A method for detecting marine life based on sonar images

Abstract

The present invention relates to the technical field of marine life detection. The present invention proposes a marine life detection method based on sonar images. By emitting multi-frequency sonar signals, multi-frequency sonar images are acquired in a sea area to be detected, and the multi-frequency sonar images are divided into sonar image sequences with strong information stability, weak information stability and information fluctuation. A fusion coefficient is calculated according to the sequence, position and frequency of the sonar images, and the fusion coefficient is used as a weight to perform information fusion on the multi-frequency sonar images. A rectangular recognition area is defined to compare and identify the multi-frequency sonar images with a marine life expert database. The present invention improves the quality of sonar images used for marine life detection by reducing noise information but retaining important information, thereby improving the accuracy of marine life detection based on sonar images.

Description

Technical Field

[0001] The present invention relates to the fields of sonar image processing technology and marine life detection technology, and in particular to a marine life detection method based on sonar images. Background Art

[0002] With the development of modern science and technology, people have begun to use various advanced equipment to detect marine life, among which sonar is the most representative. Sonar can obtain target information by emitting sound waves and receiving returned sound wave signals or only receiving ocean sound wave signals. Compared with submersibles and underwater robots, sonar's signal collection method is more cost-effective and extensive, and can detect larger areas and deeper waters in a shorter time.

[0003] Although sonar technology has made great progress in marine life detection, there are still some technical difficulties that need to be solved. For example, how to effectively use the sound wave information collected by sonar is still a key issue. Due to the complexity of the marine environment, these sound wave information are often accompanied by a lot of noise and interference. The interference of factors such as noise and strong reflection highlights the information related to marine life. Sonar signals and sonar images may cause errors in marine life detection due to noise, strong reflection and other phenomena. Therefore, solving the problem of noise, strong reflection and other phenomena affecting the quality of sonar images is the key to marine life detection and analysis. It is necessary to conduct in-depth research on how to construct high-quality sonar images and image analysis methods. Summary of the invention

[0004] The purpose of the present invention is to provide a method for detecting marine life by analyzing sonar images through image analysis technology, so as to solve the technical problem that the existing detection methods are not effective.

[0005] In order to solve the above problems, the technical solution of the present invention is: a method for detecting marine life based on sonar images, the characteristics and methods of which include the following:

[0006] S100: Use the sonar system to obtain multi-frequency sonar images in the sea area to be detected;

[0007] S200: dividing the multi-frequency sonar image into a plurality of information strong stable sonar image sequences, information weak stable sonar image sequences or information fluctuation sonar image sequences;

[0008] S300: calculating the fusion coefficient of each sonar image and performing information fusion on the multi-frequency sonar images;

[0009] S400: Define a rectangular identification area for the sonar image after information fusion, and compare and identify it with the expert database of marine organisms.

[0010] Further, in step S100, the sea area to be detected is scanned by the sonar system, specifically: a series of excitation sound wave signals of different frequencies are emitted to the sea area to be detected by the sonar system, where the frequency refers to the working frequency of the sonar system, that is, the acoustic frequency of the underwater acoustic transducer, and these sound wave signals of different frequencies include low-frequency sound wave signals (sound wave frequency is less than or equal to 20kHz), medium-frequency sound wave signals (sound wave frequency is greater than 20kHz, but less than 100kHz) and high-frequency sound wave signals (sound wave frequency is greater than or equal to 100kHz); compared with the traditional sonar system Different from the sonar system which only uses a single frequency sound wave signal for detection, the multi-frequency sound wave signal can provide richer information; the reflected multi-frequency echo signal is received by the sonar system's receiver, and at each frequency, each frame of the echo signal received by the sonar system is recorded as a column of longitudinal graph data; the received multi-frequency echo signal is subjected to phase correction and amplitude modulation correction processes to extract relevant information of each frequency echo data; the above-mentioned process of sonar sending and receiving sound wave signals is repeated until the set number of times is reached, thereby obtaining the corrected multi-frequency echo signal;

[0011] Further, the echo signal of each frequency in the corrected multi-frequency echo signal is independently spliced ​​into grayscale images according to the receiving time sequence as multiple sonar images, wherein the multiple sonar images under multiple frequencies are recorded as multi-frequency sonar images, and the corresponding grayscale value is allocated according to the echo signal strength value. The larger the signal strength value, the closer the grayscale value is to 255, and the smaller the signal strength value, the closer the grayscale value is to 0. The echo signals of all frames of each frequency constitute a corresponding sonar image. If there are sound waves of N frequencies and their echo signals, there are N sonar images, wherein N is a non-negative integer and the minimum value is 0; the more times the sonar system is set to repeat the process of emitting sound waves and receiving echoes, the longer the time sequence;

[0012] Taking into account the computational load and memory limitations, the number of excitation sound waves of different frequencies emitted by the sonar system, the number of times the echoes are received (i.e., the time series length of the echoes received by the sonar system) and the sampling frequency of the sound wave signal can be adjusted according to different sea areas, sea area conditions and the needs of the detection task; when conducting marine biological detection in shallow waters, the time series length of the echoes received by the sonar system is shortened or the sampling frequency of the medium and low frequency sound waves emitted by the sonar system is reduced. For example, in a certain sampling, in deep water areas, the sampling frequencies for medium frequency sound waves are 20kHz, 30kHz, 40kHz, 50kHz, 60kHz, 70kHz, 80kHz, 90kHz, and the time series length of the echoes received is 1000; in shallow water areas, the sampling frequencies for medium frequency waves are changed to 20kHz, 40kHz, 60kHz, 80kHz, and the time series length of the echoes received is changed to 500, reducing the sampling frequency of medium frequency waves and shortening the time series length of the echoes received;

[0013] The present invention does not directly fuse the multi-frequency sonar images in a weighted summation manner. There will always be some noisy sonar image information in the multi-frequency sonar images. This fusion method is likely to introduce additional noise interference to the sonar images after information fusion, thereby affecting the effect of using sonar images for marine life detection.

[0014] Preferably, in step S200, the multi-frequency sonar images are divided into a sequence of information strong stable sonar images, a sequence of information weak stable sonar images, or a sequence of information fluctuation sonar images, and the steps are as follows:

[0015] S201, sort the multi-frequency sonar images in ascending order of frequency to obtain a set G; the number of multi-frequency sonar images in the set G is denoted as I, where I is a positive integer, and i is the sequence number of the element in the set G, where i∈[0, I-1] is an integer, and G i represents the i-th multi-frequency sonar image in the set G;

[0016] S202, calculate the average or median of the maximum grayscale value and the average grayscale value of the grayscale image of the multi-frequency sonar image in the set G as thresholds Thr1 and Thr2 respectively, and calculate the thresholds of two grayscale images G with adjacent serial numbers in the set G in ascending order. i and G i+1 Pixel grayscale value, where SUM is the operation of finding the sum of grayscale values ​​in the image, SUM(G i ) indicates G i The sum of the grayscale values ​​of pixels in the image, SUM(G i+1 ) represents the G in the image i+1The sum of the grayscale values ​​of pixels; there are I-1 elements in the list Red, j is the serial number of the difference element in the list Red, j∈[0,I-2] and is an integer, Red j Represents the jth element in the list Red;

[0017] S203, set set P to be an empty set (the elements of set P can be a set composed of one or more difference elements in list Red), define the difference elements in list Red whose absolute values ​​are greater than Thr2 as boundary difference elements; use the boundary difference elements as the division boundary (two or more boundary difference elements with consecutive serial numbers are also regarded as a boundary) to divide the non-boundary difference elements in list Red into different sets (the absolute value of the boundary difference element is greater than the threshold Thr2, and the information fluctuation between the two sonar images corresponding to the boundary difference element is large, indicating that the acoustic signal is strongly interfered and contains a lot of noise, so the boundary difference elements are excluded when dividing the differences in list Red), and each set is added to set P as a subset; let k be the serial number of the subset in set P, k∈[0, K-1] and is an integer, P k represents the kth subset in the set P; the specific steps are as follows:

[0018] (The elements in the list Red reflect the information fluctuation of two adjacent sonar images of different frequencies in the set G. When two adjacent sonar images of corresponding frequencies have large information fluctuations (the appearance of boundary difference elements), in addition to being strongly interfered, it also shows that the sonar image information of the front and rear frequency bands corresponding to the boundary difference elements is greatly different. Therefore, the list Red is divided using the boundary difference elements as the dividing boundary);

[0019] Set the initial value of integer variable Ac and Boolean variable TCin to 0, the initial value of j to 0, and traverse the value range of j from small to large, repeating the following traversal process until j=I-2:

[0020] When j=0:

[0021] If | Red j |≤Thr2, let j and Ac increase by 1 respectively; if |Red j |>Thr2, set TCin to 1 and increase the value of j by 1;

[0022] When j≠0:

[0023] If | Red j |≤Thr2, set TCin to 0, and increase the values ​​of j and Ac by 1 respectively;

[0024] When | Red jWhen Thr2 and TCin is 0, take the elements from Red to Red in the list Red as a subset and add them to the set P, reset the value of Ac to 0 and set TCin to 1, and increment the value of j by 1; j-Ac to Red j-1 in the set P, reset the value of Ac to 0 and set TCin to 1, and increment the value of j by 1;

[0025] When |Red j |>Thr2 and TCin is 1, increment the value of j by 1;

[0026] S204. Divide the sonar image sequences in the set G into information strongly stable sonar image sequences, information weakly stable sonar image sequences or information fluctuating sonar image sequences according to the information fluctuation situation of the sonar image sequences. Set the initial value of k to 0, traverse the value range of k from small to large, and repeat the following traversal process until k = K - 1:

[0027] Calculate the average value or median Ame of the absolute values of all elements in the set Pk. Sort the sonar images in the set G that calculate the difference elements of the set Pk in ascending order according to their corresponding frequencies and remove the sonar image with the largest corresponding frequency to obtain the sorted sonar images. If Ame < 0.5×Thr1, divide the sorted sonar images into an information strongly stable sonar image sequence. If 0.5×Thr1 ≤ Ame < Thr1, divide the sorted sonar images into an information weakly stable sonar image sequence. If Ame ≥ Thr1, divide the sorted sonar images into an information fluctuating sonar image sequence;

[0028] After finishing the traversal of the set P, obtain multiple information strongly stable sonar image sequences, information weakly stable sonar image sequences and information fluctuating sonar image sequences of the sonar images in the set G. Denote that there are m sonar images in the set G that are not divided into information strongly stable sonar image sequences, information weakly stable sonar image sequences or information fluctuating sonar image sequences, where m ∈ [0, I - 1] and is an integer. Remove these m undivided sonar images from the set G (the undivided sonar images correspond to the boundary difference elements in Red. The acoustic wave signals of these sonar images have large fluctuations and a lot of noise due to interference (such as strong reflection). Adding these sonar images to the information fusion of multi-frequency sonar images is not conducive to using sonar images for marine biological detection). At this time, the number of sonar images in the set G is I - m;

[0029] Furthermore, in step S300, calculate the fusion coefficient i of each sonar image G in the set G , and its calculation method is as follows:

[0030] ;

[0031] where ,

[0032] , SUM is the operation of finding the sum of the grayscale values ​​of the image, e is a natural constant, ln is the logarithm with the constant e as the base, u and s represent G i The end sequence number and the start sequence number of the sonar image sequence (information strong stable sonar image sequence or information weak stable sonar image sequence or information fluctuation sonar image sequence) in the set G, F i , F u and F s Represents the sonar image G i , G u and G s The corresponding frequency of

[0033] (This method of calculating the fusion coefficient takes into account the impact of multiple variables and factors (the gray value (or energy value) of each pixel of the sonar image, the frequency corresponding to the sonar image, the frequency segment of the sonar image sequence to which the sonar image belongs, and the information fluctuation between the sonar image and sonar images with adjacent frequencies) on the sonar image information in a more comprehensive manner, and can more robustly filter out the interference of noise and strong reflection on the sonar image without losing important information);

[0034] Further, in step S300, the fusion coefficients of all sonar images in the set G are normalized, and the normalized fusion coefficient is used as the weight of each sonar image. The grayscale value of each sonar image in the set G is weighted and summed with the normalized fusion coefficient as the weight to obtain the information fused sonar image. The information fused sonar image is denoted as Y, and the calculation method is as follows:

[0035] ;

[0036] The beneficial effects are as follows: after dividing the multi-frequency sonar images into different sets, the fusion coefficients of the sonar images of different frequencies are calculated by considering the influence of various factors on the sonar image information, so that the importance of the sonar image information of each frequency can be more accurately evaluated; some small organisms and large phytoplankton will present pseudo-noise points in the sonar images that are the same as noise. In order to distinguish the pseudo-noise points from the real noise, after calculating the fusion coefficients of the sonar images of different frequencies, the multi-frequency sonar images are fused to accurately remove the real noise and retain the pseudo-noise points; screening out more effective information as the final fusion information can effectively alleviate the influence of the noise information in the sonar images on the fusion of the sonar image information; at the same time, using the multi-frequency sonar images for information fusion can also make the sonar image information richer, so that the sonar system can have a larger detection range in the sea area to be detected and detect more diverse marine life;

[0037] Further, in step S400, the sonar grayscale image obtained in S300 is converted into an RGB three-channel sonar image according to the size of its pixel grayscale value. The color of the pixel with a larger grayscale value in the sonar grayscale image is closer to red after conversion, and the color of the pixel with a smaller grayscale value is closer to blue after conversion. The RGB three-channel sonar image obtained after conversion is traversed according to a rectangular frame of a predetermined window size. During the traversal, each time the rectangular frame reaches a sonar image area, the sonar image of the area is compared and identified with an expert database of marine organisms (the expert database includes a bow model library of marine organisms). If the comparison and identification is successful, the type and distribution position of the marine organisms in the area are obtained according to the result of the comparison and identification, thereby completing the detection of marine organisms in the sonar image area. Otherwise, it means that there is no marine organism that needs to be detected in the sonar image area.

[0038] The present invention also provides a marine life detection system based on sonar images, which includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the marine life detection method based on sonar images are implemented. The marine life detection system based on sonar images can be run on computing devices such as desktop computers, laptop computers, PDAs, and cloud data centers. The executable system may include, but is not limited to, a processor, a memory, and a server cluster.

[0039] The beneficial effects of the present invention are as follows: the present invention provides a method for detecting marine life based on sonar images, which integrates various information and comprehensively considers the influence of various factors on the detection of marine life, optimizes the sonar image information by combining the information of each frequency, reduces the interference of factors such as noise and strong reflection, and highlights the information related to marine life, reduces the phenomenon of marine life detection errors caused by noise, strong reflection and other phenomena in sonar signals and sonar images, and improves the accuracy of marine life detection using sonar image information. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The above and other features of the present invention will become more obvious by describing in detail the embodiments shown in the accompanying drawings. The same reference numerals in the accompanying drawings of the present invention represent the same or similar elements. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other accompanying drawings can be obtained based on these accompanying drawings without creative work. In the accompanying drawings:

[0041] Figure 1 Shown is a flow chart of a method for detecting marine life based on sonar images. DETAILED DESCRIPTION

[0042] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention, so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0043] like Figure 1 The following is a flow chart of a method for detecting marine life based on sonar images. Figure 1 A method for detecting marine life based on sonar images according to an embodiment of the present invention is described below. The method comprises the following steps:

[0044] S100: Use the sonar system to obtain multi-frequency sonar images in the sea area to be detected;

[0045] S200: dividing the multi-frequency sonar image into a plurality of information strong stable sonar image sequences, information weak stable sonar image sequences or information fluctuation sonar image sequences;

[0046] S300: calculating the fusion coefficient of each sonar image and performing information fusion on the multi-frequency sonar images;

[0047] S400: Define a rectangular identification area for the sonar image after information fusion, and compare and identify it with the expert database of marine organisms.

[0048] Further, in step S100, a sea area to be detected with an area of ​​20 km×20 km is scanned by a sonar system, specifically: a series of excitation sound wave signals of different frequencies are emitted to the sea area to be detected by the sonar system, where the frequency refers to the working frequency of the sonar system, that is, the acoustic frequency of the hydroacoustic transducer, and these sound wave signals of different frequencies include low-frequency sound wave signals (sound wave frequency is less than or equal to 20 kHz), medium-frequency sound wave signals (sound wave frequency is greater than 20 kHz, but less than 100 kHz) and high-frequency sound wave signals (sound wave frequency is greater than or equal to 100 kHz); and Unlike traditional sonar systems that only use single-frequency sound wave signals for detection, multi-frequency sound wave signals can provide richer information; the reflected multi-frequency echo signal is received by the receiver of the sonar system, and at each frequency, each frame of the echo signal received by the sonar system is recorded as a column of longitudinal graph data; the received multi-frequency echo signal is subjected to phase correction and amplitude modulation correction processes to extract relevant information of each frequency echo data; the above-mentioned process of sonar sending and receiving sound wave signals is repeated until the set number of times is reached, thereby obtaining the corrected multi-frequency echo signal;

[0049] Further, the echo signal of each frequency in the corrected multi-frequency echo signal is independently spliced ​​into grayscale images according to the receiving time sequence as multiple sonar images, wherein the multiple sonar images under multiple frequencies are recorded as multi-frequency sonar images, and the corresponding grayscale values ​​are allocated according to the echo signal strength values. The larger the signal strength value, the closer the grayscale value is to 255, and the smaller the signal strength value, the closer the grayscale value is to 0. The echo signals of all frames of each frequency constitute a corresponding sonar image. In this embodiment, there are sound waves of 1000 frequencies and their echo signals, and then there are 1000 sonar images.

[0050] The present invention does not directly fuse the multi-frequency sonar images in a weighted summation manner. There will always be some noisy sonar image information in the multi-frequency sonar images. This fusion method is likely to introduce additional noise interference to the sonar images after information fusion, thereby affecting the effect of using sonar images for marine life detection.

[0051] Preferably, in step S200, the multi-frequency sonar images are divided into a sequence of information strong stable sonar images, a sequence of information weak stable sonar images, or a sequence of information fluctuation sonar images, and the steps are as follows:

[0052] S201, sort the multi-frequency sonar images in ascending order of frequency to obtain a set G; the number of multi-frequency sonar images in the set G is denoted as I (in this embodiment, I is an integer 1000), where I is a positive integer, i is the sequence number of the element in the set G, i∈[0, I-1] and is an integer, G i represents the i-th multi-frequency sonar image in the set G;

[0053] S202, calculate the average or median of the maximum grayscale value and the average grayscale value of the grayscale image of the multi-frequency sonar image in the set G as thresholds Thr1 and Thr2 respectively, and calculate the thresholds of two grayscale images G with adjacent serial numbers in the set G in ascending order. i and G i+1 Pixel gray value, G i and G i+1 The difference of the sum of pixel gray values ​​is calculated as SUM(G i+1 )-SUM(G i ) or SUM(G i )-SUM(G i+1 ), where SUM is the operation of finding the sum of the gray values ​​in the image, SUM(G i ) indicates G i The sum of the grayscale values ​​of pixels in the image, SUM(G i+1 ) represents the G in the image i+1The sum of the grayscale values ​​of pixels; there are I-1 elements in the list Red, j is the serial number of the difference element in the list Red, j∈[0,I-2] and is an integer, Red j Represents the jth element in the list Red;

[0054] S203, set set P to be an empty set (the elements of set P can be a set consisting of one or more difference elements in list Red), define the difference elements in list Red whose absolute values ​​are greater than Thr2 as boundary difference elements; use the boundary difference elements as the dividing boundary (two or more boundary difference elements with consecutive serial numbers are also regarded as a boundary) to divide the non-boundary difference elements in list Red into different sets, and each set is added to set P as a subset; let k be the serial number of the subset in set P, k∈[0, K-1] and is an integer, P k represents the kth subset in the set P; the specific steps are as follows:

[0055] Set the initial value of integer variable Ac and Boolean variable TCin to 0, the initial value of j to 0, and traverse the value range of j from small to large, repeating the following traversal process until j=I-2:

[0056] When j=0:

[0057] If | Red j |≤Thr2, let j and Ac increase by 1 respectively; if |Red j |>Thr2, set TCin to 1 and increase the value of j by 1;

[0058] When j≠0:

[0059] If | Red j |≤Thr2, set TCin to 0, and increase the values ​​of j and Ac by 1 respectively;

[0060] When | Red j |>Thr2 and TCin is 0, Red in the list Red j-Ac To Red j-1 Add the elements of as a subset to the set P, reset the value of Ac to 0 and TCin to 1, and increase the value of j by 1;

[0061] When | Red j |>Thr2 and TCin is 1, increase the value of j by 1;

[0062] S204. Divide the sonar image sequences in set G into information strongly stable sonar image sequences, information weakly stable sonar image sequences, or information fluctuating sonar image sequences according to the information fluctuation situation of the sonar image sequences. Set the initial value of k to 0, traverse the value range of k from small to large, and repeat the following traversal process until k = K - 1:

[0063] Calculate set P k The average value or median Ame of the absolute values of all elements, and sort the sonar images of the difference elements in set G obtained from set P k in ascending order according to their corresponding frequencies and remove the sonar image with the largest corresponding frequency to obtain the sorted sonar images. If Ame < 0.5 × Thr1, divide the sorted sonar images into an information strongly stable sonar image sequence; if 0.5 × Thr1 ≤ Ame < Thr1, divide the sorted sonar images into an information weakly stable sonar image sequence; if Ame ≥ Thr1, divide the sorted sonar images into an information fluctuating sonar image sequence;

[0064] After finishing the traversal of set P, obtain multiple information strongly stable sonar image sequences, information weakly stable sonar image sequences, and information fluctuating sonar image sequences of the sonar images in set G. Denote that there are m sonar images in set G that are not divided into information strongly stable sonar image sequences, information weakly stable sonar image sequences, or information fluctuating sonar image sequences, where m ∈ [0, I - 1] and is an integer. Remove these m undivided sonar images from set G. At this time, the number of sonar images in set G is I - m;

[0065] Furthermore, in step S300, calculate the fusion coefficient i of each sonar image G in set G , and its calculation method is as follows:

[0066] ;

[0067] where ,

[0068] , SUM is the operation of summing the image grayscale values, e is the natural constant, ln is the logarithm with the constant e as the base, and u and s respectively represent the ending serial number and starting serial number of the sonar image sequence (information strongly stable sonar image sequence or information weakly stable sonar image sequence or information fluctuating sonar image sequence) where G i is located in set G, F i , F u and F s respectively represent the corresponding frequencies of sonar images G i , G u and G s ;

[0069] Further, in step S300, the fusion coefficients of all sonar images in the set G are normalized, and the normalized fusion coefficient is used as the weight of each sonar image. The grayscale value of each sonar image in the set G is weighted and summed with the normalized fusion coefficient as the weight to obtain the information fused sonar image. The information fused sonar image is denoted as Y, and the calculation method is as follows:

[0070] ;

[0071] Further, in step S400, the sonar grayscale image obtained in S300 is converted into an RGB three-channel sonar image according to the size of its pixel grayscale value. The color of the pixel with a larger grayscale value in the sonar grayscale image is closer to red after conversion, and the color of the pixel with a smaller grayscale value is closer to blue after conversion. The RGB three-channel sonar image obtained after conversion is traversed according to a rectangular frame of a predetermined window size. During the traversal, each time the rectangular frame reaches a sonar image area, the sonar image of the area is compared and identified with an expert database of marine organisms (the expert database includes a bow model library of marine organisms). If the comparison and identification is successful, the type and distribution position of the marine organisms in the area are obtained according to the result of the comparison and identification, thereby completing the detection of marine organisms in the sonar image area. Otherwise, it means that there is no marine organism that needs to be detected in the sonar image area.

[0072] An embodiment of the present invention provides a method for detecting marine life based on sonar images. The method for detecting marine life based on sonar images in this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned embodiment of the method for detecting marine life based on sonar images are implemented.

[0073] The method for detecting marine life based on sonar images can be run on computing devices such as desktop computers, laptops, PDAs, and cloud servers. The system that can run the method for detecting marine life based on sonar images may include, but is not limited to, processors and memories. Those skilled in the art will understand that the example is only an example of a method for detecting marine life based on sonar images, and does not constitute a limitation on a method for detecting marine life based on sonar images. It may include more or fewer components than the example, or a combination of certain components, or different components. For example, the system for detecting marine life based on sonar images may also include input and output devices, network access devices, buses, etc.

[0074] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the marine biological detection method based on sonar images, and uses various interfaces and lines to connect the various parts of the entire marine biological detection method based on sonar images.

[0075] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the marine biological detection method based on sonar images by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0076] Although the description of the present invention has been quite detailed and has been described in particular with respect to several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, so as to effectively cover the intended scope of the present invention. In addition, the present invention is described above with the embodiments foreseeable by the inventors, and its purpose is to provide a useful description, and those non-substantial changes to the present invention that are not currently foreseen may still represent equivalent changes of the present invention.

Claims

1. A method for detecting marine life based on sonar images, characterized in that: The method comprises the following steps: S100: Use the sonar system to obtain multi-frequency sonar images in the sea area to be detected; S200: dividing the multi-frequency sonar image into a plurality of information strong stable sonar image sequences, information weak stable sonar image sequences or information fluctuation sonar image sequences; S300: calculating the fusion coefficient of each sonar image and performing information fusion on the multi-frequency sonar images; S400: Define a rectangular recognition area for the sonar image after information fusion, and compare and identify it with the expert database of marine organisms; In S200, the multi-frequency sonar images are divided into a sequence of information strong stable sonar images, a sequence of information weak stable sonar images, or a sequence of information fluctuation sonar images, and the steps are as follows: S201, sort the multi-frequency sonar images in ascending order of frequency to obtain a set G; the number of multi-frequency sonar images in the set G is denoted as I, where I is a positive integer, and i is the sequence number of the element in the set G, where i∈[0, I-1] is an integer, and G i represents the i-th multi-frequency sonar image in the set G; S202, calculate the average or median of the maximum grayscale value and the average grayscale value of the grayscale image of the multi-frequency sonar image in the set G as thresholds Thr1 and Thr2 respectively, and calculate the thresholds of two grayscale images G with adjacent serial numbers in the set G in ascending order. i and G i+1 The difference of the sum of the pixel gray values ​​forms a list Red. There are I-1 elements in the list Red. j is the serial number of the difference element in the list Red. j∈[0, I-2] is an integer. Redj represents the jth element in the list Red. S203, set the set P to be an empty set, define the difference elements in the list Red whose absolute values ​​are greater than Thr2 as boundary difference elements; use the boundary difference elements as the dividing boundary to divide the non-boundary difference elements in the list Red into different sets, and each set is added to the set P as a subset; let k be the sequence number of the subset in the set P, k∈[0, K-1] and is an integer, P k represents the kth subset in the set P; S204, traversing the set P to divide the sonar image sequences in the set G into a plurality of information strong stable sonar image sequences, information weak stable sonar image sequences or information fluctuation sonar image sequences.

2. A method for detecting marine life based on sonar images according to claim 1, characterized in that: In step S100, the sea area to be detected is scanned by the sonar system, specifically: a series of excitation sound wave signals of different frequencies are emitted to the sea area to be detected by the sonar system, and the reflected multi-frequency echo signals are received by the receiver of the sonar system. At each frequency, the echo signal of each frame received by the sonar system is recorded as a column of longitudinal graph data; the received multi-frequency echo signal is subjected to phase correction and amplitude modulation correction to extract relevant information of the echo data of each frequency; the above-mentioned process of emitting and receiving sound wave signals by the sonar is repeated until the set number of times is reached, thereby obtaining the corrected multi-frequency echo signal.

3. The method for detecting marine life based on sonar images according to claim 1, characterized in that: In S100, the echo signals of each frequency in the corrected multi-frequency echo signals are independently spliced ​​into grayscale images according to the receiving time sequence as multiple sonar images, and the corresponding grayscale values ​​are allocated according to the echo signal strength values. The larger the signal strength value, the closer the grayscale value is to 255, and the smaller the signal strength value, the closer the grayscale value is to 0. The echo signals of all frames of each frequency constitute a corresponding sonar image.

4. The method for detecting marine life based on sonar images according to claim 1, characterized in that: In S203, the steps of dividing the non-boundary difference elements in the list Red into different sets by using the boundary difference elements as the dividing boundaries are as follows: Set the initial value of integer variable Ac and Boolean variable TCin to 0, the initial value of j to 0, and traverse the value range of j from small to large, repeating the following traversal process until j=I-2: When j=0: If | Red j |≤Thr2, let j and Ac increase by 1 respectively; if |Red j |>Thr2, set TCin to 1 and increase the value of j by 1; When j≠0: If | Red j |≤Thr2, set TCin to 0, and increase the values ​​of j and Ac by 1 respectively; When | Red j |>Thr2 and TCin is 0, Red in the list Red j-Ac To Red j-1 Add the elements of as a subset to the set P, reset the value of Ac to 0 and TCin to 1, and increase the value of j by 1; When | Red j |>Thr2 and TCin is 1, increase the value of j by 1.

5. The method for detecting marine life based on sonar images according to claim 1, characterized in that: In S204, the specific steps of traversing the set P to divide the sonar image sequences in the set G into a plurality of information strong stable sonar image sequences, information weak stable sonar image sequences or information fluctuation sonar image sequences are as follows: Set the initial value of k to 0, traverse the value range of k from small to large, and repeat the following traversal process until k=K-1: Calculate set P k The average value or median Ame of the absolute values of all elements, and calculate set P from set G k Sort the sonar images of the difference elements in ascending order according to their corresponding frequencies and remove the sonar image with the largest corresponding frequency to obtain the sorted sonar images; if Ame < 0.5 × Thr1, divide the sorted sonar images into an information strongly stable sonar image sequence; if 0.5 × Thr1 ≤ Ame < Thr1, divide the sorted sonar images into an information weakly stable sonar image sequence; if Ame ≥ Thr1, divide the sorted sonar images into an information fluctuating sonar image sequence.

6. The method for detecting marine life based on sonar images according to claim 1, characterized in that: In S300, each sonar image G in the set G is calculated i The fusion coefficient , which is calculated as follows: ； in , , SUM is the operation of finding the sum of the grayscale values ​​of the image, e is a natural constant, ln is the logarithm with the constant e as the base, u and s represent G i The end and start numbers of the sonar image sequence in the set G, F i , F u and F s Represents the sonar image G i , G u and G s The corresponding frequency of In S300, the fusion coefficients of all sonar images in the set G are normalized, and the normalized fusion coefficient is used as the weight of each sonar image. The grayscale value of each sonar image in the set G is weighted and summed with the normalized fusion coefficient as the weight to obtain the information fused sonar image. The information fused sonar image is denoted as Y, and the calculation method is as follows: 。 7. The method for detecting marine life based on sonar images according to claim 1, characterized in that: In S400, the sonar grayscale image obtained in S300 is converted into an RGB three-channel sonar image according to the size of its pixel grayscale value. The pixel with a larger grayscale value in the sonar grayscale image is closer to red after conversion, and the pixel with a smaller grayscale value is closer to blue after conversion. The RGB three-channel sonar image obtained after conversion is traversed according to a rectangular frame of a predetermined window size. Whenever the rectangular frame reaches a sonar image area, the sonar image of the area is compared and identified with an expert database of marine organisms.

Images