A method for intelligent recovery and positioning of underwater robots

Abstract

The present invention relates to the field of image processing technology, and in particular to a method for intelligent recovery and positioning of an underwater robot, comprising: obtaining a water surface image, obtaining a suspected light source area through iterative threshold segmentation; obtaining the similarity between every two suspected light source areas according to the grayscale difference and shape difference in the suspected light source area, grouping the suspected light source areas, and recording the group with the most regular shape of the convex polygon formed by the suspected light source areas in each group as the actual light source group; and obtaining the moving direction of the underwater robot during the recovery process according to the actual light source group. In the process of intelligent recovery of the underwater robot, the present invention obtains the suspected light source through threshold segmentation, and distinguishes the positioning light source from the pseudo light source according to the grayscale change and the regularity of the shape of the light source group, so as to avoid the pseudo light source from affecting the positioning result and improve the accuracy of the intelligent recovery and positioning of the underwater robot.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to an intelligent recovery and positioning method for an underwater robot. Background Art

[0002] With the rapid development of electronics, computers, sensors and other technologies, underwater robot technology has gradually matured and has been widely used in the fields of marine development, scientific investigation, etc. The positioning and recovery of underwater robots is also an important part of the application of underwater robot technology.

[0003] When recovering an underwater robot, a light source is usually used as a positioning mark. After the underwater robot captures the light source information through visual sensors such as cameras, the light source information can be processed to extract position information, such as distance or angle. This position information provides the underwater robot with an approximate positioning direction. When a single light source is used as a positioning mark, if there are materials with better reflective effects near the light source, these reflected lights may interfere with the underwater robot's judgment of the positioning light source, making it difficult to effectively distinguish the positioning light source from the interfering reflected light, thereby reducing the accuracy of positioning. Summary of the invention

[0004] The invention provides an underwater robot intelligent recovery positioning method to solve the problem that the existing reflected light affects the positioning accuracy.

[0005] The intelligent recovery and positioning method of an underwater robot of the present invention adopts the following technical scheme:

[0006] An embodiment of the present invention provides an underwater robot intelligent recovery and positioning method, the method comprising the following steps:

[0007] Acquire a water surface image based on a preset positioning light source;

[0008] Perform iterative threshold segmentation on the water surface image to obtain all suspected light source areas;

[0009] According to the difference in grayscale fluctuation degree and shape contour similarity between every two suspected light source areas, the similarity degree of every two suspected light source areas is obtained; according to the similarity degree of every two suspected light source areas, two similar light source groups are obtained; the convex polygon area surrounded by the centroids of all suspected light source areas in each similar light source group is recorded as the light source array polygon of each similar light source group; according to the shape regularity of the light source array polygon of each similar light source group, the possibility of each similar light source group being the actual positioned light source group is obtained, and the possibility of each similar light source group being the actual positioned light source group is recorded as the positioned light source group probability of each similar light source group; the similar light source group with the maximum positioned light source group probability is recorded as the actual light source group;

[0010] According to the actual light source group, the moving direction of the underwater robot during the recovery process is obtained to realize the intelligent recovery and positioning of the underwater robot.

[0011] Furthermore, the obtaining of all suspected light source areas includes:

[0012] An initial grayscale threshold, a noise threshold and a step size are preset, and the initial grayscale threshold is used as the current grayscale threshold for threshold segmentation to obtain several connected domains. The connected domains whose noise light source coefficients are less than the noise threshold in the threshold segmentation results are recorded as non-noise connected domains. When the number of non-noise connected domains is greater than twice the number of located light sources, the connected domains with the smallest noise light source coefficients that are twice the number of located light sources are recorded as suspected light source areas; when the number of non-noise connected domains is equal to twice the number of located light sources, all non-noise connected domains are recorded as suspected light source areas; when the number of non-noise connected domains is less than twice the number of located light sources, the grayscale value obtained by subtracting the step size from the current grayscale threshold is used as the new current grayscale threshold, and the water surface grayscale image is threshold segmented, and so on until all suspected light source areas are obtained.

[0013] Furthermore, the specific method for obtaining the noise light source coefficient is:

[0014] The number of pixels at the edge of each connected domain is recorded as the perimeter of the connected domain, and the number of all pixels in the connected domain is recorded as the area of ​​the connected domain; the ratio of the perimeter of each connected domain to the area of ​​each connected domain is normalized by the Sigmoid function, and the normalized result is recorded as the noise light source coefficient of each connected domain.

[0015] Furthermore, obtaining the similarity between every two suspected light source regions includes:

[0016] The similarity between the i-th suspected light source area and the j-th suspected light source area The calculation method is:

[0017]

[0018] Where: represents the Hu moment of the i-th suspected light source area, represents the Hu moment of the jth suspected light source area, Indicates the grayscale fluctuation degree of the i-th suspected light source area, Indicates the grayscale fluctuation degree of the jth suspected light source area; represents the hyperparameter, It means to find the cosine similarity function of two vectors. represents the absolute value function, Represents an exponential function with a natural constant as its base.

[0019] Furthermore, the specific method for obtaining the grayscale fluctuation degree is:

[0020] The pixels in the 8 neighborhoods of each pixel in the suspected light source area are recorded as the neighboring pixels of each pixel. The absolute value of the average of the differences in grayscale values ​​between each pixel in the i-th suspected light source area and all its neighboring pixels is taken as the neighborhood grayscale difference of each pixel in the i-th suspected light source area. The average of the neighborhood grayscale differences of all pixels in the i-th suspected light source area is obtained and recorded as the grayscale fluctuation degree of the i-th suspected light source area.

[0021] Further, the obtaining of two similar light source groups includes:

[0022] For all permutations and combinations of suspected light source areas, all suspected light source areas are divided into two light source groups. The average similarity group moments of each light source group under each permutation and combination are traversed, and the corresponding two light source groups under the permutation and combination when the sum of the average similarity group distances of the two light source groups is the largest are recorded as similar light source groups.

[0023] Furthermore, the specific method for obtaining the average similarity group distance is:

[0024] The average value of the similarity between every two suspected light source regions in the sth light source group is recorded as the average similarity group distance of the sth light source group.

[0025] Furthermore, the specific method for obtaining the probability of locating the light source group is:

[0026] The probability of locating the light source group of the s-th similar light source group The calculation method is:

[0027]

[0028] Where: represents the average similarity group distance of the s-th similar light source group, The number of edges of the light source array polygon representing the s-th similar light source group, represents the vertical point offset between the kth edge and the k+1th edge in the light source array polygon of the sth similar light source group; Represents an exponential function with a natural constant as its base.

[0029] Furthermore, the specific method for obtaining the vertical point deviation is:

[0030] Draw the perpendicular bisector of the kth side and the perpendicular bisector of the k+1th side in the light source array polygon of the sth similar light source group, and record the Euclidean distance between the intersection of the perpendicular bisector of the kth side and the perpendicular bisector of the k+1th side and the centroid of the light source array polygon of the sth similar light source group as the perpendicular point offset between the kth side and the k+1th side in the light source array polygon of the sth similar light source group.

[0031] Furthermore, obtaining the moving direction of the underwater robot during the recovery process includes:

[0032] Taking the center position of the water surface image as the current position and the centroid position of the light source array polygon of the actual light source group as the target moving position, the underwater robot moves vertically upward in the vertical direction and moves in the direction from the current position to the target moving position on the horizontal plane. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 The present invention is a flowchart of the steps of an underwater robot intelligent recovery and positioning method. DETAILED DESCRIPTION

[0035] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of an underwater robot intelligent recovery and positioning method proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0036] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0037] The specific scheme of the underwater robot intelligent recovery and positioning method provided by the present invention is described in detail below with reference to the accompanying drawings.

[0038] See also Figure 1 , which shows a flowchart of a method for intelligent recovery and positioning of an underwater robot provided by an embodiment of the present invention, the method comprising the following steps:

[0039] Step S001: Acquire a water surface image based on a preset positioning light source.

[0040] When recovering the underwater robot, a positioning light source installed on the surface of a positioning plate above the water surface is used to enable the underwater robot to capture a water surface image underwater. The movement direction of the underwater robot during recovery is determined by the position of the positioning light source in the water surface image, so that the underwater robot moves toward the position of the positioning plate. The positioning plate is usually installed above the water surface on the shore, with the side of the positioning plate having the positioning light source facing the water surface and parallel to the water surface. To prevent the stability of the positioning plate from being affected by the water flow, the positioning plate is at a certain height from the water surface. Every time positioning is required, the positioning light source will emit strong light, and the underwater robot will achieve intelligent positioning by capturing water surface images and analyzing the position of the light source. Since the water surface will refract and reflect the positioning light source, the water surface image captured by the underwater robot will contain the positioning light source and a pseudo light source generated by the reflection and refraction of the positioning light source by the water surface, and the water surface image needs to be further analyzed.

[0041] The purpose of this embodiment is to use multiple positioning light sources to realize the intelligent recovery positioning of the underwater robot. Therefore, a reflective plate is first installed on the shore. The positioning plate contains several positioning light sources. The underwater robot collects water surface images through a camera;

[0042] It should be noted that the present embodiment sets four positioning light sources, which are respectively installed at the four vertices of a square with a side length of 1 meter on the surface of the positioning plate. Other embodiments may set other positioning light source installation methods, which are not specifically limited in the present embodiment.

[0043] Step S002: Perform iterative threshold segmentation on the water surface image to obtain all suspected light source areas.

[0044] It should be noted that, compared with the non-positioned light source and pseudo-light source areas, the positioned light source and pseudo-light source appear as areas with higher grayscale values ​​in the water surface image, so the positioned light source and pseudo-light source areas are segmented from the water surface image through threshold segmentation; since part of the water body around the positioned light source and pseudo-light source areas is also illuminated, directly using the Otsu threshold segmentation method to perform threshold segmentation on the water surface image will cause the segmented positioned light source and pseudo-light source areas to be too large or irregular, affecting the accuracy of subsequent analysis, so by gradually lowering the threshold, several connected domains with shapes close to circles in the water surface image are obtained as suspected light source areas, so as to achieve accurate segmentation of the positioned light source and pseudo-light source areas; since a positioned light source will generate a pseudo-light source under the reflection and refraction of the water surface, the number of suspected light source areas finally obtained is twice the number of positioned light sources.

[0045] Specifically, the water surface image is averaged grayed, and the grayed result is recorded as the water surface gray image; the initial gray threshold, noise threshold and step size are preset, and the initial gray threshold is used as the current gray threshold to perform threshold segmentation on the water surface gray image to obtain the threshold segmentation result, the pixel points with gray values ​​greater than the current gray threshold belong to the foreground area, the gray values ​​of the pixels in the foreground area remain unchanged, the pixel points with gray values ​​less than or equal to the current gray threshold belong to the background area, and the gray values ​​of the pixels in the background area are set to 0; the set of pixels whose gray values ​​are not 0 and are connected to each other is recorded as a connected domain, and all connected domains in the foreground area of ​​the threshold segmentation result are obtained; the number of pixels at the edge of each connected domain is recorded as the perimeter of the connected domain, and the number of all pixels in the connected domain is recorded as the area of ​​the connected domain; the ratio of the perimeter of each connected domain to the area of ​​each connected domain is normalized by the Sigmoid function, and the normalized result is recorded as the noise light source coefficient of each connected domain;

[0046] The connected domains whose noise light source coefficients in the threshold segmentation results are less than the noise threshold are recorded as non-noise connected domains. When the number of non-noise connected domains is greater than twice the number of located light sources, the connected domains with the smallest noise light source coefficients that are twice the number of located light sources are recorded as suspected light source areas. When the number of non-noise connected domains is equal to twice the number of located light sources, all non-noise connected domains are recorded as suspected light source areas. When the number of non-noise connected domains is less than twice the number of located light sources, the grayscale value obtained by subtracting the step size from the current grayscale threshold is used as the new current grayscale threshold, and the water surface grayscale image is threshold segmented, and this process is repeated until all suspected light source areas are obtained.

[0047] In this embodiment, the initial grayscale threshold is preset to be 250, the step size is 5, and the noise threshold is 0.6. Other embodiments may set it to other values, which are not specifically limited in this embodiment.

[0048] Step S003, obtaining two similar light source groups; obtaining the possibility that each similar light source group is an actually positioned light source group; and obtaining an actual light source group.

[0049] It should be noted that the suspected light source area includes two types of light sources, namely, positioned light sources and pseudo light sources. The similarity between light sources of the same type is high, while the similarity between light sources of different types is low. The similarity between every two suspected light source areas is obtained based on the Hu moment and grayscale fluctuation degree of the suspected light source area. The Hu moment is invariant to the shape and contour features of the image. By comparing the similarity of the Hu moments of two suspected light source areas, the similarity of the shapes and contour features of the two suspected light source areas can be reflected. The more similar the Hu moments of the two suspected light source areas are, the more similar the shapes and contour features of the two suspected light source areas are, and the greater the similarity between the two suspected light source areas. The smaller the difference in grayscale fluctuation degree is, the greater the similarity between the two suspected light source areas. All suspected light source areas are grouped according to the similarity between every two suspected light source areas. Since the number of suspected light source areas is equal to twice the number of positioned light sources, the two light source groups with the largest average similarity within the group are obtained, and the number of light source areas contained in each group is equal to the number of positioned light sources.

[0050] Specifically, the pixel points of the 8 neighborhoods of each pixel point in the suspected light source area are recorded as the neighboring pixel points of each pixel point, the average value of the difference between the grayscale values ​​of each pixel point in the i-th suspected light source area and all the neighboring pixel points of the pixel point is taken as the neighborhood grayscale difference of each pixel point in the i-th suspected light source area, the average value of the neighborhood grayscale difference of all pixels in the i-th suspected light source area is obtained, and recorded as the grayscale fluctuation degree of the i-th suspected light source area, and the grayscale fluctuation degree of all suspected light source areas is obtained in the same way; the similarity between the i-th suspected light source area and the j-th suspected light source area The calculation method is:

[0051]

[0052] Where: represents the Hu moment of the i-th suspected light source area, represents the Hu moment of the jth suspected light source area, Indicates the grayscale fluctuation degree of the i-th suspected light source area, Indicates the grayscale fluctuation degree of the jth suspected light source area; represents a hyperparameter. In this embodiment, x=1 is used to avoid the denominator being 0. It means to find the cosine similarity function of two vectors. represents the absolute value function, represents an exponential function with a natural constant as base;

[0053] It should be noted that It represents the cosine similarity between the Hu moment of the i-th suspected light source area and the Hu moment of the j-th suspected light source area. The Hu moment reflects the shape and contour characteristics of the area. The closer the cosine similarity is to 1, the The smaller the value of , the more similar the Hu moment of the i-th suspected light source area is to the Hu moment of the j-th suspected light source area, and the more similar the i-th suspected light source area is to the j-th suspected light source area; It reflects the difference in the fluctuation degree of the grayscale values ​​of all pixels in the i-th suspected light source area and the j-th suspected light source area. The smaller the difference in the fluctuation degree, the closer the value is to 1. The smaller the value of is, the more similar the i-th suspected light source area is to the j-th suspected light source area. It should be noted that Hu moments (image moments) are well-known technologies and will not be described in detail in this embodiment.

[0054] Similarly, the similarity between every two suspected light source areas is obtained; for all permutations and combinations of suspected light source areas, all suspected light source areas are divided into two light source groups, the number of suspected light source areas in each light source group is the same as the number of located light sources, and the average value of the similarity between every two suspected light source areas in the light source group is recorded as the average similarity group distance of the light source group, and the average similarity group moment of each light source group under each permutation and combination is traversed, and the corresponding two light source groups under the permutation and combination when the sum of the average similarity group distances of the two light source groups is the largest are recorded as similar light source groups.

[0055] It should be further explained that the pseudo light source is composed of inverted images formed by reflection. Since the light is refracted and the reflection angle changes when passing through the water surface, the shape formed by the suspected light source area in the similar light source group composed of the pseudo light source will be greatly stretched, while the shape formed by the suspected light source area in the similar light source group composed of the locating light source is less deformed by water waves than the pseudo light source. Therefore, the probability of the locating light source group of each similar light source group is obtained by the regularity of the polygons composed of all the suspected light source areas in each similar light source group. The more regular the polygons composed of all the suspected light source areas in the similar light source group, the more likely it is that the similar light source group is a similar light source group composed of locating light sources. The similar light source group with the highest probability of obtaining a locating light source group is recorded as the actual light source group.

[0056] Specifically, the convex polygonal area enclosed by the centroids of all suspected light source areas in each similar light source group is recorded as the light source array polygon of each similar light source group; the possibility that the s-th similar light source group is the actual positioned light source group is recorded as the positioned light source group probability of the s-th similar light source group, and the positioned light source group probability of the s-th similar light source group is recorded as The calculation method is:

[0057]

[0058] Where: represents the average similarity group distance of the s-th similar light source group, The number of edges of the light source array polygon representing the s-th similar light source group, The perpendicular point offset between the kth side and the k+1th side in the light source array polygon of the sth similar light source group is represented, and the method for obtaining the perpendicular point offset between the kth side and the k+1th side is as follows: draw the perpendicular bisector of the kth side and the perpendicular bisector of the k+1th side in the light source array polygon of the sth similar light source group, and record the Euclidean distance between the intersection of the perpendicular bisector of the kth side and the perpendicular bisector of the k+1th side and the centroid of the light source array polygon of the sth similar light source group as the perpendicular point offset between the kth side and the k+1th side in the light source array polygon of the sth similar light source group; represents an exponential function with a natural constant as base;

[0059] It should be noted that Represents the average similarity group distance of the s-th similar light source group. The value is greater than 0. The larger the value, the greater the similarity between each two suspected light source areas in the s-th similar light source group, and the greater the probability of locating the light source group of the s-th similar light source group. Indicates the regularity of the light source array polygon of the s-th similar light source group. In this embodiment, the polygon formed by the positioning light sources is a square, and the vertical point offset of the adjacent sides of the square is 0. Therefore, the smaller the value, the greater the probability of the positioning light source group of the s-th similar light source group;

[0060] Similarly, the positioning light source group probability of each similar light source group is obtained, and the similar light source group with the largest positioning light source group probability is recorded as the actual light source group.

[0061] Step S004: According to the actual light source group, the moving direction of the underwater robot during the recovery process is obtained to achieve intelligent recovery and positioning of the underwater robot.

[0062] Taking the center position of the water surface image as the current position and the centroid position of the light source array polygon of the actual light source group as the target moving position, the underwater robot moves vertically upward in the vertical direction and moves in the direction from the current position to the target moving position on the horizontal plane; the movement mode and movement rules of the underwater robot are existing public technologies, and this embodiment only analyzes the movement direction and target.

[0063] A preset positioning interval is used. Every time a positioning interval occurs, the underwater robot takes a water surface image and re-determines the current position and the target moving position until the underwater robot returns to the water surface, thereby realizing the intelligent recovery positioning of the underwater robot. The preset positioning interval in this embodiment is 10 seconds, and other embodiments may be set to other values, which are not specifically limited in this embodiment.

[0064] It should be noted that the The model only shows negative correlation and the output of the constraint model is in In the range, As the input of this model, it can be replaced by other models with the same purpose in specific implementation. The model is used as an example for description without any specific limitation.

[0065] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for intelligent recovery and positioning of an underwater robot, characterized in that: The method comprises the following steps: Acquire a water surface image based on a preset positioning light source; Perform iterative threshold segmentation on the water surface image to obtain all suspected light source areas; According to the difference in grayscale fluctuation degree and shape contour similarity between every two suspected light source areas, the similarity degree of every two suspected light source areas is obtained; according to the similarity degree of every two suspected light source areas, two similar light source groups are obtained; the convex polygon area surrounded by the centroids of all suspected light source areas in each similar light source group is recorded as the light source array polygon of each similar light source group; according to the shape regularity of the light source array polygon of each similar light source group, the possibility of each similar light source group being the actual positioned light source group is obtained, and the possibility of each similar light source group being the actual positioned light source group is recorded as the positioned light source group probability of each similar light source group; the similar light source group with the maximum positioned light source group probability is recorded as the actual light source group; According to the actual light source group, the moving direction of the underwater robot during the recovery process is obtained to realize the intelligent recovery positioning of the underwater robot; The obtaining of two similar light source groups comprises: For all permutations and combinations of suspected light source areas, all suspected light source areas are divided into two light source groups, and the average similarity group moments of each light source group under each permutation and combination are traversed. The two light source groups corresponding to the permutation and combination when the sum of the average similarity group moments of the two light source groups is the largest are recorded as similar light source groups; The specific method for obtaining the average similarity group distance is: The average value of the similarity between every two suspected light source regions in the sth light source group is recorded as the average similarity group distance of the sth light source group.

2. According to claim 1, the method for intelligent recovery and positioning of an underwater robot is characterized in that: The obtaining of all suspected light source areas includes: An initial grayscale threshold, a noise threshold and a step size are preset, and the initial grayscale threshold is used as the current grayscale threshold for threshold segmentation to obtain several connected domains. The connected domains whose noise light source coefficients are less than the noise threshold in the threshold segmentation results are recorded as non-noise connected domains. When the number of non-noise connected domains is greater than twice the number of located light sources, the connected domains with the smallest noise light source coefficients that are twice the number of located light sources are recorded as suspected light source areas; when the number of non-noise connected domains is equal to twice the number of located light sources, all non-noise connected domains are recorded as suspected light source areas; when the number of non-noise connected domains is less than twice the number of located light sources, the grayscale value obtained by subtracting the step size from the current grayscale threshold is used as the new current grayscale threshold, and the water surface grayscale image is threshold segmented, and so on until all suspected light source areas are obtained.

3. According to claim 2, the method for intelligent recovery and positioning of an underwater robot is characterized in that: The specific method for obtaining the noise light source coefficient is: The number of pixels at the edge of each connected domain is recorded as the perimeter of the connected domain, and the number of all pixels in the connected domain is recorded as the area of ​​the connected domain; the ratio of the perimeter of each connected domain to the area of ​​each connected domain is normalized by the Sigmoid function, and the normalized result is recorded as the noise light source coefficient of each connected domain.

4. According to claim 1, the method for intelligent recovery and positioning of an underwater robot is characterized in that: The obtaining of the similarity between each two suspected light source regions includes: The similarity between the i-th suspected light source area and the j-th suspected light source area The calculation method is: Where: represents the Hu moment of the i-th suspected light source area, represents the Hu moment of the jth suspected light source area, Indicates the grayscale fluctuation degree of the i-th suspected light source area, Indicates the grayscale fluctuation degree of the jth suspected light source area; represents the hyperparameter, It means to find the cosine similarity function of two vectors. represents the absolute value function, Represents an exponential function with a natural constant as its base.

5. According to claim 4, the method for intelligent recovery and positioning of an underwater robot is characterized in that: The specific method for obtaining the grayscale fluctuation degree is: The pixels in the 8 neighborhoods of each pixel in the suspected light source area are recorded as the neighboring pixels of each pixel. The absolute value of the average of the differences in grayscale values ​​between each pixel in the i-th suspected light source area and all its neighboring pixels is taken as the neighborhood grayscale difference of each pixel in the i-th suspected light source area. The average of the neighborhood grayscale differences of all pixels in the i-th suspected light source area is obtained and recorded as the grayscale fluctuation degree of the i-th suspected light source area.

6. The method for intelligent recovery and positioning of an underwater robot according to claim 1, characterized in that: The specific method for obtaining the probability of locating the light source group is: The probability of locating the light source group of the s-th similar light source group The calculation method is: Where: represents the average similarity group distance of the s-th similar light source group, The number of edges of the light source array polygon representing the s-th similar light source group, represents the vertical point offset between the kth edge and the k+1th edge in the light source array polygon of the sth similar light source group; Represents an exponential function with a natural constant as its base.

7. The method for intelligent recovery and positioning of an underwater robot according to claim 6, characterized in that: The specific method for obtaining the vertical point deviation is: Draw the perpendicular bisector of the kth side and the perpendicular bisector of the k+1th side in the light source array polygon of the sth similar light source group, and record the Euclidean distance between the intersection of the perpendicular bisector of the kth side and the perpendicular bisector of the k+1th side and the centroid of the light source array polygon of the sth similar light source group as the perpendicular point offset between the kth side and the k+1th side in the light source array polygon of the sth similar light source group.

8. The method for intelligent recovery and positioning of an underwater robot according to claim 1, characterized in that: The obtaining of the moving direction of the underwater robot during the recovery process includes: Taking the center position of the water surface image as the current position and the centroid position of the light source array polygon of the actual light source group as the target moving position, the underwater robot moves vertically upward in the vertical direction and moves in the direction from the current position to the target moving position on the horizontal plane.

Images