first quantity.

[0043] Exemplarily, the first quantity can be 100. When the remote sensing image is divided into multiple grid areas, all the initial feature points are divided into the corresponding grid areas, that is, there are no initial feature points outside the divided grid areas.

[0044] In step S130, for each grid region, the grid region is divided into a plurality of sub-regions according to the density of the initial feature points in the grid region.

[0045] For each grid region, the density of the initial feature points in the grid region can be determined according to the ratio between the number of initial feature points in the grid region and the area of the grid region. Among them, the greater the density, the smaller and more the divided sub-regions are; the smaller the density, the larger and fewer the divided sub-regions are. Among them, each grid region is at least divided into 2 sub-regions.

[0046] Exemplarily, the grid region is evenly divided into a plurality of sub-regions according to the density of the initial feature points in the grid region and the resolution of the remote sensing image.

[0047] Optionally, the lower limit of the number of pixels occupied by the sub-region can be determined according to the resolution of the remote sensing image. When the resolution is higher, the lower limit can be higher; when the resolution is lower, the lower limit can be fewer.

[0048] Optionally, for each grid region, the first coefficient corresponding to the density of the initial feature points in the grid region and the second coefficient corresponding to the resolution of the remote sensing image can be determined, and then the number of pixels A occupied by each sub-region is determined according to the following formula 1:

[0049] A = B * k1 + C * k2 Formula 1

[0050] Among them, B represents the density of the initial feature points in the grid region, C represents the resolution of the remote sensing image, k1 represents the first coefficient, and k2 represents the second coefficient. In this way, the size of the most suitable sub-region in each grid region can be accurately calculated.

[0051] In step S140, for each sub-region, the first feature points in the sub-region are determined according to the initial feature points in the sub-region, where the first feature points at least include the initial feature points with the highest confidence in the sub-region.

[0052] There are usually multiple initial feature points in each sub-region, but not all of the initial feature points are initial feature points with high quality. Therefore, the first feature points in the sub-region are determined according to the initial feature points in the sub-region, so as to filter out the initial feature points with low quality.

[0053] Exemplarily, a confidence threshold can be preset, and the initial feature points with corresponding confidence lower than the confidence threshold are removed, so as to only retain the initial feature points with corresponding confidence greater than or equal to the confidence threshold.

[0054] Exemplarily, for each sub-region, the confidence levels of the initial feature points in the sub-region can be compared, and the initial feature points with lower confidence levels can be removed to obtain second feature points.

[0055] Exemplarily, for the initial feature points in the sub-region, the confidence levels between every two initial feature points with a distance less than the first distance threshold can be compared, and the initial feature point with a smaller confidence level among the two initial feature points can be removed until the distance between any two remaining initial feature points is not less than the first distance threshold, so as to obtain at least one first feature point.

[0056] For example, when the number of initial feature points in the sub-region is n, for the first initial feature point, the first initial feature point can be compared with the remaining n - 1 initial feature points respectively, and then the distance and the size relationship of the confidence levels between the first initial feature point and the remaining n - 1 initial feature points can be determined. The same operation as that for the first initial feature point is performed for each initial feature point in the sub-region to determine the distance and the size relationship of the confidence levels between each initial feature point and the other initial feature points in the sub-region other than this feature point. Then, for each initial feature point in the sub-region, the initial feature point to be removed corresponding to this initial feature point can be determined according to the determined distance and the size relationship of the confidence levels between this initial feature point and the other initial feature points in the sub-region. Among them, the confidence level of the initial feature point to be removed corresponding to this initial feature point is less than the confidence level of this initial feature point, and the distance between this initial feature point and the initial feature point to be removed corresponding to this initial feature point is less than the first threshold. After determining the initial feature points to be removed corresponding to each initial feature point in the sub-region respectively, these initial feature points can be removed, and the remaining initial feature points can be used as the first feature points in the sub-region.

[0057] Also for example, when the number of initial feature points in the sub-region is n, for the first initial feature point, the first initial feature point can be compared with the remaining n - 1 initial feature points respectively, and then the initial feature point to be removed corresponding to this initial feature point can be determined. Among them, the confidence level of the initial feature point to be removed corresponding to this initial feature point is less than the confidence level of this initial feature point, and the distance between this initial feature point and the initial feature point to be removed corresponding to this initial feature point is less than the first threshold. Then the initial feature point to be removed corresponding to this initial feature point is removed to update the initial feature points in the sub-region. After that, the first operation can be repeatedly executed until no initial feature point to be removed can be determined.

[0058] Among them, in each first operation, an unoperated feature point can be queried from the updated initial feature points. Among them, the unoperated initial feature points have not determined their corresponding initial feature points to be removed. Then, based on the queried unoperated initial feature point and the updated initial feature points, the same processing operation as the first initial feature point is performed to determine the initial feature points corresponding to the unoperated initial feature point, and these initial feature points are removed to update the initial feature points in this sub-region.

[0059] It can be understood that since the initial feature point with a smaller confidence among any two initial feature points with a distance less than the first threshold will be removed, the distance between any two first feature points in the finally obtained sub-region is not less than the first distance threshold. By screening the initial feature points in the sub-region according to the magnitude relationship between the distances and confidences of two initial feature points, the distance between every two first feature points in the sub-region can be prevented from being too close and the corresponding confidences are relatively high.

[0060] Exemplarily, when the sub-region is a square, the first distance threshold can be determined according to the side length of the sub-region, where the first distance threshold is not less than 1 / 2 of the side length.

[0061] In step S150, the target feature points of the remote sensing image are determined according to all the first feature points.

[0062] The target feature points are the high-quality and evenly distributed feature points expected to be obtained.

[0063] Since the first feature points have been screened and updated, all the first feature points can be directly used as the target feature points of the remote sensing image. It is also possible to screen and update the first feature points, so that the screened and updated first feature points are used as the target feature points of the remote sensing image.

[0064] Exemplarily, when there are too many first feature points, according to the expected number of target feature points, the redundant first feature points can be removed to use the remaining first feature points as the target feature points of the remote sensing image. Among them, the redundant first feature points can be determined according to the confidence distribution of the first feature points. For example, the redundant first feature points can be the first feature points with a confidence lower than the preset threshold, the first feature points with a normalized confidence lower than the preset threshold, the specified number of first feature points with the lowest confidence, etc.

[0065] In the above technical solution, after obtaining the initial feature points of the remote sensing image and the confidence of each initial feature point, the remote sensing image is divided into multiple grid regions. Then, for each grid region, the grid region is divided into multiple sub-regions according to the density of the initial feature points in the grid region. After that, for each sub-region, according to the initial feature points in the sub-region, the first feature point in the sub-region is determined, where the first feature point at least includes the initial feature point with the highest confidence in the sub-region. Finally, according to all the first feature points, the target feature point of the remote sensing image is determined. In this way, the initial feature points of the remote sensing image can be screened to obtain high-quality and evenly distributed feature points of the remote sensing image.

[0066] Exemplarily, after obtaining at least one first feature point with the highest confidence, the above method for extracting feature points of a remote sensing image may further include step S141: For the first feature points in each grid region, compare the confidence between every two second feature points whose distance is less than the first distance threshold and remove the second feature point with the smaller confidence among the two second feature points until the distance between any remaining two second feature points is not less than the first distance threshold to update at least one first feature point, where every two second feature points are respectively the first feature points in adjacent sub-regions in the grid region.

[0067] Exemplarily, for each grid region, each pair of adjacent sub-regions can be determined according to the position coordinate range of the sub-regions in the grid region. It can be understood that the edges of adjacent sub-regions can overlap.

[0068] It can be understood that there are multiple sub-regions in each grid region. Although for each sub-region, the distance between the first feature points within the sub-region is not less than the first distance threshold, the distance between the first feature points near the edge of the sub-region and the first feature points in the adjacent sub-region of the sub-region may be less than the first distance threshold. And the distance between the first feature points located in non-adjacent sub-regions must be greater than the first distance threshold. Therefore, only the confidence and distance between the first feature points located in adjacent sub-regions need to be compared to update the first feature points. For each grid region, the distance between the updated first feature points in the grid region is not less than the first threshold.

[0069] For example, the first sub-region and the second sub-region are adjacent sub-regions, and i first feature points are located in the first sub-region, and j second feature points are located in the second sub-region. The distances between the i first feature points in the first sub-region and the j first feature points in the second sub-region can be determined respectively. Then, multiple pairs of first feature points with the distance less than the first distance threshold are determined, where one of each pair of first feature points is a first feature point in the first sub-region and the other is a first feature point in the second sub-region. Then, the confidence levels between each pair of first feature points are compared, and the first feature point with a smaller confidence level in each pair of first feature points is determined. Then, the first feature point with a smaller confidence level in each pair of first feature points is removed to update the first feature points in the first sub-region and the second sub-region. Similarly, for the first feature points in each pair of adjacent sub-regions, the same operation can be performed according to the first sub-region and the second sub-region. Thus, the first feature points in the grid region can be updated.

[0070] The distances between the first feature points located in non-adjacent sub-regions must be greater than the first distance threshold. Therefore, only the confidence levels and distances between the second feature points can be compared to update the first feature points. For each grid region, the distances between the updated first feature points in the grid region are not less than the first threshold.

[0071] In the above technical solution, for the first feature points in each grid region, the confidence levels between every two second feature points with a distance less than the first distance threshold are compared, and the second feature point with a smaller confidence level among the two second feature points is removed until the distance between any remaining two second feature points is not less than the first distance threshold to update at least one first feature point, where every two second feature points are respectively the first feature points in adjacent sub-regions in the grid region. In this way, the first feature points can be further screened to avoid the distances between the updated first feature points in each grid region from being too small.

[0072] Exemplarily, after updating at least one first feature point with the highest confidence level, the above method for extracting feature points from a remote sensing image may further include step S142: for the updated first feature points, compare the confidence levels between every two third feature points with a distance less than the first distance threshold, and remove the third feature point with a smaller confidence level among the two third feature points until the distance between any remaining two third feature points is not less than the first distance threshold, where every two third feature points are respectively the updated first feature points in adjacent grid regions.

[0073] The operations performed in step S142 are similar to those performed in the above step S141, and will not be elaborated here. It can be understood that the adjacent sub-regions in the above step S141 can be replaced with adjacent grid regions, the second feature points in the above step S141 can be replaced with third feature points, and the same operations can be performed to obtain the first feature points updated again.

[0074] In the above technical solution, for the updated first feature points, the confidence levels between every two third feature points with a distance less than the first distance threshold are compared, and the third feature point with a smaller confidence level among the two third feature points is removed until the distance between any two remaining third feature points is not less than the first distance threshold. Among them, every two third feature points are respectively the updated first feature points located in adjacent grid regions. In this way, the first feature points can be further screened to avoid the situation that there are still first feature points with too small distances among the updated first feature points.

[0075] Figure 2 Fig. shows a schematic flowchart of updating the first feature points according to an embodiment of the present invention. As Figure 2 shown, before determining the target feature points, the above method may further include steps S210 to S220.

[0076] In step S210, among the first feature points, the first feature points with repeated positions in each group are determined.

[0077] There may be initial feature points in the first feature points that are repeatedly determined as the first feature points. For example, when the third sub-region and the fourth sub-region are adjacent sub-regions and there is an overlap at the edge, the initial feature point located on the overlapping edge can be used as the first feature point in the third sub-region and also as the first feature point in the fourth sub-region. At this time, two first feature points with the same position can be obtained according to this initial feature point.

[0078] The first feature points with the same coordinates can be determined as the first feature points with repeated positions in each group according to the coordinates of the first feature points.

[0079] In step S220, for each group of first feature points with repeated positions, only one of the first feature points in this group of first feature points with repeated positions is retained to update the first feature points.

[0080] For each group of first feature points with repeated positions, any one of the first feature points in this group of first feature points with repeated positions can be retained, and the other first feature points can be deleted to update the first feature points.

[0081] In the above technical solution, among the first feature points, each group of first feature points with repeated positions is determined, and then for each group of first feature points with repeated positions, only one of the first feature points in the group with repeated positions is retained to update the first feature points. In this way, redundant first feature points can be removed.

[0082] Figure 3 FIG. shows a schematic flowchart of determining the target feature points of a remote sensing image according to all the first feature points according to an embodiment of the present invention. As Figure 3 shown, the above step S150 may include steps S310 to S320.

[0083] In step S310, for each first feature point, according to the confidence level of the first feature point, the minimum confidence level among all the first feature points, and the maximum confidence level among all the first feature points, the normalized confidence level of the first feature point is determined.

[0084] The first difference between the confidence level of the first feature point and the minimum confidence level among all the first feature points can be determined, and the second difference between the maximum confidence level and the minimum confidence level among all the first feature points can be determined. The normalized confidence level of the first feature point can be determined according to the ratio of the first difference and the second difference.

[0085] Exemplarily, the normalized confidence level of the first feature point can be determined according to the following formula 2:

[0086]

[0087] where C i ′ is the normalized confidence level of the first feature point, C i is the confidence level of the first feature point, C min represents the minimum confidence level among all the first feature points, and C max represents the maximum confidence level among all the first feature points. The normalized confidence level characterizes the quality distribution of the first feature points. The higher the normalized confidence level, the higher the quality of the first feature point corresponding to the normalized confidence level.

[0088] In step S320, the first feature points with a normalized confidence level less than the first threshold are removed to obtain the target feature points of the remote sensing image.

[0089] The first threshold can be determined according to the distribution of the normalized confidence degrees of all the first feature points. For example, the first threshold can be determined according to the quantity requirement of the target feature points. When the number of determined first feature points is 100 and the expected number of target feature points is 90, the first threshold can be set to remove 10 first feature points with normalized confidence degrees less than the first threshold, and the remaining 90 first feature points are used as the target feature points of the remote sensing image. For another example, in order to avoid low quality of the first feature points, the first threshold can be set to a, and all first feature points with normalized confidence degrees less than a are removed, and only the first feature points with normalized confidence degrees not less than a are retained as the target feature points of the remote sensing image.

[0090] In the above technical solution, for each first feature point, according to the confidence degree of this first feature point, the minimum confidence degree among all the confidence degrees of the first feature points, and the maximum confidence degree among all the confidence degrees of the first feature points, the normalized confidence degree of this first feature point is determined, and then the first feature points with normalized confidence degrees less than the first threshold are removed to obtain the target feature points of the remote sensing image. Further screening and updating the first feature points according to the normalized confidence degrees can improve the quality of the target feature points of the remote sensing image.

[0091] Figure 4 The schematic diagram of the method for extracting feature points of a remote sensing image according to an embodiment of the present invention is shown.

[0092] As Figure 4 shown, the initial feature points of the remote sensing image can be extracted and the confidence degrees of the initial feature points can be determined. Then, according to the distribution of the feature points, the remote sensing image is segmented into multiple grid regions. For each grid region, the size of the sub-region in this grid region can be determined according to the resolution of the remote sensing image and the distribution density of the feature points in this grid region. For each sub-region, according to the initial feature points in this sub-region, the first feature points in this sub-region are determined, where the first feature points at least include the initial feature point with the highest confidence degree in this sub-region. When determining the first feature points in this sub-region, a sliding window can be set in advance, and the size of the sliding window is the same as the size of this sub-region. For different sub-regions, the sliding window can be placed at the position of the corresponding sub-region to use the sliding window to determine the first feature points in this sub-region according to the initial feature points in this sub-region. By way of example, the sliding window and the sub-region can be 3*3 pixel sizes. Then, for each sub-region, the confidence degrees between every two initial feature points with a distance less than the first distance threshold are compared, and the initial feature point with a smaller confidence degree among the two initial feature points is removed until the distance between any remaining two initial feature points is not less than the first distance threshold, so as to remove redundant initial feature points to obtain at least one first feature point. For example Figure 4There are 10 initial feature points in the shown sub-region. Based on these 10 initial feature points, redundant initial feature points are removed, and 3 first feature points can be obtained. Then, the normalized confidence of the first feature points is determined, and the first feature points with a normalized confidence less than the first threshold are removed to obtain the target feature points of the remote sensing image.

[0093] Figure 5 FIG. shows a schematic block diagram of a device for extracting feature points of a remote sensing image according to an embodiment of the present invention. As Figure 5 shown, the device for extracting feature points of a remote sensing image includes an extraction module 510, a grid division module 520, a sub-region division module 530, a screening module 540, and a determination module 550.

[0094] The extraction module 510 is configured to obtain the initial feature points of the remote sensing image and the confidence of each initial feature point.

[0095] The grid division module 520 is configured to divide the remote sensing image into a plurality of grid regions.

[0096] The sub-region division module 530 is configured to, for each grid region, divide the grid region into a plurality of sub-regions according to the density of the initial feature points in the grid region.

[0097] The screening module 540 is configured to, for each sub-region, determine the first feature points in the sub-region according to the initial feature points in the sub-region, where the first feature points at least include the initial feature point with the highest confidence in the sub-region.

[0098] The determination module 550 is configured to determine the target feature points of the remote sensing image according to all the first feature points.

[0099] Exemplarily, the grid division module 520 may include a first division sub-module. The first division sub-module is configured to divide the remote sensing image into a plurality of grid regions according to the initial feature points of the remote sensing image, where the number of initial feature points in each grid region is not less than the first quantity.

[0100] Exemplarily, the screening module 540 includes a first screening sub-module. The first screening sub-module is configured to, for the initial feature points in the sub-region, compare the confidence between every two initial feature points with a distance less than the first distance threshold and remove the initial feature point with the smaller confidence among the two initial feature points until the distance between any remaining two initial feature points is not less than the first distance threshold to obtain at least one first feature point.

[0101] Exemplarily, the screening module 540 may further include a second screening sub-module. The second screening sub-module is configured to, after obtaining at least one first feature point with the highest confidence, for the first feature points in each grid region, compare the confidence levels between every two second feature points with a distance less than the first distance threshold and remove the second feature point with a smaller confidence level among the two second feature points until the distance between any remaining two second feature points is not less than the first distance threshold, so as to update the at least one first feature point, where every two second feature points are respectively first feature points located in adjacent sub-regions in the grid region.

[0102] Exemplarily, the screening module 540 may further include a third screening sub-module. The third screening sub-module is also configured to, after updating the at least one first feature point with the highest confidence, for the updated first feature points, compare the confidence levels between every two third feature points with a distance less than the first distance threshold and remove the third feature point with a smaller confidence level among the two third feature points until the distance between any remaining two third feature points is not less than the first distance threshold, where every two third feature points are respectively the updated first feature points located in adjacent grid regions.

[0103] Exemplarily, the sub-region is a square, and the above remote sensing image feature point extraction device may further include a threshold calculation module. The threshold calculation module is configured to determine the first distance threshold according to the side length of the sub-region, where the first distance threshold is not less than 1 / 2 of the side length.

[0104] Exemplarily, the sub-region division module 530 includes a first division sub-module. The first division sub-module is configured to evenly divide the grid region into multiple sub-regions according to the density of the initial feature points in the grid region and the resolution of the remote sensing image.

[0105] Exemplarily, the above remote sensing image feature point extraction device may further include a first calculation sub-module and a first update sub-module. The first calculation sub-module is configured to, before determining the target feature points, determine each group of first feature points with repeated positions among the first feature points. The first update sub-module is configured to, for each group of first feature points with repeated positions, only retain one of the first feature points in the group with repeated positions to update the first feature points.

[0106] Exemplarily, the determination module 550 may include a normalized confidence calculation sub-module and a second update sub-module. The normalized confidence calculation sub-module is configured to, for each first feature point, determine the normalized confidence of the first feature point according to the confidence of the first feature point, the minimum confidence among all the confidence levels of the first feature points, and the maximum confidence among all the confidence levels of the first feature points. The second update sub-module is configured to remove the first feature points with a normalized confidence less than the first threshold to obtain the target feature points of the remote sensing image.

[0107] According to another aspect of the present invention, an electronic device is also provided. Figure 6 A schematic block diagram of an electronic device according to an embodiment of the present invention is shown. As Figure 6 shown, the electronic device includes a processor and a memory. Among them, computer program instructions are stored in the memory, and when the computer program instructions are run by the processor, they are used to execute the remote sensing image feature point extraction method described above.

[0108] In addition, according to yet another aspect of the present invention, a storage medium is also provided. Program instructions are stored on the storage medium, and when the program instructions are run by a computer or a processor, the computer or the processor is caused to execute the corresponding steps of the above-mentioned remote sensing image feature point extraction method of the embodiments of the present invention, and to implement the corresponding modules in the above-mentioned remote sensing image feature point extraction device according to the embodiments of the present invention. The storage medium may for example include a storage component of a tablet computer, a hard disk of a personal computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disc read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0109] According to another aspect of the present invention, a computer program product is also provided, including computer program instructions, and the computer program instructions are used to execute the above-mentioned remote sensing image feature point extraction method when running.

[0110] Those of ordinary skill in the art can understand the specific implementation and beneficial effects of the above-mentioned remote sensing image feature point extraction device, electronic device, storage medium, and computer program product by reading the above specific description of the remote sensing image feature point extraction method. For the sake of brevity, it will not be elaborated here.

[0111] Although example embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above example embodiments are merely exemplary and are not intended to limit the scope of the present invention thereto. Those of ordinary skill in the art can make various changes and modifications therein without departing from the scope and spirit of the present invention. All such changes and modifications are intended to be included within the scope of the present invention as claimed in the appended claims.

[0112] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0113] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed.

[0114] In the specification provided herein, a large number of specific details are set forth. However, it can be understood that the embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures, and technologies have not been shown in detail so as not to obscure the understanding of this specification.

[0115] Similarly, it should be understood that, in order to streamline the present invention and assist in understanding one or more of the various inventive aspects, in the description of the exemplary embodiments of the present invention, the various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, the methods of the present invention should not be construed as reflecting the intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected in the corresponding claims, the inventive point lies in that the corresponding technical problems can be solved by features less than all the features of a single disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into the detailed description, where each claim itself serves as a separate embodiment of the present invention.

[0116] Those skilled in the art can understand that, except for features that are mutually exclusive, any combination can be used for all the features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all the processes or units of any method or device so disclosed. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) can be replaced by an alternative feature that provides the same, equivalent, or similar purpose.

[0117] In addition, those skilled in the art can understand that, although some of the embodiments described herein include certain features included in other embodiments rather than other features, the combination of the features of different embodiments means that it is within the scope of the present invention and forms different embodiments. For example, in the claims, any one of the claimed embodiments can be used in any combination.

[0118] Each component embodiment of the present invention can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art should understand that a microprocessor or a digital signal processor (DSP) can be used in practice to implement some or all of the functions of some modules in the xx device according to the embodiments of the present invention. The present invention can also be implemented as a device program (e.g., a computer program and a computer program product) for executing part or all of the methods described herein. Such a program implementing the present invention can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, or provided on a carrier signal, or in any other form.

[0119] It should be noted that the above embodiments illustrate the present invention rather than limit the present invention, and those skilled in the art can design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in the claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention can be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In the unit claims listing several devices, several of these devices can be embodied by the same item of hardware. The use of the words first, second, and third, etc. does not denote any order. These words can be interpreted as names.

[0120] As described above, it is only the specific implementation manner or the description of the specific implementation manner of the present invention, and the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should be covered by the protection scope of the present invention. The protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A remote sensing image feature point extraction method, characterized in that: The method comprises: Obtain the initial feature points of the remote sensing image and the confidence of each initial feature point; Dividing the remote sensing image into a plurality of grid areas; For each grid area, the grid area is divided into a plurality of sub-areas according to the density of the initial feature points in the grid area; For each sub-region, determining the first feature point in the sub-region according to the initial feature points in the sub-region, wherein the first feature point includes at least the initial feature point with the highest confidence in the sub-region; According to all the first feature points, target feature points of the remote sensing image are determined.

2. The method according to claim 1, characterized in that The step of dividing the remote sensing image into a plurality of grid areas comprises: According to the initial feature points of the remote sensing image, the remote sensing image is divided into a plurality of grid areas, wherein the initial feature points in each grid area are not less than a first number.

3. The method according to claim 1, characterized in that The step of determining the first feature point in the sub-region according to the initial feature point in the sub-region includes: For the initial feature points in the sub-area, the confidence between every two initial feature points whose distance is less than the first distance threshold is compared and the initial feature point with the smaller confidence among the two initial feature points is removed until the distance between any two remaining initial feature points is not less than the first distance threshold, so as to obtain at least one first feature point.

4. The method according to claim 3, characterized in that After obtaining at least one first feature point with the highest confidence, the method further includes: For the first feature point in each grid area, the confidence between every two second feature points whose distance is less than the first distance threshold is compared and the second feature point with smaller confidence is removed from the two second feature points until the distance between any two remaining second feature points is not less than the first distance threshold, so as to update the at least one first feature point, wherein the every two second feature points are first feature points located in adjacent sub-areas in the grid area.

5. The method according to claim 4, characterized in that After updating the at least one first feature point with the highest confidence, the method further includes: For the updated first feature point, compare the confidence between every two third feature points whose distance is less than the first distance threshold, and remove the third feature point with smaller confidence among the two third feature points until the distance between any two remaining third feature points is not less than the first distance threshold, wherein the every two third feature points are the updated first feature points located in adjacent grid areas.

6. The method according to claim 3, characterized in that The sub-region is a square, and the method further includes: The first distance threshold is determined according to the side length of the sub-region, wherein the first distance threshold is not less than 1 / 2 of the side length.

7. The method according to claim 1, characterized in that The step of dividing the grid area into a plurality of sub-areas according to the density of the initial feature points in the grid area comprises: According to the density of the initial feature points in the grid area and the resolution of the remote sensing image, the grid area is evenly divided into a plurality of sub-areas.

8. The method according to claim 1, characterized in that: Before determining the target feature point, the method further includes: Among the first feature points, determining each group of first feature points with repeated positions; For each group of first feature points with repeated positions, only one of the first feature points with repeated positions is retained to update the first feature point.

9. The method according to claim 1, characterized in that: Determining the target feature points of the remote sensing image according to all the first feature points includes: For each first feature point, determine a normalized confidence of the first feature point according to the confidence of the first feature point, the minimum confidence among the confidences of all the first feature points, and the maximum confidence among the confidences of all the first feature points; The first feature points whose normalized confidence is less than a first threshold are removed to obtain target feature points of the remote sensing image.

10. A remote sensing image feature point extraction device, characterized in that: include: An extraction module is used to obtain the initial feature points of the remote sensing image and the confidence of each initial feature point; A grid division module, used for dividing the remote sensing image into a plurality of grid areas; A sub-region division module is used to divide each grid region into a plurality of sub-regions according to the density of initial feature points in the grid region; A screening module, configured to determine, for each sub-region, a first feature point in the sub-region based on the initial feature points in the sub-region, wherein the first feature point includes at least the initial feature point with the highest confidence in the sub-region; The determination module is used to determine the target feature points of the remote sensing image according to all the first feature points.

11. An electronic device comprising a processor and a memory, characterized in that: The memory stores computer program instructions, which are used by the processor to execute the remote sensing image feature point extraction method according to any one of claims 1 to 9 when the processor is running the computer program instructions.

12. A storage medium having program instructions stored thereon, characterized in that: The program instructions are used to execute the remote sensing image feature point extraction method as described in any one of claims 1 to 9 when running.

13. A computer program product comprising computer program instructions, characterized in that The computer program instructions are used to execute the remote sensing image feature point extraction method according to any one of claims 1 to 9 when running.

Images