Multi-channel whale-dolphin whistling detection method and passive sonar system
Through the multi-channel whale and dolphin whisker sound detection method in time domain downsampling and envelope calculation, the problem of high computational complexity in the existing technology is solved, and the detection of whale and dolphin whisker sounds and marine biological monitoring are realized.
Patent Information
- Application Number
- CN202411720205.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-28
- Publication Date
- 2025-02-25
AI Technical Summary
The prior art has high computational complexity in monitoring whale and dolphin sing sounds, making it difficult to monitor the sing sounds of various marine organisms on a large scale.
The multi-channel whale and dolphin whisker sound detection method is used to reduce the calculation amount by performing sound wave downsampling, envelope calculation, amplitude, pulse width and frequency judgment in the time domain, and to process the sound wave signals of different frequency bands in multiple channels respectively.
The detection of various characteristics of whales and porcines has been realized, and the calculation volume has been reduced. It is suitable for marine ecological environment protection and marine mammal abundance survey.
Smart Images
Figure CN119517093A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of ocean acoustic monitoring technology, and in particular to a multi-channel cetacean call detection method and a passive sonar system. Background Art
[0002] With the increasing attention paid to marine ecological protection around the world, the protection and monitoring of marine mammals such as cetaceans have become an important topic in marine biological research. The calls of cetaceans are an important means for them to communicate, navigate, and locate prey in the marine environment. Therefore, accurate and real-time monitoring of the calls of cetaceans is of great significance for protecting these species, understanding their behavioral patterns, and conducting scientific research.
[0003] At present, the monitoring of cetacean calls mainly relies on passive sonar technology. Passive sonar systems can receive sound waves emitted by underwater sound sources. Because they do not cause noise pollution, they are more suitable for monitoring marine life such as cetaceans. Traditional passive sonar recognition algorithms usually target sounds with specific spectral characteristics. By modeling and detecting specific spectral characteristics, the algorithm can accurately identify the sounds of cetaceans and reduce interference from other noise or stray sound sources. However, this type of algorithm requires precise prior knowledge, has a high computational complexity, and has obvious deficiencies when investigating marine biological resources, and cannot monitor a variety of marine life on a large scale. Summary of the invention
[0004] The purpose of this application is to provide a multi-channel cetacean call detection method and a passive sonar system, which can reduce the amount of calculation and realize the detection of multiple characteristic calls of cetaceans at the same time.
[0005] To achieve the above objectives, this application provides the following solutions:
[0006] In the first aspect, the present application provides a multi-channel cetacean call detection method, comprising: setting a plurality of channels for receiving acoustic wave signals of cetacean calls of different frequency bands; setting a noise collection mode and a detection mode to be performed sequentially in each cycle; when executing the noise collection mode, downsampling the sound waves of the frequency band received by each channel, and determining the sound wave envelope after downsampling; determining the amplitude threshold of each channel according to the sound wave envelope after downsampling of each channel; after starting to execute the detection mode, for each channel that receives the sound wave to be detected, downsampling the sound wave to be detected of the frequency band received by the channel, and determining The method comprises the following steps: determining the sound wave envelope to be detected after downsampling; identifying the number of sound wave envelope sample points to be detected that continuously exceed the amplitude threshold of each channel in the cycle period; if the number of identified sound wave envelope sample points to be detected is greater than or equal to the preset pulse width threshold, determining that the sound wave to be detected in the frequency band received by the channel is the sound wave signal of the cetacean call, and defining the channel as the target channel; when there is only one target channel, taking the frequency band of the sound wave signal of the target channel after downsampling as the frequency range of the cetacean call; when there are multiple target channels, fusing the frequency bands of the sound wave signals of the multiple target channels after downsampling to determine the frequency range of the cetacean call.
[0007] In a second aspect, the present application provides a passive sonar system, which adopts the above-mentioned multi-channel whale and dolphin call detection method.
[0008] According to the specific embodiments provided in this application, this application has the following technical effects:
[0009] The present application provides a multi-channel cetacean call detection method and a passive sonar system, which reduces the amount of computation compared to the traditional passive sonar recognition algorithm for cetacean call detection in the frequency domain by downsampling, envelope calculation, amplitude, pulse width and frequency discrimination of sound waves in the time domain; at the same time, the sound wave signal processing of cetacean calls in different frequency bands is performed on multiple channels, thereby realizing the detection of multiple characteristic calls of cetaceans. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0011] Figure 1 A schematic diagram of a flow chart of a multi-channel cetacean call detection method provided in an embodiment of the present application;
[0012] Figure 2 A schematic diagram of the principle of a multi-channel cetacean call detection method provided in one embodiment of the present application;
[0013] Figure 3 A schematic diagram of a sliding average provided for another embodiment of the present application. DETAILED DESCRIPTION
[0014] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0015] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0016] In an exemplary embodiment, Figure 1 As shown, a multi-channel cetacean call detection method is provided, comprising the following steps 101 to 107. Among them:
[0017] Step 101: multiple channels are set for receiving acoustic wave signals of cetacean calls in different frequency bands.
[0018] Step 102: setting the noise collection mode and the detection mode to be performed sequentially in each cycle.
[0019] Step 103: When executing the noise collection mode, down-sample the sound waves in the receiving frequency band of each channel, and determine the envelope of the down-sampled sound waves.
[0020] Step 104: Determine the amplitude threshold of each channel according to the down-sampled sound wave envelope of each channel.
[0021] Step 105: After the detection mode is started, for each channel that receives the sound wave to be detected, the sound wave to be detected in the frequency band received by the channel is downsampled, and the envelope of the sound wave to be detected after downsampling is determined; the number of sample points of the sound wave envelope to be detected that continuously exceeds the amplitude threshold of each channel in the cycle period is identified; if the number of sample points of the sound wave envelope to be detected is greater than or equal to the preset pulse width threshold, it is determined that the sound wave to be detected in the frequency band received by the channel is a sound wave signal of the cetacean call, and the channel is defined as a target channel.
[0022] Step 106: When there is only one target channel, the frequency band of the downsampled sound wave signal of the target channel is used as the frequency range of the cetacean call.
[0023] Step 107: When there are multiple target channels, the frequency bands of the downsampled sound waves of the multiple target channels are fused to determine the frequency range of the cetacean call.
[0024] By implementing the above steps 101 to 107, the amount of computation is reduced compared to the detection of cetacean calls in the frequency domain by the traditional passive sonar recognition algorithm by downsampling the sound waves, envelope calculation, amplitude, pulse width and frequency discrimination in the time domain; at the same time, the sound wave signal processing of cetacean calls in different frequency bands is performed in multiple channels respectively, thereby realizing the detection of multiple characteristic calls of cetaceans.
[0025] In another exemplary embodiment of the present application, the frequency and pulse width of cetacean calls have a large span, ranging from 10 Hz to more than 100 kHz; the pulse width is as short as hundreds of microseconds and as long as tens of seconds. The present application processes signals of different frequency bands separately. According to the frequency range and pulse width characteristics of cetacean calls, the received signal is divided into multiple frequency bands. Then the above step 101 can be specifically as follows:
[0026] The first channel is set to receive the sound wave signal of the whale call sound of the first frequency band, the second channel is set to receive the sound wave signal of the whale call sound of the second frequency band, the third channel is set to receive the sound wave signal of the whale call sound of the third frequency band, and the fourth channel is set to receive the sound wave signal of the whale call sound of the fourth frequency band; the frequency range of the first frequency band is greater than or equal to 10 Hz and less than 100 Hz, the frequency range of the second frequency band is greater than or equal to 100 Hz and less than 1 kHz, the frequency range of the third frequency band is greater than or equal to 1 kHz and less than 10 kHz, and the frequency range of the fourth frequency band is greater than or equal to 10 kHz and less than 100 kHz.
[0027] In another exemplary embodiment of the present application, the method of the present application is divided into two modes: a noise collection mode and a detection mode. In each cycle, the duration of receiving sound waves in the noise collection mode is 5 seconds, and the duration of receiving sound waves in the detection mode is 30 minutes.
[0028] In another exemplary embodiment of the present application, in order to reduce the amount of data and improve processing efficiency, the sound waves of each frequency band are downsampled. Specifically, different sampling rates are used for different frequency bands. Then the above step 103 of downsampling the sound waves of each channel receiving frequency band can be replaced by the following steps 201 to 202:
[0029] Step 201: Determine that the sampling rate of the first channel is 1k, the sampling rate of the second channel is 10k, the sampling rate of the third channel is 100k, and the sampling rate of the fourth channel is 1MHz.
[0030] Step 202: down-sample the sound waves in the receiving frequency band of each channel according to the sampling rate of each channel.
[0031] In another exemplary embodiment of the present application, in order to remove high-frequency noise in the sound wave and extract the main features of the signal, a sliding window technique is used to smooth the downsampled signal and calculate the envelope of the sound wave. The length of the sliding window is adjusted according to the frequency, and the maximum window length is set to 30 times the lowest frequency period of the channel. Figure 2 As shown, in the noise acquisition mode, after the sampled data enters the channel, it is first downsampled by taking points at intervals according to the new sampling frequency of each channel. If the sampling rate is reduced to 1 / 10 of the original sampling rate, every 10 data points are extracted, and the 0th, 10th, 20th... data points of the original sampled data are taken to form new data for subsequent operations. After downsampling, the data is first taken for absolute value, and then the data is smoothed using a sliding window. Then, the determination of the acoustic envelope after downsampling in the above step 103 can be replaced by the following steps 301 to 305:
[0032] Step 301: taking the absolute value of the downsampled sound wave amplitude to obtain a positive sound wave.
[0033] Step 302: Set the sliding window duration of each channel to three times the minimum frequency of the channel receiving frequency band.
[0034] Step 303: According to the sliding window duration of each channel, a sliding window technique is used to perform window sliding interception on the positive sound waves.
[0035] Step 304: Calculate the average amplitude of the positive sound waves in each sliding intercepted window.
[0036] Step 305: sequentially connect the amplitude means of all windows to obtain the downsampled sound wave envelope.
[0037] Table 1 shows the parameters of each channel. According to the sampling data frequency band, there are five channels in total. The first three channels are downsampled by 1000 times, 100 times and 10 times respectively, and the sampling rates become 1k, 10k and 100k, which are 10 times the upper limit of the frequency of this channel respectively. The sampling rate of the last two channels remains unchanged at 1MHz. The sliding window length is set to 3 times the minimum frequency of the channel, that is, the maximum period. This window length is equivalent to 30 times the period of the maximum frequency of the channel.
[0038] Table 1 Channel parameters
[0039] Channel number Frequency lower limit / Hz Upper frequency limit / Hz Channel sampling rate / Hz Sliding window duration / s Window length points 1 10 100 1k 0.3 300 2 100 1k 10k 0.03 300 3 1k 10k 100k 0.003 300 4 10k 100k 1M 0.0003 300 5 100k —— 1M 0.00003 30
[0040] In another exemplary embodiment of the present application, the calculation formula for the amplitude mean of the positive sound waves in each sliding intercepted window is:
[0041]
[0042] Wherein, env(n) is the absolute mean of the amplitudes from sample point n-N+1 to sample point n, N is the duration of the sliding window, m is the index of the sampling point, x(m) is the amplitude of the sound wave at sampling point m, and w(nm) is the sliding window.
[0043] In the first formula, env(n) is the envelope value at sample point n, that is, the envelope at the nth point is calculated. It also represents the current position of the sliding window, and n is equivalent to the last point of the sliding window. m is the index of the sampling point, that is, the mth data collected. m and n are used to describe the position of the original signal and the position after the envelope calculation, respectively. In the second formula, as n changes, the window moves, and the original signal sample points in the window also change, so the mean of different signal sample points can be calculated.
[0044] When the sliding window length is N and the step length is k, the first sliding operation calculates the mean of the N data from point x(n-N+1) to point x(n) at sample point n, the second sliding operation calculates the mean of point x(n-N+1+k) to point x(n+k), and so on. Figure 3 As shown, the sliding window length is 5 and the step size is 3. The first sliding calculates the mean of x(1) to x(5) at sample point 5, the second sliding calculates the mean of x(4) to x(8), and the third sliding calculates the mean of x(7) to x(11). This mean is the signal envelope.
[0045] In another exemplary embodiment of the present application, the preset pulse width threshold of each channel is a multiple of the sliding window time length of the respective channel.
[0046] In another exemplary embodiment of the present application, 5 seconds of data are collected as noise when calculating the amplitude threshold, and then the smoothed noise is averaged. The amplitude threshold is set according to the distribution characteristics of the noise. After the noise is collected, the amplitude threshold is determined by multiplying it by an adjustable parameter. If the signal envelope exceeds this threshold and continues to exceed the set pulse width threshold, it is determined to be a whale call signal. Then the above step 104 can be specifically as follows: multiply the mean of the sound wave envelope after downsampling of each channel by the adjustable parameter to obtain the amplitude threshold of each channel.
[0047] Exemplarily, the pulse width threshold may be set to an integer multiple or a fraction of the sliding window duration N.
[0048] In another exemplary embodiment of the present application, the sound waves to be detected received by each channel in the above step 105 may be in two situations: sound waves with only noise, or sound wave signals of cetacean calls mixed with sound waves with noise.
[0049] In another exemplary embodiment of the present application, referring to Figure 2It can be seen that each channel in the detection mode and noise acquisition mode has undergone downsampling, absolute value taking and smoothing operations. Therefore, the downsampling, absolute value taking and smoothing operations in the detection mode can be referred to the noise acquisition mode and will not be repeated here. Different from the noise acquisition mode: the detection mode further combines the pulse width and frequency characteristics for judgment. By setting the pulse width threshold, the number of envelope sample points that continuously exceed the amplitude threshold is detected. If the number of sample points exceeds the set pulse width threshold, it is regarded as a valid signal. At the same time, considering the frequency distribution of cetacean calls, fine classification can be performed according to the frequency response of different channels to improve the accuracy of signal recognition. Among them, the envelope sample points can be consistent with the previous sample points, or they can be customized.
[0050] In another exemplary embodiment of the present application, in a multi-channel system, when multiple channels detect signals at the same time, the present application adopts a decision strategy to determine the final signal output. Specifically, if multiple channels detect signals at the same time, data fusion is performed according to the overlap and intensity of the signals, and a comprehensive detection result is finally output. This multi-channel data fusion method can effectively improve the reliability of signal detection and reduce false positives and false negatives. Then the above step 107 can be replaced by the following steps 401 to 402:
[0051] Step 401: Perform frequency domain analysis on the downsampled sound wave signals of each target channel to determine the peak frequency of each target channel.
[0052] Step 402: Select the maximum and minimum values of the peak frequencies of all target channels, and use the frequency band formed by the maximum and minimum values of the peak frequencies as the frequency range of the cetacean calls.
[0053] This application collects and processes signals in different frequency bands to flexibly adjust parameters according to different frequencies to adapt to different characteristics of signals. Compared with traditional recognition methods for specific sound features, this application only downsamples and averages signals in the time domain, and only performs frequency domain analysis during data fusion and channel judgment. Compared with Fourier transform processing in the frequency domain, the amount of calculation is lower, and multiple sounds can be detected at the same time, which is more conducive to the assessment of marine resource diversity.
[0054] The method of the present application can be used for acoustic monitoring of marine mammals such as cetaceans and dolphins, and is applicable to the fields of marine ecological environment protection, marine mammal abundance surveys, etc.
[0055] Based on the same inventive concept, the embodiment of the present application also provides a passive sonar system that applies the multi-channel cetacean call detection method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations of one or more passive sonar system embodiments provided below can refer to the limitations of the multi-channel cetacean call detection method above, and will not be repeated here.
[0056] In an exemplary embodiment, a passive sonar system is provided, wherein the passive sonar system adopts the above-mentioned multi-channel cetacean call detection method.
[0057] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0058] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A multi-channel cetacean call detection method, characterized in that: include: A plurality of channels are provided for receiving acoustic wave signals of cetacean calls of different frequency bands; Set the noise collection mode and detection mode to be performed sequentially in each cycle; When executing the noise collection mode, down-sample the sound waves of the receiving frequency band of each channel, and determine the envelope of the down-sampled sound waves; Determine the amplitude threshold of each channel according to the sound wave envelope after downsampling of each channel; After the detection mode is started, for each channel that receives the sound wave to be detected, downsampling the sound wave to be detected in the frequency band received by the channel is performed, and the envelope of the sound wave to be detected after the downsampling is determined; Identify the number of sound wave envelope sample points to be detected that continuously exceed the amplitude threshold of each channel within the cycle period; if the identified number of sound wave envelope sample points to be detected is greater than or equal to the preset pulse width threshold, determine that the sound wave to be detected in the frequency band received by the channel is a sound wave signal of a cetacean call, and define the channel as a target channel; When there is only one target channel, the frequency band of the downsampled sound wave signal of the target channel is used as the frequency range of the cetacean call; When there are multiple target channels, the frequency bands of the downsampled sound waves of the multiple target channels are fused to determine the frequency range of the cetacean calls.
2. The multi-channel cetacean call detection method according to claim 1, characterized in that: Multiple channels are set up for receiving acoustic signals of cetacean calls in different frequency bands, specifically including: The first channel is set to receive the sound wave signal of the whale call sound of the first frequency band, the second channel is set to receive the sound wave signal of the whale call sound of the second frequency band, the third channel is set to receive the sound wave signal of the whale call sound of the third frequency band, and the fourth channel is set to receive the sound wave signal of the whale call sound of the fourth frequency band; the frequency range of the first frequency band is greater than or equal to 10 Hz and less than 100 Hz, the frequency range of the second frequency band is greater than or equal to 100 Hz and less than 1 kHz, the frequency range of the third frequency band is greater than or equal to 1 kHz and less than 10 kHz, and the frequency range of the fourth frequency band is greater than or equal to 10 kHz and less than 100 kHz.
3. The multi-channel cetacean call detection method according to claim 2, characterized in that: Downsample the sound waves in the receiving frequency band of each channel, including: Determine that the sampling rate of the first channel is 1k, the sampling rate of the second channel is 10k, the sampling rate of the third channel is 100k, and the sampling rate of the fourth channel is 1MHz; According to the sampling rate of each channel, the sound waves in the receiving frequency band of each channel are downsampled.
4. The multi-channel cetacean call detection method according to claim 1, characterized in that: Determine the sound wave envelope after downsampling, including: Take the absolute value of the amplitude of the downsampled sound wave to obtain a positive sound wave; Set the sliding window duration of each channel to 3 times the minimum frequency of the channel receiving frequency band; According to the sliding window duration of each channel, the sliding window technology is used to perform window sliding interception on the positive sound waves; Calculate the mean amplitude of the positive sound waves in each sliding intercepted window; The amplitude means of all windows are connected in sequence to obtain the downsampled sound wave envelope.
5. The multi-channel cetacean call detection method according to claim 4, characterized in that: The calculation formula for the mean amplitude of the positive sound wave in each sliding intercepted window is: Wherein, env(n) is the absolute mean of the amplitudes from sample point n-N+1 to sample point n, N is the duration of the sliding window, m is the index of the sampling point, x(m) is the amplitude of the sound wave at sampling point m, and w(nm) is the sliding window.
6. The multi-channel cetacean call detection method according to claim 4, characterized in that: The preset pulse width threshold of each channel is a multiple of the sliding window duration of the respective channel.
7. The multi-channel cetacean call detection method according to claim 1, characterized in that: According to the sound wave envelope after downsampling of each channel, the amplitude threshold of each channel is determined, including: The amplitude threshold of each channel is obtained by multiplying the mean value of the downsampled sound wave envelope of each channel by the adjustable parameter.
8. The multi-channel cetacean call detection method according to claim 1, characterized in that: When there are multiple target channels, the frequency bands of the downsampled sound waves of the multiple target channels are fused to determine the frequency range of the cetacean call, including: Perform frequency domain analysis on the downsampled acoustic wave signals of each target channel to determine the peak frequency of each target channel; The maximum and minimum values of the peak frequencies of all target channels are selected, and the frequency band formed by the maximum and minimum values of the peak frequencies is used as the frequency range of the cetacean calls.
9. The multi-channel cetacean call detection method according to claim 1, characterized in that: In each cycle, the duration of receiving sound waves in the noise collection mode is 5 seconds, and the duration of receiving sound waves in the detection mode is 30 minutes.
10. A passive sonar system, characterized in that: The passive sonar system adopts the multi-channel cetacean call detection method according to any one of claims 1 to 9.