Underwater small target identification method using acousto-optic fusion

Abstract

The invention relates to the technical field of underwater small target recognition, and discloses an underwater small target recognition method using acousto-optic fusion, which comprises the following steps: acquiring an acoustic image of a target area through a side-scan sonar, and processing the acoustic image by using a background segmentation technology to obtain an acoustic recognition result; obtaining an expert rule, and performing three-section fuzzy, weight calculation, normalization and probability classification on the acoustic recognition result by using the expert rule to obtain a successful recognition probability of each underwater small target; for the underwater small target whose successful identification probability is greater than or equal to a preset probability threshold value, directly taking the acoustic identification result as an identification result of the underwater small target and outputting the identification result; and for the underwater small target of which the successful identification probability is less than a preset probability threshold, performing secondary identification by using the optical camera, and outputting an optical identification result as an identification result of the underwater small target. According to the method, the purpose of improving the accuracy of underwater small target recognition on the basis of not sacrificing time complexity is achieved.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of underwater small target recognition, and particularly to an underwater small target recognition method using acoustic-optical fusion. Background Art

[0002] With the continuous development of marine technology, the detection requirements for underwater small targets in the shallow sea environment by humans have gradually increased. Acoustic sensors (side-scan sonar) are frequently used as important devices for underwater target recognition. Using side-scan sonar can achieve underwater detection over a relatively wide range and has a very high recognition accuracy for larger targets (such as sunken ships, etc.). However, its prominent drawback still exists, that is, the recognition accuracy for small targets (such as frogmen, mines, unmanned underwater vehicles, etc.) is relatively low.

[0003] In order to improve the recognition accuracy of underwater small targets, the currently commonly used solution is to increase the resolution of the sonar. However, high-precision side-scan sonar itself has a large volume and requires higher power. The key problem in this process is how to obtain the position and size information of small targets. An optical camera can itself observe small targets by changing the focal length, but the imaging range of the optical camera is extremely limited. Directly using optical recognition not only increases the recognition time but also has the problem of being difficult to detect over a relatively wide range.

[0004] Scientifically fusing acoustic recognition and optical recognition for underwater small target recognition is a feasible idea, that is, using optical recognition to improve the accuracy of acoustic recognition. The key steps here are how to process the results of acoustic recognition and under what conditions to use optical recognition and how to scientifically and effectively process the results of optical recognition. So far, no research team has proposed an effective solution. Summary of the Invention

[0005] In order to solve the above technical problems, the embodiments of the present application propose an underwater small target recognition method using acoustic-optical fusion, which makes full use of the breadth of acoustic recognition and the accuracy of optical recognition, and combines expert rules to achieve the purpose of improving the recognition accuracy of underwater small targets without sacrificing time complexity.

[0006] To achieve the above object, an embodiment of the present application proposes an underwater small target recognition method using acoustic-optical fusion. The method includes: obtaining an acoustic image of a target area through a side-scan sonar, and processing the acoustic image using background segmentation technology to obtain an acoustic recognition result; wherein, the acoustic recognition result includes the center point coordinates, pixel area, and approximate shape of the underwater small target; obtaining expert rules, and using the expert rules to perform three-stage fuzzy, weight calculation, normalization, and probability classification on the acoustic recognition result to obtain the successful recognition probability of each underwater small target; sequentially determining whether the successful recognition probability of each underwater small target is less than a preset probability threshold; for underwater small targets with a successful recognition probability greater than or equal to the preset probability threshold, directly outputting the acoustic recognition result as the recognition result of the underwater small target; for underwater small targets with a successful recognition probability less than the preset probability threshold, performing secondary recognition using an optical camera and outputting the optical recognition result as the recognition result of the underwater small target.

[0007] To achieve the above object, an embodiment of the present application also proposes an underwater small target recognition system using acoustic-optical fusion. The system includes: an acoustic recognition module, an expert processing module, a judgment module, an optical recognition module, and an output module; the acoustic recognition module is used to obtain an acoustic image of a target area through a side-scan sonar, and process the acoustic image using background segmentation technology to obtain an acoustic recognition result, wherein the acoustic recognition result includes the center point coordinates, pixel area, and approximate shape of the underwater small target; the expert processing module is used to obtain expert rules, and use the expert rules to perform three-stage fuzzy, weight calculation, normalization, and probability classification on the acoustic recognition result to obtain the successful recognition probability of each underwater small target; the judgment module is used to sequentially determine whether the successful recognition probability of each underwater small target is less than a preset probability threshold, and for underwater small targets with a successful recognition probability greater than or equal to the preset probability threshold, instructing the output module to directly output the acoustic recognition result as the recognition result of the underwater small target, and for underwater small targets with a successful recognition probability less than the preset probability threshold, calling the optical recognition module for processing; the optical recognition module is used to perform secondary recognition on underwater small targets with a successful recognition probability less than the preset probability threshold using an optical camera to obtain an optical recognition result, and instructing the output module to output the optical recognition result as the recognition result of the underwater small target.

[0008] To achieve the above object, an embodiment of the present application also proposes an electronic device. The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute an underwater small target recognition method using acoustic-optical fusion as described above.

[0009] To achieve the above object, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which when executed by a processor, can implement a method for underwater small target recognition using acoustic-optic fusion as described above.

[0010] A method for underwater small target recognition using acoustic-optic fusion proposed by an embodiment of the present application first performs acoustic recognition on a target area through a side-scan sonar, combines expert rules, and based on the center point coordinates, pixel area, and approximate shape of the underwater small target in the acoustic recognition result, determines whether to perform optical recognition. For underwater small targets with a relatively high successful recognition probability in the acoustic recognition result, the acoustic recognition result is directly output. For underwater small targets with a relatively low successful recognition probability in the acoustic recognition result, an optical camera is called for optical recognition, and the optical recognition result is output as the recognition result of the underwater small target. This design effectively combines the breadth of acoustic recognition and the accuracy of optical recognition, trusts the acoustic recognition result with a relatively high successful recognition probability, and performs secondary recognition on the acoustic recognition result with a relatively low successful recognition probability through an optical camera, thus well achieving the purpose of improving the accuracy of underwater small target recognition without sacrificing time complexity. In addition, the introduction of expert rules improves the scientificity and standardization of acoustic-optic fusion, and the design of three-stage fuzzy and probability classification well evaluates the successful recognition probability of underwater small targets, which effectively improves the universality and generalization ability of underwater small target recognition.

[0011] In some optional embodiments, background segmentation technology is used to process the acoustic image to obtain the acoustic recognition result, including: removing the background from the acoustic image to obtain the center point coordinates of the underwater small target; extracting the pixel area occupied by the underwater small target, where the pixel area of the underwater small target is the total number of pixel points occupied by the underwater small target; approximating the shape of the underwater small target to obtain the approximate shape of the underwater small target, and the approximate shape of the underwater small target is represented by the number of sides. If the approximate shape of the underwater small target is circular, the number of sides is recorded as 0. The acoustic recognition result includes a lot of content. The present application selects the center point coordinates, pixel area, and approximate shape to participate in subsequent expert rule processing, which can effectively improve the speed of successful recognition probability evaluation. In addition, the number of sides can simply and intuitively characterize the approximate shape of the underwater small target.

[0012] In some optional embodiments, the expert rule is used to connect acoustic recognition and optical recognition. The design of the expert rule includes two parts: input fuzzification and probability classification. The expert rule uses a three-stage fuzzy method to achieve input fuzzification and a weighted sum and normalization method to achieve probability classification. The basis for probability classification comes from the offline recognition results of existing acoustic images.

[0013] In some alternative embodiments, a three-stage fuzzy method is adopted to achieve input fuzzification, including: using the acoustic recognition result as the input of the three-stage fuzzy of the expert rule; calculating the relative distance between the underwater small target and the side-scan sonar based on the central point coordinates of the underwater small target, performing three-stage fuzzy on the relative distance, dividing it into three parts: long distance, medium distance, and short distance according to the size of the relative distance, and assigning values of 1, 2, and 3 in sequence; calculating the area ratio of the pixel area of the underwater small target to the total area of the acoustic image, performing three-stage fuzzy on the area ratio, dividing it into three parts: small area, medium area, and large area according to the size of the area ratio, and assigning values of 1, 2, and 3 in sequence; performing three-stage fuzzy on the approximate shape of the underwater small target, dividing it into three parts: multi-sided, few-sided, and circular according to the number of sides of the approximate shape, and assigning values of 1, 2, and 3 in sequence. The accuracy of acoustic recognition is correlated with the relative distance of the underwater small target from the side-scan sonar, the area ratio of the underwater small target imaging, and the approximate shape of the underwater small target. Acoustic recognition has a high recognition accuracy for underwater small targets with a short relative distance, a large area ratio, and a regular shape, and a relatively low recognition accuracy for underwater small targets with a long relative distance, a small area ratio, and an irregular shape. Therefore, through different assignments, it can be well quantified into the three-stage fuzzy results of relative distance, pixel area, and approximate shape.

[0014] In some alternative embodiments, a weighted sum and normalization method is adopted to achieve probability classification, including: respectively assigning weighted weights to the three-stage fuzzy results of relative distance, pixel area, and approximate shape; according to the assigned weighted weights, performing a weighted sum on the three-stage fuzzy results of relative distance, pixel area, and approximate shape, and normalizing the weighted sum result to between 0 and 1 to finally obtain the successful recognition probability of each underwater small target.

[0015] In some alternative embodiments, the preset probability threshold is 0.85, and the weighted weights respectively assigned to the three-stage fuzzy results of relative distance, pixel area, and approximate shape are 0.4, 0.4, and 0.2 in sequence.

[0016] In some alternative embodiments, for underwater small targets with a successful recognition probability less than a preset probability threshold, secondary recognition is performed using an optical camera, and the optical recognition result is output as the recognition result of the underwater small target, including: obtaining the position information of the underwater small target with a successful recognition probability less than the preset probability threshold, where the position information of the underwater small target is derived from the acoustic recognition result; calculating the relative distance between the optical camera and the underwater small target, and calculating the zoom ratio of the optical camera according to the relative distance to increase the pixel area, so as to obtain an optical image; performing shape approximation on the underwater small target in the obtained optical image to obtain the approximate shape of the underwater small target, and the approximate shape of the underwater small target is represented by the number of sides. If the approximate shape of the underwater small target is circular, the number of sides is recorded as 0; outputting the position information of the underwater small target, as well as the pixel area and approximate shape of the underwater small target obtained by optical recognition, as the recognition result of the underwater small target, and restoring the focal length of the optical camera to the default focal length. The purpose of optical recognition is to improve the recognition success rate. Therefore, it is necessary to calculate the zoom ratio of the optical camera according to the relative distance to increase the pixel area and ensure the accuracy of the optical recognition result. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the related art, the following will briefly introduce the drawings required for use in the description of the embodiments of the present application or the related art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 is a flowchart of a method for recognizing underwater small targets using acoustic-optical fusion provided in an embodiment of the present application;

[0019] Figure 2 is a modular schematic diagram of a method for recognizing underwater small targets using acoustic-optical fusion provided in an embodiment of the present application;

[0020] Figure 3 is a schematic diagram of the principle of an expert rule provided in an embodiment of the present application;

[0021] Figure 4 is a schematic diagram of the structure of a system for recognizing underwater small targets using acoustic-optical fusion provided in another embodiment of the present application;

[0022] Figure 5 is a schematic diagram of the structure of an electronic device provided in another embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will elaborate on each embodiment of this application in conjunction with the accompanying drawings. However, those of ordinary skill in the art can understand that in each embodiment of this application, many technical details are presented to help readers better understand this application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in this application can still be implemented. The division of the following embodiments is for convenience of description and should not impose any limitation on the specific implementation of this application. Under the premise of no contradiction, each embodiment can be combined and cross-referenced with each other.

[0024] An embodiment of this application proposes an underwater small target recognition method using acoustic-optical fusion, which is applied to an electronic device. Here, the electronic device can be a terminal or a server. In this embodiment and the following embodiments, the electronic device is exemplified by a server. The following will specifically describe the implementation details of an underwater small target recognition method using acoustic-optical fusion proposed in this embodiment. The following content is only implementation details provided for convenience of understanding and is not necessary for implementing this solution.

[0025] The specific process of an underwater small target recognition method using acoustic-optical fusion proposed in this embodiment can be as Figure 1 shown, and its modular schematic diagram is as Figure 2 shown. The method specifically includes:

[0026] Step 101: Obtain an acoustic image of the target area through a sidescan sonar, and process the acoustic image using background segmentation technology to obtain an acoustic recognition result. Here, the acoustic recognition result includes the center point coordinates, pixel area, and approximate shape of the underwater small target.

[0027] In specific implementation, the server first needs to obtain an acoustic image of the target area through a sidescan sonar, and then process the acoustic image using image processing technologies including background segmentation technology to obtain an acoustic recognition result of the acoustic image. The acoustic recognition result includes, but is not limited to, the center point coordinates, pixel area, and approximate shape of the underwater small target, etc.

[0028] In one example, the server uses background segmentation technology to remove the background from the acoustic image and obtain the central point coordinates of the underwater small target. Subsequently, based on the central point coordinates of the underwater small target, the pixel area occupied by the underwater small target is extracted. The pixel area of the underwater small target is the total number of pixel points occupied by the underwater small target. Finally, the shape of the underwater small target is approximated to obtain the approximate shape of the underwater small target. The approximate shape of the underwater small target is represented by the number of sides. For example, if the shape of the underwater small target is approximated to a triangle, the number of sides is recorded as 3. If the shape of the underwater small target is approximated to a rectangle, the number of sides is recorded as 4. It should be particularly noted that if the approximate shape of the underwater small target is a circle, the number of sides is recorded as 0.

[0029] It can be understood that the acoustic recognition results include a lot of content. In this embodiment, the central point coordinates, pixel area, and approximate shape are selected to participate in the subsequent expert rule processing, which can effectively improve the speed of the successful recognition probability evaluation. In addition, the number of sides can simply and intuitively represent the approximate shape of the underwater small target.

[0030] Step 102, obtain the expert rules, and use the expert rules to perform three-stage fuzzy, weight calculation, normalization, and probability classification on the acoustic recognition results to obtain the successful recognition probabilities of each underwater small target.

[0031] In a specific implementation, after the server completes the acoustic recognition, it needs to obtain the expert rules, and use the obtained expert rules to perform three-stage fuzzy, weight calculation, normalization, and probability classification on the acoustic recognition results to obtain the successful recognition probabilities of each underwater small target. The higher the successful recognition probability, the more credible the acoustic recognition result. On the contrary, it indicates that the credibility of the acoustic recognition result is not high.

[0032] It should be noted that the essential role of the expert rules is to connect acoustic recognition and optical recognition. The design of the expert rules includes two parts: input fuzzification and probability classification. The expert rules use the three-stage fuzzy method to achieve input fuzzification, and use the weighted and normalization method to achieve probability classification. The basis of probability classification comes from the offline recognition results of existing acoustic images.

[0033] In one example, the principle of the expert rules can be as Figure 3 shown.

[0034] In one example, after the server completes the acoustic recognition, the acoustic recognition results are used as the input for the three-stage fuzzy of the expert rules.

[0035] First is the three - stage fuzzification of the relative distance. The server calculates the relative distance between the underwater small target and the sidescan sonar based on the central point coordinates of the underwater small target, and then performs three - stage fuzzification on the relative distance. It is divided into three parts: long distance, medium distance, and short distance according to the size of the relative distance, and is assigned values of 1, 2, and 3 in sequence. Since the closer the distance, the more reliable the acoustic recognition result, the three - stage fuzzification result of the short distance is assigned the value of 3. It is worth mentioning that the three - stage fuzzification standard of the relative distance can be selected and set by those skilled in the art according to actual needs. For example, a relative distance less than 1 meter is fuzzified as short distance, greater than 1 meter and less than 2 meters is fuzzified as medium distance, and greater than 2 meters is fuzzified as long distance.

[0036] Next is the three - stage fuzzification of the pixel area (area ratio). The server calculates the area ratio of the pixel area of the underwater small target to the total area of the acoustic image, and performs three - stage fuzzification on the area ratio. It is divided into three parts: small area, medium area, and large area according to the size of the area ratio, and is assigned values of 1, 2, and 3 in sequence. Since the larger the area ratio, the more reliable the acoustic recognition result, the three - stage fuzzification result of the large area is assigned the value of 3. It is worth mentioning that the three - stage fuzzification standard of the pixel area can be selected and set by those skilled in the art according to actual needs. For example, an area ratio less than 0.1 is fuzzified as a small area, greater than 0.1 and less than 0.5 is fuzzified as a medium area, and greater than 0.5 is fuzzified as a large area.

[0037] Finally is the three - stage fuzzification of the approximate shape (number of sides). The server performs three - stage fuzzification on the approximate shape of the underwater small target. It is divided into three parts: multi - sided, few - sided, and circular according to the number of sides, and is assigned values of 1, 2, and 3 in sequence. Since the more regular the shape, the more reliable the acoustic recognition result, the three - stage fuzzification result of the circular shape is assigned the value of 3. It is worth mentioning that the three - stage fuzzification standard of the approximate shape can be selected and set by those skilled in the art according to actual needs. For example, a number of sides equal to 0 is fuzzified as a circular shape, greater than 2 and less than 5 is fuzzified as few - sided, and greater than 5 (including 5) is fuzzified as multi - sided.

[0038] The accuracy of acoustic recognition is correlated with the relative distance between the underwater small target and the sidescan sonar, the area ratio of the underwater small target imaging, and the approximate shape of the underwater small target. Acoustic recognition has a high recognition accuracy for underwater small targets with a short relative distance, a large area ratio, and a regular shape, but has a relatively low recognition accuracy for underwater small targets with a long relative distance, a small area ratio, and an irregular shape. Therefore, through different assignments, it can be well quantified into the three - stage fuzzification results of relative distance, pixel area, and approximate shape.

[0039] In one example, after the server completes the input fuzzification using expert rules, it will assign weighted weights to the three-stage fuzzy results of relative distance, pixel area, and approximate shape respectively. Subsequently, according to the assigned weighted weights, it will perform weighted summation on the three-stage fuzzy results of relative distance, pixel area, and approximate shape, and normalize the weighted summation result to between 0 and 1, finally obtaining the successful recognition probability of each underwater small target.

[0040] In one example, the weighted weights assigned by the server to the three-stage fuzzy results of relative distance, pixel area, and approximate shape are 0.4, 0.4, and 0.2 in sequence. Suppose the assignments of the three-stage fuzzy results of relative distance, pixel area, and approximate shape of an underwater small target are 3, 1, and 3 respectively. Then the weighted summation result of this underwater small target is 3×0.4 + 1×0.4 + 3×0.2 = 2.2. Correspondingly, the normalized value of the weighted summation result of this underwater small target is 2.2 / 3 = 0.73.

[0041] Step 103: Judging in sequence whether the successful recognition probability of each underwater small target is less than a preset probability threshold.

[0042] In specific implementation, after the server obtains the successful recognition probabilities of each underwater small target, it needs to judge in sequence whether the successful recognition probability of each underwater small target is less than a preset probability threshold. For an underwater small target whose successful recognition probability is greater than or equal to the preset probability threshold, its acoustic recognition result is credible. For an underwater small target whose successful recognition probability is less than the preset probability threshold, its acoustic recognition result is not credible.

[0043] In one example, the preset probability threshold is 0.85.

[0044] Step 104: For an underwater small target whose successful recognition probability is greater than or equal to the preset probability threshold, directly output the acoustic recognition result as the recognition result of the underwater small target.

[0045] In specific implementation, for an underwater small target whose successful recognition probability is greater than or equal to the preset probability threshold, its acoustic recognition result is credible, and the server can directly output the acoustic recognition result as the recognition result of the underwater small target.

[0046] Step 105: For an underwater small target whose successful recognition probability is less than the preset probability threshold, perform secondary recognition using an optical camera and output the optical recognition result as the recognition result of the underwater small target.

[0047] In specific implementation, for an underwater small target whose successful recognition probability is less than the preset probability threshold, its acoustic recognition result is not credible. At this time, the server needs to perform secondary recognition using an optical camera and output the optical recognition result as the recognition result of the underwater small target.

[0048] In one example, the server obtains the position information of underwater small targets with a successful recognition probability less than a preset probability threshold. The position information of the underwater small targets is derived from the acoustic recognition result. Subsequently, the relative distance between the optical camera and the underwater small target is calculated, and the zoom ratio of the optical camera is calculated according to the relative distance to increase the pixel area, so as to obtain an optical image. Next, the shape of the underwater small target is approximated in the obtained optical image to obtain the approximate shape of the underwater small target. The approximate shape of the underwater small target is represented by the number of sides. If the approximate shape of the underwater small target is circular, the number of sides is recorded as 0. Finally, the position information of the underwater small target, as well as the pixel area and approximate shape of the underwater small target obtained by optical recognition, are output as the recognition result of the underwater small target, and the focal length of the optical camera is restored to the default focal length.

[0049] It should be noted that the purpose of optical recognition is to improve the recognition success rate. Therefore, it is necessary to calculate the zoom ratio of the optical camera according to the relative distance to increase the pixel area and ensure the accuracy of the optical recognition result.

[0050] A method for recognizing underwater small targets using acoustic-optical fusion proposed in this embodiment first performs acoustic recognition on the target area through a side-scan sonar. Combining expert rules, based on the center point coordinates, pixel area, and approximate shape of the underwater small targets in the acoustic recognition result, a judgment is made on whether to perform optical recognition. For underwater small targets with a relatively high successful recognition probability in the acoustic recognition result, the acoustic recognition result is directly output. For underwater small targets with a relatively low successful recognition probability in the acoustic recognition result, an optical camera is called for optical recognition, and the optical recognition result is output as the recognition result of the underwater small target. This design effectively combines the breadth of acoustic recognition and the accuracy of optical recognition, trusts the acoustic recognition result with a relatively high successful recognition probability, and performs secondary recognition on the acoustic recognition result with a relatively low successful recognition probability through an optical camera, thus well achieving the purpose of improving the recognition accuracy of underwater small targets without sacrificing time complexity. In addition, the introduction of expert rules improves the scientificity and standardization of acoustic-optical fusion, and the design of three-stage fuzzy and probability classification well judges the successful recognition probability of underwater small targets, which effectively improves the universality and generalization ability of underwater small target recognition.

[0051] The step division of the above various methods is only for clear description. When implemented, they can be combined into one step, or some steps can be split into multiple steps. As long as the same logical relationship is included, they are all within the protection scope of this application; adding insignificant modifications to the algorithm or process or introducing insignificant designs, but not changing the core design of its algorithm and process, are all within the protection scope of this application.

[0052] Another embodiment of the present application proposes an underwater small target recognition system using acoustic-optical fusion. The implementation details of an underwater small target recognition system using acoustic-optical fusion proposed in this embodiment will be specifically described below. The following content is only for facilitating understanding of the implementation details and is not necessary for implementing this solution.

[0053] The specific structure of an underwater small target recognition system using acoustic-optical fusion proposed in this embodiment can be as Figure 4 shown, including: an acoustic recognition module 201, an expert processing module 202, a judgment module 203, an optical recognition module 204, and an output module 205.

[0054] The acoustic recognition module 201 is used to obtain an acoustic image of the target area through a side-scan sonar, process the acoustic image using background segmentation technology, and obtain an acoustic recognition result. The acoustic recognition result includes the center point coordinates, pixel area, and approximate shape of the underwater small target.

[0055] The expert processing module 202 is used to obtain expert rules and perform three-stage fuzzy processing, weight calculation, normalization, and probability classification on the acoustic recognition result using the expert rules to obtain the successful recognition probability of each underwater small target.

[0056] The judgment module 203 is used to sequentially judge whether the successful recognition probability of each underwater small target is less than a preset probability threshold. For underwater small targets with a successful recognition probability greater than or equal to the preset probability threshold, it instructs the output module 205 to directly output the acoustic recognition result as the recognition result of the underwater small target. For underwater small targets with a successful recognition probability less than the preset probability threshold, it calls the optical recognition module 204 for processing.

[0057] The optical recognition module 204 is used to perform secondary recognition on underwater small targets with a successful recognition probability less than the preset probability threshold using an optical camera, obtain an optical recognition result, and instruct the output module 205 to output the optical recognition result as the recognition result of the underwater small target.

[0058] It is worth mentioning that each module and module group involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of the present application, units that are not closely related to solving the technical problems proposed in the present application are not introduced in this embodiment, but this does not mean that there are no other units in this embodiment.

[0059] It is not difficult to find that this embodiment is a system embodiment corresponding to the above method embodiment, and this embodiment can be implemented in cooperation with the above method embodiment. The relevant technical details and technical effects mentioned in the above method embodiment are still valid in this embodiment. To avoid repetition, they will not be elaborated here. Correspondingly, the relevant technical details mentioned in this embodiment can also be applied to the above method embodiment.

[0060] Another embodiment of the present application proposes an electronic device, as Figure 5 shown, including: at least one processor 301; and a memory 302 communicatively connected to the at least one processor 301; wherein, the memory 302 stores instructions executable by the at least one processor 301, and the instructions are executed by the at least one processor 301 to enable the at least one processor 301 to execute a method for underwater small target recognition using acoustic-optical fusion as described in the above method embodiment.

[0061] Among them, the memory and the processor are connected by a bus. The bus can include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art, so they will not be further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be an element or multiple elements, such as multiple receivers and transmitters, and provides a unit for communicating with various other devices on the transmission medium.

[0062] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interface, voltage regulation, power management, and other control functions. And the memory can be used to store the data used by the processor when executing operations.

[0063] Another embodiment of the present application proposes a computer-readable storage medium storing a computer program, which when executed by a processor, can implement a method for underwater small target recognition using acoustic-optical fusion as described in the above method embodiment.

[0064] That is, those skilled in the art can understand that all or part of the steps in the above embodiments can be completed by instructing relevant hardware through a program. This program is stored in a storage medium, including several instructions to enable a device (such as a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the methods described in various embodiments of the present application. And the aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories, random access memories, magnetic disks, or optical discs that can store program codes.

[0065] Those of ordinary skill in the art can understand that the above embodiments are all specific embodiments for implementing the present application. In actual applications, various changes can be made to them in form and details without departing from the spirit and scope of the present application.

Claims

1. An underwater small target recognition method using acoustic-optical fusion, characterized in that Including: Obtain the acoustic image of the target area through side-scan sonar, and process the acoustic image using background segmentation technology to obtain the acoustic recognition result; among them, the acoustic recognition result includes the center point coordinates, pixel area, and approximate shape of underwater small targets; Obtain the expert rules, and use the expert rules to perform three-stage fuzzy, weight calculation, normalization, and probability classification on the acoustic recognition result to obtain the successful recognition probability of each underwater small target; Successively determine whether the successful recognition probability of each underwater small target is less than the preset probability threshold; For the underwater small targets with a successful recognition probability greater than or equal to the preset probability threshold, directly output the acoustic recognition result as the recognition result of the underwater small target; For the underwater small targets with a successful recognition probability less than the preset probability threshold, use an optical camera for secondary recognition and output the optical recognition result as the recognition result of the underwater small target.

2. The underwater small target recognition method using acoustic-optical fusion according to claim 1, characterized in that Using background segmentation technology to process the acoustic image to obtain the acoustic recognition result, including: Remove the background from the acoustic image to obtain the center point coordinates of the underwater small target; Extract the pixel area occupied by the underwater small target; among them, the pixel area of the underwater small target is the total number of pixel points occupied by the underwater small target; Approximate the shape of the underwater small target to obtain the approximate shape of the underwater small target. The approximate shape of the underwater small target is expressed as the number of sides. If the approximate shape of the underwater small target is circular, the number of sides is recorded as 0.

3. The underwater small target recognition method using acoustic-optic fusion according to claim 2, characterized in that The expert rules are used to connect acoustic recognition and optical recognition. The design of the expert rules includes two parts: input fuzzification and probability classification. The input fuzzification is implemented in a three-stage fuzzy manner, and the probability classification is implemented in a weighted and normalized manner. The basis for probability classification comes from the offline recognition results of existing acoustic images.

4. A method for underwater small target recognition using acoustic-optical fusion according to claim 3, characterized in that, Implement input fuzzification in a three-stage fuzzy manner, including: Use the acoustic recognition result as the input of the three-stage fuzzy of the expert rules; Calculate the relative distance between the underwater small target and the side-scan sonar based on the center point coordinates of the underwater small target, perform three-stage fuzzy on the relative distance, divide it into three parts: long distance, medium distance, and short distance according to the size of the relative distance, and assign values of 1, 2, and 3 in sequence; Calculate the area ratio of the pixel area of the underwater small target to the total area of the acoustic image, perform three-stage fuzzy on the area ratio, divide it into three parts: small area, medium area, and large area according to the size of the area ratio, and assign values of 1, 2, and 3 in sequence; Perform three-stage fuzzy on the approximate shape of the underwater small target, divide it into three parts: multi-sided, few-sided, and circular according to the number of sides of the approximate shape, and assign values of 1, 2, and 3 in sequence.

5. A method for underwater small target recognition using acoustic-optic fusion according to claim 4, characterized in that, Implement probability classification in a weighted and normalized manner, including: Assign weighted weights to the three-stage fuzzy results of relative distance, pixel area, and approximate shape respectively; According to the assigned weighted weights, perform weighted summation on the three-stage fuzzy results of relative distance, pixel area, and approximate shape, and normalize the weighted summation result to between 0 and 1 to finally obtain the successful recognition probability of each underwater small target.

6. A method for underwater small target recognition using acoustic-optic fusion according to claim 5, characterized in that, The preset probability threshold is 0.85, and the weighted weights assigned to the three-stage fuzzy results of relative distance, pixel area, and approximate shape are 0.4, 0.4, and 0.2 in sequence.

7. A method for underwater small target recognition using acoustic-optical fusion according to any one of claims 1 to 6, characterized in that, For underwater small targets with a successful recognition probability less than the preset probability threshold, an optical camera is used for secondary recognition, and the optical recognition result is output as the recognition result of the underwater small target, including: Obtain the position information of the underwater small target with a successful recognition probability less than the preset probability threshold, and the position information of the underwater small target is derived from the acoustic recognition result; Calculate the relative distance between the optical camera and the underwater small target, and calculate the zoom ratio of the optical camera according to the relative distance to increase the pixel area, so as to obtain an optical image; Approximate the shape of the underwater small target in the obtained optical image to obtain the approximate shape of the underwater small target. The approximate shape of the underwater small target is expressed as the number of sides. If the approximate shape of the underwater small target is circular, the number of sides is recorded as 0; Output the position information of the underwater small target, as well as the pixel area and approximate shape of the underwater small target obtained by optical recognition, as the recognition result of the underwater small target, and restore the focal length of the optical camera to the default focal length.

8. An underwater small target recognition system using the fusion of acoustic and optical signals, characterized in that, Including: An acoustic recognition module, an expert processing module, a judgment module, an optical recognition module, and an output module; The acoustic recognition module is used to obtain the acoustic image of the target area through a side-scan sonar, and process the acoustic image using background segmentation technology to obtain the acoustic recognition result. Among them, the acoustic recognition result includes the center point coordinates, pixel area, and approximate shape of the underwater small target; The expert processing module is used to obtain expert rules, and perform three-stage fuzzy, weight calculation, normalization, and probability classification on the acoustic recognition result using the expert rules to obtain the successful recognition probability of each underwater small target; The judgment module is used to sequentially judge whether the successful recognition probability of each underwater small target is less than the preset probability threshold. For underwater small targets with a successful recognition probability greater than or equal to the preset probability threshold, it instructs the output module to directly output the acoustic recognition result as the recognition result of the underwater small target. For underwater small targets with a successful recognition probability less than the preset probability threshold, it calls the optical recognition module for processing; The optical recognition module is used to perform secondary recognition on the underwater small target with a successful recognition probability less than the preset probability threshold using an optical camera to obtain the optical recognition result, and instructs the output module to output the optical recognition result as the recognition result of the underwater small target.

9. An electronic device, characterized in that, Including: At least one processor; And, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a method for recognizing underwater small targets using acoustic-optical fusion as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it can implement a method for recognizing underwater small targets using acoustic-optical fusion as described in any one of claims 1 to 7.

Images