Forward-looking sonar target online identification method and system in extremely shallow water environment

Abstract

The invention discloses a forward-looking sonar target online identification method and system in an extremely shallow water environment, and relates to the field of underwater target identification. The objective of the invention is to solve the problems of few detail features, fuzzy contour, extremely many imaging background false targets and high target similarity of existing underwater sonar detection shoal and near-shore targets. Comprising the following steps: S1, acquiring a sonar image in a shallow water environment based on a foresight sonar, and denoising the image in combination with median filtering and wavelet transform; s2, performing optimization processing on the filtered image by adopting fuzzy enhancement, segmentation and morphological processing, and extracting the contour of a target; s3, according to the position of the suspected target in the current frame of sonar image, assuming that the motion of the platform between the front frame and the rear frame is uniform linear motion, and calculating the position of the next frame of suspected target in combination with the attitude and motion information of the platform; and associating the position of the suspected target in the next frame of sonar image with the calculated position data of the suspected target through a nearest neighbor data association algorithm to obtain a real target.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater target recognition, and in particular to an online recognition method and system for forward-looking sonar targets in an extremely shallow water environment. Background Art

[0002] The vast ocean contains rich resources and has huge development potential. With the development of the world, marine resources have attracted the attention and attention of various countries, and they have begun to vigorously develop marine resources and develop marine economy. In the development of the ocean, the detection of the underwater world can allow people to obtain richer and more accurate data and information about the ocean, and accelerate people's understanding and development of the ocean. Therefore, with the growing demand for underwater detection, underwater target recognition has become one of the hot spots in recent years. At present, underwater target recognition focuses on the research of deep-water seabed areas, while there are fewer studies on underwater target recognition in shallow coastal areas. Compared with deep-water seabed areas, sound waves have a strong scattering effect in shallow coastal areas. Sonar images have problems such as high noise, non-single imaging background, low contrast, color deviation, image distortion, etc. There are fewer target features in sonar images, blurred contours, and they are easily submerged by noise. In addition, there are many rocks on the bottom of the shallow lake, and there are many false targets in the imaging background and they are highly similar to the targets. Therefore, it is difficult to extract target features from sonar images and identify targets.

[0003] Therefore, in view of the sonar image recognition problems and difficulties of the above-mentioned shallow and near-shore targets, there is an urgent need for an online forward-looking sonar image recognition technology for shallow and near-shore tiny targets. Summary of the invention

[0004] The technical problems to be solved by the present invention are:

[0005] The existing underwater sonar detection images have problems such as large noise, non-single imaging background, low contrast, easy color deviation, image distortion, etc., as well as few detailed features of shallow near-shore targets, blurred outlines, a large number of false targets in the imaging background and high similarity with the targets.

[0006] The present invention adopts the following technical solutions to solve the above technical problems:

[0007] The present invention provides a method for online recognition of forward-looking sonar targets in an extremely shallow water environment, comprising the following steps:

[0008] S1. The forward-looking sonar is mounted on an amphibious unmanned carrier platform, sonar images in shallow water environments are collected based on the forward-looking sonar, and the images are denoised by combining median filtering and wavelet transform;

[0009] S2, using fuzzy enhancement, segmentation and morphological processing to optimize the filtered image and extract the contour of the target;

[0010] S3, according to the position of the suspected target in the sonar image of the current frame, assuming that the movement of the platform between the previous and next two frames is uniform linear motion, and combining the posture and motion information of the platform, the position of the suspected target in the next frame is calculated;

[0011] The position of the suspected target in the next frame of sonar image is associated with the inferred position data of the suspected target through the nearest neighbor data association algorithm, and the associated suspected target in the next frame is retained. During the platform movement, false targets are gradually removed and real targets are screened out.

[0012] Furthermore, the specific process in S2 includes: first, blur enhancement, image segmentation and image dilation are performed on the filtered image to enhance the contour features of the target; then, the image is subjected to secondary segmentation, contour detection and contour area screening to remove the false area and further enhance the contour features; finally, the image is subjected to secondary dilation and contour detection to extract the contour of the target.

[0013] Furthermore, the fuzzy enhancement process in S2 is as follows:

[0014] The filtered M× image R is used as a fuzzy matrix and expressed as:

[0015]

[0016] in Represents a pixel x ij The degree of having a certain characteristic, μ ij The fuzzy membership function defined for this feature is:

[0017]

[0018] in e is the exponential fuzzy factor, F d is the reciprocal fuzzy factor, x max is the maximum gray value of the pixel in image R, and the membership degree is transformed into:

[0019]

[0020] μ ij ′=T r (μ ij )=T1(T r-1 (μ ij )),r=1,2,…

[0021] x ij ′=G -1 (μ ij ′),x ij ′∈′

[0022] After r membership transformations, the enhanced image ′ is obtained by the inverse membership function to enhance the contrast between the target and the background.

[0023] Furthermore, the image segmentation processing in S2 is specifically as follows: binarization processing is performed on each row of data of the sonar image based on the 3σ criterion, that is, each row of data representing distance information in the sonar image is divided into a group of data to be detected, and the mean and standard deviation of each group of data to be detected are obtained; the pixel values ​​whose values ​​are distributed in (μ-3σ, μ+3σ) are set to 255, and the pixel values ​​whose values ​​are distributed outside (μ-3σ, μ+3σ) are set to 0.

[0024] Furthermore, the contour detection in S2 is specifically: using findContours function to perform contour detection.

[0025] Furthermore, the contour area screening in S2 is specifically as follows: suspected targets are screened by calculating the area within the contour to remove smaller and larger false targets.

[0026] Furthermore, S3 includes the following steps:

[0027] S31: Solve the direction and distance of the suspected target in the sonar coordinate system. The calculation formula is:

[0028] θ=(x osp *b r -f / 2)*r d

[0029] r=(y osp *r r )

[0030] where x osp ,y osp are the x and y coordinates of the suspected target in the sonar image coordinate system, b r 、r r are azimuth resolution and distance resolution respectively, f is the sonar horizontal opening angle, r d is a constant for converting degrees to radians, θ and r are the azimuth and distance of the suspected target in the sonar coordinate system, respectively;

[0031] The coordinates of the suspected target in the sonar coordinate system are solved by the azimuth and distance of the suspected target in the sonar coordinate system. The specific calculation formula is:

[0032] x os =r*cos(θ)

[0033] y os =r*sin(θ)

[0034] z 0s =Du

[0035] where x os ,y os 、z 0s are the x, y, and z coordinates of the suspected target in the sonar coordinate system, D u The depth of the multi-habitat platform;

[0036] S32: For the suspected target position in the sonar coordinate system in S31, solve the position of the target in the platform center of gravity coordinate system. The calculation formula is:

[0037] x ou =x os +x c

[0038] y ou =y os +y c

[0039] z ou =z 0s +z c

[0040] where x ou ,y ou 、z ou are the x, y, and z coordinates of the suspected target in the platform's center of gravity coordinate system, respectively. c ,y c 、z c They are the x, y, and z position compensation from the sonar to the center of gravity of the platform;

[0041] S33: The position of the target in the north-east coordinate system is calculated based on the position of the suspected target in the platform center of gravity coordinate system obtained in S32. The calculation formula is:

[0042] p oNE =R*p ou +p uNE

[0043] where p oNE is the north-east depth position coordinate of the target, p ou is the position coordinate of the target in the platform center of gravity coordinate system, p uNE is the northeastern deep position coordinate of the platform in the northeastern coordinate system;

[0044] R is the six-degree-of-freedom rotation matrix of the platform:

[0045]

[0046] Where φ, θ, and ψ are the roll angle, pitch angle, and heading angle of the platform, respectively;

[0047] S34: Assuming that the motion of the platform between the two frames is uniform linear motion, combining the roll angle, pitch angle, heading angle and speed information of the platform, the coordinates of the suspected target in the previous frame in the north-east coordinate system obtained in S33 are calculated through the kinematic model of the underwater vehicle, and the coordinates of the suspected target in the next frame in the north-east coordinate system are calculated;

[0048] p oNE ′=R*p v +p nNE

[0049] p v =v*t

[0050] where p oNE ′ is the north-east depth coordinate of the suspected target in the next frame, v is the x, y, and z three-dimensional velocity of the platform, and t is the interval between two frames;

[0051] S35: For the suspected target in the next frame of sonar image, perform calculations of S31-S33 to obtain the coordinates of the suspected target in the next frame in the north-east coordinate system, and use the nearest neighbor data association algorithm to associate the coordinates with the estimated suspected target coordinates in S34, retain the associated suspected target in the next frame, and gradually screen the real target during the platform movement.

[0052] Furthermore, removing false targets in S6 includes the following process:

[0053] Establish a container current_vector to store the current feature information, a container infer_vector to store the current feature position inference information, and a container target_floww_vector to store the continuous frame matching feature information; calculate the center of gravity of the current frame feature and store it in current_vector in turn, store the position inference information of the center of gravity of the current frame in infer_vector, and because there is no previous frame information to match it at the beginning, store the center of gravity of the current frame feature in target_floww_vector in turn, and assign a corresponding feature ID number to each feature; then perform calculations for the next frame;

[0054] Solve the center of gravity of the current frame and store it in the container current_vector; take out the feature information in the container current_vector in turn, and use the nearest neighbor association algorithm to associate and match it with the inferred information of the previous frame; if the corresponding association condition is met, store the feature in the position of the corresponding feature of the previous frame in the container target_floww_vector and mark it with the same ID; if the association condition is not met, store it in the container target_floww_vector as a new feature and assign a new feature ID;

[0055] According to the platform motion information, the position of the feature center of the current frame in the container current_vector is calculated in turn and stored in the container infer_vector; the features that have not been updated for three consecutive frames in the container target_floww_vector are regarded as false features and deleted from the container target_floww_vector.

[0056] A forward-looking sonar target online recognition system in an extremely shallow water environment, the system having a program module corresponding to the steps of the method described in any one of the above technical solutions, and executing the steps of the forward-looking sonar target online recognition method in an extremely shallow water environment during operation.

[0057] A computer-readable storage medium stores a computer program, wherein the computer program is configured to implement the steps of the online forward-looking sonar target identification method in an extremely shallow water environment described in any one of the above technical solutions when called by a processor.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] The present invention designs a noise filtering method combining median filtering with wavelet transform, effectively removes most of the salt and pepper noise and high-frequency noise in the sonar image, uses fuzzy enhancement to enhance the contrast between the target and the background, separates the shoal background from the suspected target through image segmentation, and targets the problem of online target recognition, and performs target recognition based on platform motion information. The present invention dynamically and online screens suspected targets and identifies real targets during the platform movement process, thereby improving the online recognition accuracy of shoal nearshore targets.

[0060] The present invention is suitable for identifying tiny targets near the shore of shallow beaches at the bottom of lakes and seabeds, and has a wide sonar detection and identification range and a high recognition rate. After being used on an amphibious unmanned carrier platform equipped with corresponding equipment, it can undertake tasks such as underwater obstacle avoidance, underwater target search and rescue, and landing and exploration of uninhabited islands. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 is a sonar image optimization processing flow chart in an embodiment of the present invention;

[0062] Figure 2 This is a main flow chart of target recognition based on platform motion information data association in an embodiment of the present invention;

[0063] Figure 3 This is a diagram showing the optimization processing of shoal nearshore sonar images and the recognition effect of small targets in an embodiment of the present invention. DETAILED DESCRIPTION

[0064] In order to enable those skilled in the art to better understand the scheme of the present invention, exemplary implementations or embodiments of the present invention will be described below in conjunction with the accompanying drawings. Obviously, the described implementations or embodiments are only implementations or embodiments of a part of the present invention, not all of them. Based on the implementations or embodiments of the present invention, all other implementations or embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present invention.

[0065] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0066] Example 1

[0067] Combination Figures 1 to 3 As shown, the present invention provides a method for online recognition of forward-looking sonar targets in an extremely shallow water environment, comprising the following steps:

[0068] S1. The forward-looking sonar is mounted on an amphibious unmanned carrier platform. The forward-looking sonar is mounted on an amphibious unmanned carrier platform. Sonar images in shallow water environments are collected based on the forward-looking sonar, and the images are denoised by combining median filtering and wavelet transform.

[0069] The median filter uses the median value of the current pixel instead of the pixel value of the retaining wall, which can filter out isolated points and salt and pepper noise and retain the edge features in the image. The wavelet transform improves the image blurring problem caused by the median filter, and can retain the image detail information and filter out high-frequency noise. As shown in Table 1, the same sonar image data set is processed by median filtering, wavelet transform filtering, median filtering combined with wavelet transform filtering, mean filtering, Gaussian filtering and bilateral filtering, and the PSNR peak signal-to-noise ratio of the filtering results is calculated.

[0070] Table 1

[0071]

[0072]

[0073] From the PSNR results of the above filtering algorithms, it can be seen that the filtering algorithm combining median filtering and wavelet transform of the present invention has the best noise filtering effect.

[0074] S2, such as Figure 1 and Figure 3 As shown, fuzzy enhancement, segmentation and morphological processing are used to optimize the filtered image and extract the contour of the target.

[0075] First, for the filtered sonar image, the edge contour features, detail resolution, etc. are fuzzy enhanced, namely:

[0076] The filtered M× image R is used as a fuzzy matrix and expressed as:

[0077]

[0078] in Represents a pixel x ij The degree of having a certain characteristic, μ ij The fuzzy membership function defined for this feature is:

[0079]

[0080] where F e is the exponential fuzzy factor, F d is the reciprocal fuzzy factor, x max is the maximum gray value of the pixel in image R, and the membership degree is transformed into:

[0081]

[0082] μ ij ′=T r (μ ij )=T1(T r-1 (μ ij )),r=1,2,…

[0083] x ij ′=G -1 (μ ij ′),x ij ′∈R′

[0084] After r membership transformations, the enhanced image R′ is obtained by the inverse membership function to enhance the contrast between the target and the background.

[0085] The image is segmented, specifically: threshold segmentation of sonar images is performed based on the 3σ criterion of grayscale. The 3σ criterion is based on the equal precision repeated measurement of normal distribution, and the interference or noise caused by singular data is difficult to satisfy the normal distribution. 2In the normal distribution, the probability that the value is distributed in (μ-σ,μ+σ) is 0.6826, the probability that the value is distributed in (μ-2σ,μ+2σ) is 0.9544, and the probability that the value is distributed in (μ-3σ,μ+3σ) is 0.9974. For the random error of the normal distribution, the probability of falling outside ±3σ is only 0.27%, so if the absolute value νi of the residual error of a certain measurement value in a set of measurement data is greater than 3σ, then the measurement value is a bad value and should be eliminated. Usually, the error equal to ±3σ is taken as the limit error. Each line of sonar image data is screened by the 3σ criterion, leaving the highlight part returned by hard materials such as the target and eliminating most of the noise and interference.

[0086] Based on the 3σ criterion, each row of sonar image data is binarized, that is, each row of data representing distance information in the sonar image is divided into a group of data to be detected, and the mean and standard deviation of each group of data to be detected are calculated:

[0087]

[0088] where μ i is the mean of the data to be tested in the i-th row of the sonar image, σ i is the standard deviation of the data to be tested in the i-th row of the sonar image. ij is the grayscale value of the pixel in the i-th row and j-th column in the sonar image.

[0089] The pixel values ​​whose values ​​are distributed in (μ-3σ, μ+3σ) are set to 255, and the pixel values ​​whose values ​​are distributed outside (μ-3σ, μ+3σ) are set to 0. The highlights returned by hard objects such as the target are retained, and most of the background noise and interference are eliminated.

[0090] The image is expanded by using a structural element and calculating the maximum value of the overlapping area of ​​the structural element and the image. This expands the boundary of the object and increases the size of the object. The specific calculation formula is:

[0091] (A⊕B)(x,y)=max (i,j)∈B {A(x+i,y+j)}

[0092] Where A is the original image and B is the structural element.

[0093] A cross-shaped structure element is used in the processing, and the value in the structure element is set to 1. The dilation operation is to slide the cross-shaped structure element on the image, and set the grayscale value of the image pixel at the center of the structure element to the maximum value of the pixel in the image area corresponding to the area where the structure element value is 1. Based on the above processing results, the contour features of the real target in the sonar image are further enhanced, and the tiny noise points and false targets in the sonar image are removed.

[0094] Then, the image is segmented again based on the 3σ criterion of grayscale.

[0095] Use the findContours function in opencv to perform contour detection on the image, extract the target contour, and screen suspected targets by calculating the area within the contour to remove small and large false targets.

[0096] Finally, the image is also dilated twice.

[0097] Also use the findContours function in opencv to perform contour detection on the image and finally extract the contour of the target.

[0098] S3. Identify the target based on the platform’s position and motion information.

[0099] After the sonar image is optimized, there will still be suspected targets of varying sizes left. In order to identify the real targets from these suspected targets, these suspected targets in the sonar image need to be screened.

[0100] The false targets in the suspected targets in the sonar image circled by contour detection change randomly in each frame of the sonar image, while the movement of the real targets in each frame of the sonar image is related to the movement of the platform, so the false targets in the sonar image can be effectively removed based on the platform's movement information. Figure 2 As shown, the specific steps include:

[0101] S31: Solve the direction and distance of the suspected target in the sonar coordinate system. The calculation formula is:

[0102] θ=(x osp *b r -f / 2)*r d

[0103] r=(y osp *r r )

[0104] where x osp ,y osp are the x and y coordinates of the suspected target in the sonar image coordinate system, b r 、r r are azimuth resolution and distance resolution respectively, f is the sonar horizontal opening angle, r d is a constant for converting degrees to radians, θ and r are the azimuth and distance of the suspected target in the sonar coordinate system, respectively;

[0105] The coordinates of the suspected target in the sonar coordinate system are solved by the azimuth and distance of the suspected target in the sonar coordinate system. The specific calculation formula is:

[0106] xos =r*cos(θ)

[0107] y os =r*sin(θ)

[0108] z 0s =D u

[0109] where x os ,y os 、z 0s are the x, y, and z coordinates of the suspected target in the sonar coordinate system, D u The depth of the multi-habitat platform;

[0110] S32: For the suspected target position in the sonar coordinate system in S31, solve the position of the target in the platform center of gravity coordinate system. The calculation formula is:

[0111] x ou =x os +x c

[0112] y ou =y os +y c

[0113] z ou =z 0s +z c

[0114] where x ou ,y ou 、z ou are the x, y, and z coordinates of the suspected target in the platform's center of gravity coordinate system, respectively. c ,y c 、z c They are the x, y, and z position compensation from the sonar to the center of gravity of the platform;

[0115] S33: The position of the target in the north-east coordinate system is calculated based on the position of the suspected target in the platform center of gravity coordinate system obtained in S32. The calculation formula is:

[0116] p oNE =R*p ou +p uNE

[0117] where p oNE is the north-east depth position coordinate of the target, p ou is the position coordinate of the target in the platform center of gravity coordinate system, p uBE is the northeastern deep position coordinate of the platform in the northeastern coordinate system;

[0118] R is the six-degree-of-freedom rotation matrix of the platform:

[0119]

[0120] Where φ, θ, and ψ are the roll angle, pitch angle, and heading angle of the platform, respectively;

[0121] S34: Assuming that the motion of the platform between the two frames is uniform linear motion, combining the roll angle, pitch angle, heading angle and speed information of the platform, the coordinates of the suspected target in the previous frame in the north-east coordinate system obtained in S33 are calculated through the kinematic model of the underwater vehicle, and the coordinates of the suspected target in the next frame in the north-east coordinate system are calculated;

[0122] p oNE ′=R*p v +p oNE

[0123] p v =v*t

[0124] where p oNE ′ is the north-east depth coordinate of the suspected target in the next frame, v is the x, y, and z three-dimensional velocity of the platform, and t is the interval between two frames;

[0125] S35: For the suspected target in the next frame of sonar image, perform calculations of S31-S33 to obtain the coordinates of the suspected target in the next frame in the north-east coordinate system, and use the nearest neighbor data association algorithm to associate the coordinates with the estimated suspected target coordinates in S34, retain the associated suspected target in the next frame, and gradually screen the real target during the platform movement.

[0126] Create a container current_vector to store the current feature information, a container infer_vector to store the current feature position inference information, and a container target_floww_vector to store the continuous frame matching feature information. Initially, calculate the center of gravity of the current frame feature and store it in current_vector in sequence, and store the position inference information of the center of gravity of the current frame in infer_vector. Since there is no previous frame information to match it at the beginning, the center of gravity of the current frame feature is stored in target_floww_vector in sequence, and each feature is assigned a corresponding feature ID number. Then calculate the next frame.

[0127] Solve the center of gravity of the current frame and store it in the container current_vector. Take out the feature information in the container current_vector in turn, and use the nearest neighbor association algorithm to associate and match it with the inferred information of the previous frame. If the corresponding association conditions are met, the feature is stored in the position of the corresponding feature of the previous frame in the container target_floww_vector and marked with the same ID. If the association conditions are not met, it is stored in the container target_floww_vector as a new feature and a new feature ID is assigned.

[0128] According to the platform motion information, the center of gravity of the current frame feature in the container current_vector is calculated in turn and stored in the container infer_vector. The features in the container target_floww_vector that have not been updated for three consecutive frames (that is, the matching features are not found in the next frame for three consecutive frames) are considered as false features and deleted from the container target_floww_vector.

[0129] After the fifth frame, the feature with the longest matching length in the container target_floww_vector is regarded as the true target each time.

[0130] As shown in Table 2, 7 groups of sonar images and data obtained in the experiment were selected, and the above-mentioned forward-looking sonar target online recognition method in extremely shallow water environment was used to perform sonar image optimization processing and target recognition on these 7 groups of data, and the feature extraction rate and target recognition rate of the method were statistically analyzed. The results are shown in Table 2.

[0131] Table 2

[0132]

[0133]

[0134] From the high feature extraction rate in Table 2, it can be seen that the above sonar image processing method can perform good noise processing on the sonar image of the shoal and nearshore. From the recognition rate, it can be seen that the above target recognition technology based on platform motion information data association can better identify small targets on the shoal and nearshore.

[0135] The method (algorithm) for online recognition of forward-looking sonar targets in an extremely shallow water environment proposed in the present invention is the underlying technical core of the present invention, and various products can be derived based on the algorithm.

[0136] Based on the method proposed in the present invention, a forward-looking sonar target online recognition system in an extremely shallow water environment is developed using a programming language. The system has a program module corresponding to the steps of the above-mentioned technical solution, and executes the steps of the above-mentioned forward-looking sonar target online recognition method in an extremely shallow water environment during operation.

[0137] The computer program of the developed system (software) is stored on a computer-readable storage medium, and the computer program is configured to implement the steps of the above-mentioned method for online identification of forward-looking sonar targets in an extremely shallow water environment when called by a processor, that is, the present invention is materialized on a carrier to become a computer program product.

[0138] Various implementations of the systems and techniques described herein can be realized in digital electronic circuit systems, integrated circuit systems, dedicated ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0139] The computer programs (also referred to as programs, software, software applications, or codes) of the present invention include machine instructions for programmable processors, and these computer programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or device (e.g., disk, optical disk, memory, programmable logic device PLD) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0140] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. An online recognition method for forward-looking sonar targets in extremely shallow water environments. It is characterized in that The steps include: S1. The forward-looking sonar is mounted on an amphibious unmanned carrier platform, sonar images in shallow water environments are collected based on the forward-looking sonar, and the images are denoised by combining median filtering and wavelet transform; S2, using fuzzy enhancement, segmentation and morphological processing to optimize the filtered image and extract the contour of the target; S3, according to the position of the suspected target in the sonar image of the current frame, assuming that the movement of the platform between the previous and next two frames is uniform linear motion, and combining the posture and motion information of the platform, the position of the suspected target in the next frame is calculated; The position of the suspected target in the next frame of sonar image is associated with the inferred position data of the suspected target through the nearest neighbor data association algorithm, and the associated suspected target in the next frame is retained. During the platform movement, false targets are gradually removed and real targets are screened out.

2. The online recognition method for forward-looking sonar targets in extremely shallow water environment according to claim 1, It is characterized in that The specific process in S2 includes: first, blur enhancement, image segmentation and image dilation are performed on the filtered image to enhance the contour features of the target; then, the image is subjected to secondary segmentation, contour detection and contour area screening to remove the false area and further enhance the contour features; finally, the image is subjected to secondary dilation and contour detection to extract the contour of the target.

3. The online recognition method for forward-looking sonar targets in extremely shallow water environment according to claim 1, It is characterized in that The fuzzy enhancement processing in S2 is as follows: The filtered M×N image R is used as a fuzzy matrix and expressed as: in Represents a pixel x ij The degree of having a certain characteristic, μ ij The fuzzy membership function defined for this feature is: where F e is the exponential fuzzy factor, F d is the reciprocal fuzzy factor, x max is the maximum gray value of the pixel in image R, and the membership degree is transformed into: m ij ′ =T r (m ij )=T 1 (T r-1 (m ij )),r=1,2,… x ij ′ =G -1 (m ij ′ ),x ij ′ ∈R ′ After r membership transformations, the enhanced image R′ is obtained by the inverse membership function to enhance the contrast between the target and the background.

4. The online recognition method for forward-looking sonar targets in extremely shallow water environment according to claim 1, It is characterized in that The image segmentation process in S2 is specifically as follows: binarization is performed on each row of the sonar image based on the 3σ criterion, that is, each row of data representing distance information in the sonar image is divided into a group of data to be detected, and the mean and standard deviation of each group of data to be detected are obtained; The pixel values ​​whose values ​​are distributed in (μ-3σ, μ+3σ) are set to 255, and the pixel values ​​whose values ​​are distributed outside (μ-3σ, μ+3σ) are set to 0.

5. The online recognition method for forward-looking sonar targets in an extremely shallow water environment according to claim 1, It is characterized in that The contour detection in S2 is specifically as follows: contour detection is performed using the findContours function.

6. The online recognition method for forward-looking sonar targets in extremely shallow water environment according to claim 1, It is characterized in that The contour area screening in S2 is specifically as follows: suspected targets are screened by calculating the area within the contour to remove smaller and larger false targets.

7. The online recognition method for forward-looking sonar targets in extremely shallow water environment according to claim 1, It is characterized in that S3 includes the following steps: S31: Solve the direction and distance of the suspected target in the sonar coordinate system. The calculation formula is: θ=(x osp *b r -f / 2)*r d r＝(y osp *r r ) where x osp ,y osp are the x and y coordinates of the suspected target in the sonar image coordinate system, b r 、r r are azimuth resolution and distance resolution respectively, f is the sonar horizontal opening angle, r d is a constant for converting degrees to radians, θ and r are the azimuth and distance of the suspected target in the sonar coordinate system, respectively; The coordinates of the suspected target in the sonar coordinate system are solved by the azimuth and distance of the suspected target in the sonar coordinate system. The specific calculation formula is: x os =r*cos(θ) y os =r*sin(θ) z 0s =D u where x os ,y os 、z 0s are the x, y, and z coordinates of the suspected target in the sonar coordinate system, D u For the depth of the multi-habitat platform; S32: For the suspected target position in the sonar coordinate system in S31, solve the position of the target in the platform center of gravity coordinate system. The calculation formula is: x ou =x os +x c and ou =and os +y c With ou =from 0s +with c where x ou ,y ou 、z ou are the x, y, and z coordinates of the suspected target in the platform's center of gravity coordinate system, respectively. c ,y c 、z c They are the x, y, and z position compensation from the sonar to the center of gravity of the platform; S33: The position of the target in the north-east coordinate system is calculated based on the position of the suspected target in the platform center of gravity coordinate system obtained in S32. The calculation formula is: p oNE =R*p ou +p uNE where p oNE is the north-east depth position coordinate of the target, p ou is the position coordinate of the target in the platform center of gravity coordinate system, p uNE is the northeastern deep position coordinate of the platform in the northeastern coordinate system; R is the six-degree-of-freedom rotation matrix of the platform: Where φ, θ, and ψ are the roll angle, pitch angle, and heading angle of the platform, respectively; S34: Assuming that the motion of the platform between the two frames is uniform linear motion, combining the roll angle, pitch angle, heading angle and speed information of the platform, the coordinates of the suspected target in the previous frame in the north-east coordinate system obtained in S33 are calculated through the kinematic model of the underwater vehicle, and the coordinates of the suspected target in the next frame in the north-east coordinate system are calculated; p oNE ′=R*p v +p oNE p v =v*t where p oNE ′ is the north-east depth coordinate of the suspected target in the next frame, v is the x, y, and z three-dimensional velocity of the platform, and t is the interval between two frames; S35: For the suspected target in the next frame of sonar image, perform calculations of S31-S33 to obtain the coordinates of the suspected target in the next frame in the north-east coordinate system, and use the nearest neighbor data association algorithm to associate the coordinates with the estimated suspected target coordinates in S34, retain the associated suspected target in the next frame, and gradually screen the real target during the platform movement.

8. The online recognition method for forward-looking sonar targets in extremely shallow water environment according to claim 7, It is characterized in that Removing false targets in S6 includes the following process: Establish a container current_vector to store the current feature information, a container infer_vector to store the current feature position inference information, and a container target_floww_vector to store the continuous frame matching feature information; calculate the center of gravity of the current frame feature and store it in current_vector in sequence, store the position inference information of the center of gravity of the current frame in infer_vector, and because there is no previous frame information to match it at the beginning, store the center of gravity of the current frame feature in target_floww_vector in sequence, and assign a corresponding feature ID number to each feature; Then the calculation of the next frame is performed; Solve the center of gravity of the current frame and store it in the container current_vector; take out the feature information in the container current_vector in turn, and use the nearest neighbor association algorithm to associate and match it with the inferred information of the previous frame; if the corresponding association condition is met, store the feature in the position of the corresponding feature of the previous frame in the container target_floww_vector and mark it with the same ID; if the association condition is not met, store it in the container target_floww_vector as a new feature and assign a new feature ID; According to the platform motion information, the position of the feature center of the current frame in the container current_vector is calculated in turn and stored in the container infer_vector; the features that have not been updated for three consecutive frames in the container target_floww_vector are regarded as false features and deleted from the container target_floww_vector.

9. An online recognition system for forward-looking sonar targets in extremely shallow water environments. It is characterized in that The system has a program module corresponding to the steps of the method described in any one of claims 1 to 8, and executes the steps of the online forward-looking sonar target recognition method in an extremely shallow water environment when running.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of the online forward-looking sonar target identification method in an extremely shallow water environment according to any one of claims 1 to 8 when called by a processor.

Images