An underwater positioning method for underwater sampling robot based on sonar image processing

Abstract

The present invention provides an underwater positioning method for an underwater sampling robot based on sonar image processing, which belongs to the technical field of underwater robot real-time positioning. The method includes the following steps: collecting the weighted azimuth contour of the underwater sampling robot in advance, obtaining a coded contour line image source, using sonar to continuously collect seabed images, performing contour processing on the collected images, comparing the seabed pattern contour line image with the contour line image source to find out whether there is an underwater sampling robot contour, determining the position of the underwater sampling robot, and tracking the underwater sampling robot in real time. By collecting the omnidirectional contour of the underwater sampling robot in advance, and then obtaining a comparison image source, and then when collecting the image, the closed-loop contour is processed in a polar coordinate manner, and the image similarity is obtained according to two discrete Fourier and inverse Fourier changes, so that the entire calculation process is simpler, the processing speed is faster, the calculation amount is smaller, and real-time positioning and tracking is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater robot real-time positioning, and in particular to an underwater positioning method for an underwater sampling robot based on sonar image processing. Background Art

[0002] Underwater sampling robots are devices used to detect the underwater world, but when they sink into deep water, they often cannot communicate directly with the water surface and cannot be located in real time. With the development of sonar technology, image sonar has become the main equipment and technical means for searching robots on the lake bottom, but in actual applications, there are still technical and usage deficiencies in detecting robots through sonar images. First, compared with optical images, the acoustic imaging structure is rough and the fineness is poor. The low contrast of the target and the unclear features are the fatal defects of sonar images. It often requires professionally trained personnel to carefully interpret the sonar image to identify the robot; secondly, sonar images generally have a search field width of nearly one thousand meters, while the maximum scale of most buried equipment is generally only a few meters. It is difficult for the human eye to quickly find the image target within such a large imaging range; thirdly, in terms of usage, due to the lack of positioning auxiliary system of image sonar, when a suspicious target is found, it is often necessary to repeatedly search and confirm the target position.

[0003] In addition, if we only rely on manual reading of images and search for underwater robots through image sonar, the correct recognition rate of the robot is not only affected by the performance of detection equipment such as sonar images, but also closely related to whether the equipment operator is experienced. In general, the current robot detection method requires very professional sonar image interpreters to complete the work. Due to the inevitable changes in personnel and positions, it often takes a lot of energy and cost to train personnel. In addition, the professional knowledge and experience of interpreting image sonar are difficult to replicate, and it is difficult to cope with the situation of a sharp increase in workload in a short period of time. There are problems such as high cost and low efficiency. On the other hand, the level and accuracy of personnel's interpretation of sonar images are not only directly related to knowledge and experience, but also to unquantifiable factors such as physical state, mental and emotional state during work, which is easy to randomly lead to misjudgment and wrong judgment, and there are problems such as high randomness and uncontrollable errors. In addition, due to the lack of effective robot positioning means, there are also problems such as much repetitive work and low efficiency. Therefore, it is necessary to design an underwater positioning method for underwater sampling robots based on sonar image processing. Summary of the invention

[0004] The purpose of the present invention is to provide an underwater positioning method for an underwater sampling robot based on sonar image processing to solve the technical problems mentioned in the background technology.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] An underwater positioning method for an underwater sampling robot based on sonar image processing, the method comprising the following steps:

[0007] Step 1: Obtain a full range of sonar images of the underwater sampling robot;

[0008] Step 2: Identify the edge contours and inner lines of the edge contours of all sonar images to obtain contour line drawings;

[0009] Step 3: Collect all contour line drawings, and then encode the contour line drawings to obtain encoded contour line drawing sources;

[0010] Step 4: Use sonar to continuously collect seafloor images;

[0011] Step 5: Perform contour processing on the collected image to obtain a contour line map of the seabed pattern;

[0012] Step 6: Compare the seabed pattern contour line map with the contour line map source to find out whether there is an underwater sampling robot outline;

[0013] Step 7: Compare the robot outline with the outline line source for a second time to determine the underwater sampling robot;

[0014] Step 8: Determine the position of the underwater sampling robot, adjust the position of the sonar image acquisition device, and track the underwater sampling robot in real time.

[0015] Furthermore, the specific process of step 1 is: placing the underwater sampling robot into the deep sea, and then determining the precise position of the underwater sampling robot through manual positioning, wherein a flat seabed area is selected, and a sonar collecting vessel is used to collect sonar images underwater on the water surface, and the sonar collecting vessel collects sonar images directly above the water surface, and then the underwater sampling robot performs circular motion with the position of the robot as the center, and at the same time, the sonar is turned on to collect seabed images to obtain a sonar image set, wherein the radius of the circular motion increases from small to large, and each image collected by the circular motion is saved separately, and the radius of the circular motion and the depth of the sea water are set.

[0016] Furthermore, the specific process of step 2 is:

[0017] All sonar image sets are processed by recognition. The recognition process is to recognize the contours and edge lines of the image to obtain a contour atlas. Then the contour of the underwater sampling machine is manually marked, while the lines within the contour are retained to obtain a contour line map.

[0018] Further, the process of step 3 is: take the collected circle radius as the folder name, put the corresponding collection circle contour line drawing into the folder, and then encode the specific position information of the underwater sampling robot where the collection ship is located at the time of collection on the corresponding contour line drawing, and under each contour line drawing there is the position information of the collection point relative to the underwater sampling robot, and all the coded collection position information contour line drawings are summarized as the coded contour line drawing source;

[0019] Place each image of the coded contour line source on the two-dimensional coordinate axis, select the contour of the underwater sampling robot as the center point corresponding to the origin of the two-dimensional coordinate axis, and then set the contour of the underwater sampling robot to k points, where k is an integer multiple greater than 64. The polar coordinates of the points set on the contour are n is a positive integer greater than or equal to k. n Extract the polar coordinate function l(k) = (r1, r2, r3···r n ), then the polar coordinate function l of all the coded contour line source is obtained t (k)=(r1,r2,r3···r n ), t is the number of coded contour line graphs, which is a positive integer.

[0020] Furthermore, the specific process of step 4 is: when the underwater sampling robot is placed in the deep sea for sampling, the sonar collection ship is turned on to collect images of the seabed of the underwater sampling robot on the water surface, and a single-hole sonar collection device is used on the sonar collection ship for collection.

[0021] Furthermore, the specific process of step 5 is: pre-processing the collected seabed image to obtain a grayscale image, and then extracting contours and lines from the grayscale image to obtain a seabed pattern contour line map. When performing line processing, when the difference between the pixel points of two separated points is greater than 5, it is considered that there is a line interval between the two points.

[0022] Furthermore, the specific process of step 6 is: contour tracking and recognition of the seabed pattern contour line drawing, marking the recognized closed loop contour, selecting the closed loop contour, and then placing the closed loop contour on the two-dimensional coordinate axis, making the center point of the closed loop contour correspond to the origin of the two-dimensional coordinate axis, setting the closed loop contour to k points, k is an integer multiple value greater than 64, and the polar coordinates of the points set on the contour are obtained as follows: n is a positive integer greater than or equal to k. n Extract the polar coordinate function l, (k) = (r1', r2', r3'···r n '), then l, (k) = (r1', r2', r3'···r n') and the polar coordinate function l of all coded contour line source t (k)=(r1,r2,r3···r n )Fourier transform is performed to obtain a discrete function l”(k), and then the discrete function l”(k) is inversely Fourier transformed to obtain a contour similarity function, and the maximum value of the function is found. When the maximum value is greater than or equal to the set value, the contour of the closed loop is determined to be the contour of the underwater sampling robot.

[0023] Furthermore, the specific process of step 7 is: comparing the lines inside the determined contour of the underwater sampling robot with the lines inside the contour of the coded contour line source, and when the comparison similarity of the internal lines is greater than the set value, it is reconfirmed as the contour of the underwater sampling robot.

[0024] Furthermore, the specific process of step 8 is: reversely compare the determined outline of the underwater sampling robot with the coded outline line image source to obtain the coded outline line image with the greatest similarity, and then parse the direction angle of the coded outline line image to confirm the direction of the underwater sampling robot, and then collect the outline of the underwater sampling robot in the entire image for the next time to determine the position of the underwater sampling robot. The position change of the outline of the underwater sampling robot in the previous image is used to determine the moving position of the underwater sampling robot, and then repeatedly collect image recognition to track and locate the underwater sampling robot.

[0025] The present invention has the following beneficial effects due to the adoption of the above technical solution:

[0026] The present invention collects the omnidirectional contour of the underwater sampling robot in advance, and then obtains a comparative image source. Then, when collecting the image, the closed-loop contour is processed in a polar coordinate manner, and the image similarity is obtained based on two discrete Fourier and inverse Fourier changes. This makes the entire calculation process simpler, the processing speed is faster, the calculation amount is smaller, and real-time positioning and tracking is achieved. The comparison speed is more than 5 times faster than that of traditional network neural training, and accurate underwater sampling robot positioning and tracking can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and preferred embodiments. However, it should be noted that many details listed in the specification are only for the purpose of enabling the reader to have a thorough understanding of one or more aspects of the present invention, and these aspects of the present invention can be implemented even without these specific details.

[0029] like Figure 1As shown, an underwater positioning method for an underwater sampling robot based on sonar image processing comprises the following steps:

[0030] Step 1: Obtain all-round sonar images of the underwater sampling robot. Place the underwater sampling robot in the deep sea, and then determine the precise position of the underwater sampling robot through manual positioning. Select a flat seabed area, use a sonar collection ship to collect sonar images underwater on the water surface, and the sonar collection ship collects sonar images directly above the water surface. Then, the underwater sampling robot moves in a circle with the position of the robot as the center. At the same time, the sonar is turned on to collect seabed images to obtain a sonar image set. The radius of the circular motion ranges from small to large. The images collected by each circular motion are saved separately, and the radius of the circular motion and the depth of the seawater are set. Generally, the radius of a circle collected differs by about 1 kilometer, and then the image is recognized.

[0031] Step 2: Identify the edge contours and internal lines of all sonar images to obtain contour line drawings. All sonar image sets are identified and processed. The identification process is to identify the contours and edge lines of the image to obtain a contour atlas. Then, the contour of the underwater sampling machine is manually marked, and the lines within the contour are retained to obtain a contour line drawing. The lines within the contour are used as image comparison evidence for secondary comparison confirmation, making the subsequent comparison more accurate, which is more accurate than traditional comparison technology.

[0032] Step 3: Collect all contour line drawings, and then encode the contour line drawings to obtain the coded contour line drawing source. Use the radius of the collected circle as the folder name, put the contour line drawings of the corresponding collected circle into the folder, and then encode the specific position information of the underwater sampling robot where the collection ship is located during the collection on the corresponding contour line drawing. Under each contour line drawing, there is the position information of the collection point relative to the underwater sampling robot, and all the coded collection position information contour line drawings are summarized as the coded contour line drawing source. Each folder is named with a radius, such as 2 kilometers, and the pictures in it are the images collected with a radius of 2 kilometers.

[0033] Place each image of the coded contour line source on the two-dimensional coordinate axis, select the contour of the underwater sampling robot as the center point corresponding to the origin of the two-dimensional coordinate axis, and then set the contour of the underwater sampling robot to k points, where k is an integer multiple greater than 64. The polar coordinates of the points set on the contour are n is a positive integer greater than or equal to k. n Extract the polar coordinate function l(k) = (r1, r2, r3···r n ), then the polar coordinate function l of all the coded contour line source is obtained t(k)=(r1,r2,r3···r n ), t is the number of coded contour line graphs, which is a positive integer. k is generally set to 256, but can be larger. A larger value will result in higher accuracy, but the amount of calculation will be more complex. The value of k is proportional to the amount of calculation and accuracy. After testing and comparison, taking 256 can achieve an accuracy of about 98%, which can already meet the positioning requirements well.

[0034] Step 4: Use sonar to continuously collect seabed images. When the underwater sampling robot is placed in the deep sea for sampling, the sonar collection ship is turned on to collect images of the seabed of the underwater sampling robot on the water surface. The sonar collection ship uses a single-hole sonar collection device for collection.

[0035] Step 5: Perform contour processing on the collected image to obtain a contour line map of the seabed pattern. The collected seabed image is preprocessed to obtain a grayscale image, and then contours and lines are extracted from the grayscale image to obtain a contour line map of the seabed pattern. When performing line processing, when the difference between the pixels of two separated points is greater than 5, it is considered that there is a line interval between the two points.

[0036] Step 6: Compare the seabed pattern contour line drawing with the contour line drawing source to find out whether there is an underwater sampling robot contour. Perform contour tracking and recognition on the seabed pattern contour line drawing, mark the recognized closed loop contour, select the closed loop contour, and then put the closed loop contour on the two-dimensional coordinate axis, correspond the center point of the closed loop contour to the origin of the two-dimensional coordinate axis, set the closed loop contour to k points, k is an integer multiple value greater than 64, and the polar coordinates of the points set on the contour are obtained as n is a positive integer greater than or equal to k. n Extract the polar coordinate function l'(k) = (r1', r2', r3'···r n '), then l, (k) = (r1', r2', r3'···r n ') and the polar coordinate function l of all coded contour line source t (k)=(r1,r2,r3···r n )Fourier transform is performed to obtain a discrete function l”(k), and then the discrete function l”(k) is inversely Fourier transformed to obtain a contour similarity function, and the maximum value of the function is found. When the maximum value is greater than or equal to the set value, the contour of the closed loop is determined to be the contour of the underwater sampling robot.

[0037] Step 7: The robot outline is compared with the outline line source for a second time to determine the underwater sampling robot. The lines inside the outline of the determined underwater sampling robot are compared with the lines inside the outline of the coded outline line source. When the comparison similarity of the internal lines is greater than the set value, it is confirmed as the outline of the underwater sampling robot for a second time.

[0038] Step 8: Determine the orientation of the underwater sampling robot, adjust the orientation of the sonar image acquisition device, and track the underwater sampling robot in real time. The determined contour of the underwater sampling robot is reversely compared with the coded contour line map source to obtain the coded contour line map with the greatest similarity, and then the direction angle of the coded contour line map is parsed to confirm the direction of the underwater sampling robot. Then, in the next acquisition, the position of the underwater sampling robot's contour in the entire image and the position change of the underwater sampling robot's contour in the previous image are determined to determine the moving position of the underwater sampling robot, and then the image recognition is repeatedly collected to track and locate the underwater sampling robot. According to the direction, the location where the next position will appear can be predicted, so that the next contour recognition can be more accurate. At the same time, when recognizing the closed-loop contour, the closed-loop wheel library of the predicted point is used as the preferred comparison contour, which greatly improves the comparison efficiency.

[0039] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An underwater positioning method for an underwater sampling robot based on sonar image processing, characterized in that: The method comprises the following steps: Step 1: Obtain a full range of sonar images of the underwater sampling robot; Step 2: Identify the edge contours and inner lines of the edge contours of all sonar images to obtain contour line drawings; Step 3: Collect all contour line drawings, and then encode the contour line drawings to obtain encoded contour line drawing sources; Step 4: Use sonar to continuously collect seafloor images; Step 5: Perform contour processing on the collected image to obtain a contour line map of the seabed pattern; Step 6: Compare the seabed pattern contour line map with the contour line map source to find out whether there is an underwater sampling robot outline; Step 7: Compare the robot outline with the outline line source for a second time to determine the underwater sampling robot; Step 8: Determine the position of the underwater sampling robot, adjust the position of the sonar image acquisition device, and track the underwater sampling robot in real time. The process of step 3 is: take the radius of the collected circle as the folder name, put the corresponding contour line drawing of the collected circle into the folder, and then encode the specific position information of the underwater sampling robot where the collection ship is located during the collection on the corresponding contour line drawing. Under each contour line drawing is the position information of the collection point relative to the underwater sampling robot, and all the coded collection position information contour line drawings are summarized as the coded contour line drawing source; Place each image of the coded contour line source on the two-dimensional coordinate axis, select the contour of the underwater sampling robot as the center point corresponding to the origin of the two-dimensional coordinate axis, and then set the contour of the underwater sampling robot to k points, where k is an integer multiple greater than 64. The polar coordinates of the points set on the contour are: n is a positive integer greater than or equal to k. n Extract the polar coordinate function l(k) = (r1, r2, r3···r n ), then the polar coordinate function l of all the coded contour line source is obtained t (k)=(r1,r2,r3···r n ), t is the number of coded contour line graphs, which is a positive integer; The specific process of step 6 is: track and identify the contour of the seabed pattern line drawing, mark the identified closed loop contour, select the closed loop contour, and then put the closed loop contour on the two-dimensional coordinate axis, correspond the center point of the closed loop contour to the origin of the two-dimensional coordinate axis, set the closed loop contour to k points, k is an integer multiple value greater than 64, and the polar coordinates of the points set on the contour are obtained as follows: n is a positive integer greater than or equal to k. Extract it to get the polar coordinate function Then Polar coordinate function l with all coded contour line source t (k)=(r1,r2,r3···r n ) Fourier transform to obtain the discrete function l ,, (k), then the discrete function l ,, (k) Perform an inverse Fourier transform to obtain a contour similarity function, find the maximum value of the function, and when the maximum value is greater than or equal to a set value, determine that the closed-loop contour is the contour of the underwater sampling robot.

2. The underwater positioning method of an underwater sampling robot based on sonar image processing according to claim 1 is characterized in that: The specific process of step 1 is: placing the underwater sampling robot into the deep sea, and then determining the precise position of the underwater sampling robot through manual positioning, wherein a flat seabed area is selected, and a sonar collection vessel is used to collect sonar images underwater on the water surface, and the sonar collection vessel collects sonar images directly above the water surface, and then the underwater sampling robot performs circular motion with the position of the robot as the center, and at the same time, the sonar is turned on to collect seabed images to obtain a sonar image set, wherein the radius of the circular motion increases from small to large, and each image collected by the circular motion is saved separately, and the radius of the circular motion and the depth of the sea water are set.

3. The underwater positioning method of an underwater sampling robot based on sonar image processing according to claim 2 is characterized in that: The specific process of step 2 is: All sonar image sets are processed by recognition. The recognition process is to recognize the contours and edge lines of the image to obtain a contour atlas. Then the contour of the underwater sampling machine is manually marked, while the lines within the contour are retained to obtain a contour line map.

4. The underwater positioning method of an underwater sampling robot based on sonar image processing according to claim 3 is characterized in that: The specific process of step 4 is: when the underwater sampling robot is placed in the deep sea for sampling, the sonar collection ship is turned on to collect images of the seabed of the underwater sampling robot on the water surface, and a single-hole sonar collection device is used on the sonar collection ship for collection.

5. The underwater positioning method of an underwater sampling robot based on sonar image processing according to claim 4 is characterized in that: The specific process of step 5 is: pre-process the collected seabed image to obtain a grayscale image, then extract the contour and lines of the grayscale image to obtain a seabed pattern contour line map. When performing line processing, when the difference between the pixel points of two separated points is greater than 5, it is considered that there is a line interval between the two points.

6. The underwater positioning method of an underwater sampling robot based on sonar image processing according to claim 5, characterized in that: The specific process of step 7 is: compare the lines inside the determined contour of the underwater sampling robot with the lines inside the contour of the coded contour line source. When the comparison similarity of the internal lines is greater than the set value, it is reconfirmed as the contour of the underwater sampling robot.

7. The underwater positioning method of an underwater sampling robot based on sonar image processing according to claim 6 is characterized in that: The specific process of step 8 is: reversely compare the determined outline of the underwater sampling robot with the coded outline line image source to obtain the coded outline line image with the greatest similarity, and then parse the direction angle of the coded outline line image to confirm the direction of the underwater sampling robot, and then collect the outline of the underwater sampling robot in the entire image for the next time, and determine the position change of the outline of the underwater sampling robot in the previous image to determine the moving position of the underwater sampling robot, and then repeatedly collect image recognition to track and locate the underwater sampling robot.

Images