CN112885362A - Target identification method, system, device and medium based on radiation noise - Google Patents
Target identification method, system, device and medium based on radiation noise Download PDFInfo
- Publication number
- CN112885362A CN112885362A CN202110049764.XA CN202110049764A CN112885362A CN 112885362 A CN112885362 A CN 112885362A CN 202110049764 A CN202110049764 A CN 202110049764A CN 112885362 A CN112885362 A CN 112885362A
- Authority
- CN
- China
- Prior art keywords
- signal
- frequency spectrum
- framing
- radiation noise
- resonance component
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000005855 radiation Effects 0.000 title claims abstract description 37
- 238000001228 spectrum Methods 0.000 claims abstract description 61
- 238000009432 framing Methods 0.000 claims abstract description 42
- 230000005236 sound signal Effects 0.000 claims abstract description 27
- 238000003062 neural network model Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000012545 processing Methods 0.000 claims abstract description 12
- 238000003860 storage Methods 0.000 claims abstract description 9
- 230000006870 function Effects 0.000 claims description 22
- 238000001914 filtration Methods 0.000 claims description 19
- 230000000877 morphologic effect Effects 0.000 claims description 13
- 238000000354 decomposition reaction Methods 0.000 claims description 9
- 230000009466 transformation Effects 0.000 claims description 8
- 230000037433 frameshift Effects 0.000 claims description 6
- 230000003595 spectral effect Effects 0.000 claims description 2
- 230000001131 transforming effect Effects 0.000 claims description 2
- 230000010355 oscillation Effects 0.000 abstract description 7
- 230000001052 transient effect Effects 0.000 abstract description 5
- 230000002349 favourable effect Effects 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 12
- 230000008901 benefit Effects 0.000 description 7
- 238000011161 development Methods 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 239000010410 layer Substances 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 241000251468 Actinopterygii Species 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000007123 defense Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 241000251729 Elasmobranchii Species 0.000 description 1
- 241000282414 Homo sapiens Species 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 239000011229 interlayer Substances 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 239000013535 sea water Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L17/00—Speaker identification or verification
- G10L17/26—Recognition of special voice characteristics, e.g. for use in lie detectors; Recognition of animal voices
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L17/00—Speaker identification or verification
- G10L17/18—Artificial neural networks; Connectionist approaches
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
Abstract
The invention provides a target identification method, a system, a device and a storage medium based on radiation noise, wherein the method comprises the following steps: acquiring an audio signal of first radiation noise; decomposing the audio signal to obtain a high-resonance component signal; performing framing and windowing processing on the high-resonance component signal to obtain a framing signal; generating a frequency spectrum according to the framing signal, and arranging the frequency spectrum in a time domain to obtain frequency spectrum samples; and training according to the frequency spectrum sample to obtain a neural network model, acquiring second radiation noise, and identifying and classifying according to the neural network model. The method filters transient non-oscillation signals and Gaussian white noise in the radiation noise to a certain extent, is more favorable for extracting the characteristics of the signals, solves the problem that a sample is seriously interfered by the marine environment, can realize intelligent and accurate identification of the radiation noise, and can be widely applied to the technical field of acoustic identification.
Description
Technical Field
The invention relates to the technical field of acoustic recognition, in particular to a target recognition method, a target recognition system and a storage medium based on radiation noise.
Background
With the continuous development of science and technology, the understanding of the sea by human beings is deepened gradually. On one hand, the development of marine economy becomes an important engine for national economy development, a wide and endless sea contains abundant mineral resources, marine biological resources, seawater chemical resources and the like, and the advantages of marine resources are converted into economic advantages by improving the capability of exploring and developing the marine resources, so that the development speed of the economy is accelerated, and the marine economy development method has a vital significance. On the other hand, the wide ocean area provides great ocean strategic depth in military affairs, and great tests are set for early warning defense of ocean territory and ocean resource protection. In economic activities, submarine exploration, oil platform monitoring, economic fish school detection and the like all need to identify and detect underwater targets, whether underwater acoustic signals are applied to identify and classify the underwater targets in ocean defense or not can be accurately judged whether the underwater targets belong to common aquatic fishes or unknown submarines or even torpedoes, and the method is very key to the survival and the battle of naval vessels. Thus, the need to defend mastership independently and territorial integrity, both economically for marine development purposes and military and political, has prompted technologists to continually improve the accuracy and efficiency of identification of underwater targets.
However, in the prior art, manual identification and sorting are still performed in most cases, the work intensity is high, the efficiency is low, and meanwhile, identification errors are generated with high probability in the identification process.
Disclosure of Invention
In view of the above, to at least partially solve one of the above technical problems, an embodiment of the present invention aims to provide an efficient, convenient and highly accurate target identification method based on radiation noise, and also provides a corresponding system, an apparatus and a computer-readable storage medium for implementing the method.
In a first aspect, the present invention provides a method for identifying a target based on radiation noise, comprising the steps of:
acquiring an audio signal of first radiation noise;
decomposing the audio signal to obtain a high-resonance component signal;
performing framing and windowing processing on the high-resonance component signal to obtain a framing signal;
generating a frequency spectrum according to the framing signal, and arranging the frequency spectrum in a time domain to obtain frequency spectrum samples;
and training according to the frequency spectrum sample to obtain a neural network model, acquiring second radiation noise, and identifying and classifying according to the neural network model.
In a possible embodiment of the present disclosure, the step of decomposing the audio signal to obtain a high-resonance component signal includes:
performing feature enhancement on the audio signal to obtain an enhanced signal;
and carrying out wavelet transformation on the enhanced signal, and analyzing morphological components of the wavelet-transformed signal to obtain the high-resonance component signal.
In a possible embodiment of the present disclosure, the step of performing frame windowing on the high-resonance component signal to obtain a frame signal includes:
determining frame shift, and intercepting the high-resonance component signal according to the frame shift to obtain a high-resonance component sub-signal;
and generating the framing signal according to the high-resonance component quantum signal and a window function, and removing singular sample data in the framing signal.
In a possible embodiment of the present disclosure, the generating a spectrum from the framing signal, and arranging spectrum samples in a time domain according to the spectrum includes:
carrying out Fourier transform on the framing signals to obtain frequency spectrum signals;
and generating a logarithmic magnitude spectrum according to the frequency spectrum signal and arranging the logarithmic magnitude spectrum on a time domain to obtain a frequency spectrum sample.
In a possible embodiment of the solution of the present application, the wavelet transform is a Q wavelet transform, and the step of wavelet transforming the enhancement signal includes:
and determining sparsity representation of the complex signal according to the wavelet basis function and the Q factor of the Q wavelet transform, and reconstructing the enhanced signal to obtain a reconstructed signal.
In a possible embodiment of the present disclosure, the determining sparsity representation of a complex signal according to wavelet basis functions and Q factors of the Q wavelet transform, and reconstructing the enhanced signal includes:
constructing a high-pass filter and a low-pass filter, and constructing a plurality of analysis filter groups according to the high-pass filter and the low-pass filter;
filtering the enhanced signal through the analysis filter bank to obtain a high-pass filtering signal and a low-pass filtering signal;
and iteratively inputting the low-pass filtering signal into the analysis filter bank to obtain a secondary low-pass filtering signal, and constructing the reconstruction signal according to the high-pass filtering signal and the secondary low-pass filtering signal.
In a possible embodiment of the present disclosure, the step of obtaining the high-resonance component signal from the wavelet-transformed signal by morphological component analysis includes:
obtaining a wavelet coefficient of the high resonance component of the reconstruction signal through morphological component analysis;
and extracting the high-resonance component signal from the reconstructed signal according to the wavelet coefficient.
In a second aspect, the present invention further provides a target identification system based on radiation noise, including: a signal acquisition unit configured to acquire an audio signal of the first radiation noise;
the signal decomposition unit is used for decomposing the audio signal to obtain a high-resonance component signal;
the framing processing unit is used for performing framing windowing processing on the high-resonance component signal to obtain a framing signal;
the time-frequency transformation unit is used for generating a frequency spectrum according to the framing signals and obtaining frequency spectrum samples according to the frequency spectrum in a time domain;
the model training unit is used for obtaining a neural network model according to the frequency spectrum sample training;
and the noise identification unit is used for acquiring the second radiation noise and carrying out identification and classification according to the neural network model.
In a third aspect, an embodiment of the present invention further provides a target identification apparatus based on radiation noise, including:
at least one processor;
at least one memory for storing at least one program;
when the at least one program is executed by the at least one processor, the at least one processor is caused to execute the method for radiated noise based object recognition in the first aspect.
In a fourth aspect, the present invention also provides a storage medium, in which a processor-executable program is stored, and the processor-executable program is used for executing the method in the first aspect when being executed by a processor.
Advantages and benefits of the present invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention:
the method is based on the method, transient non-oscillation signals and Gaussian white noise in the radiation noise are filtered to a certain extent by decomposing the audio signals of the radiation noise, so that the signal is more favorably subjected to characteristic extraction, and the problem that a sample is seriously interfered by the marine environment is solved; in addition, windowing and framing are carried out on the signals, the signals are converted into frequency spectrum signals to be trained to obtain a neural network model, characteristic signals are identified in a targeted mode, and intelligent and accurate radiation noise identification is achieved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flowchart illustrating steps of an embodiment of a method for identifying a target based on radiated noise according to the present invention;
FIG. 2 is a schematic diagram of the wavelet transform process of the present invention;
FIG. 3 is a schematic diagram of the structure of a convolutional neural network model in the present invention;
fig. 4 is a schematic structural diagram of a target recognition system based on radiation noise according to the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention. The step numbers in the following embodiments are provided only for convenience of illustration, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.
In a first aspect, as shown in fig. 1, the present application provides a method for target identification based on radiated noise, comprising steps S01-S05:
and S01, acquiring an audio signal of the first radiation noise.
Specifically, the embodiment acquires a certain number of samples of the radiated noise, and divides the audio signal in the samples into an offline label data set and an online test data set, which are respectively used as the offline training data samples of the subsequent neural network model and the online training data samples of the model.
And S02, decomposing the audio signal to obtain a high-resonance component signal.
Specifically, the offline tag data set in step S01 is selected, and the embodiment uses a Resonance-based Sparse Signal Decomposition algorithm (RSSD) to preprocess the Signal, and sequentially reads the obtained high-Resonance component signals. The sparse signal decomposition process based on resonance filters transient non-oscillation signals and Gaussian white noise in radiation noise to a certain extent, is more favorable for carrying out feature extraction on signals, and high resonance components output by decomposition are respectively used as the input of a subsequent cyclic neural network and a line spectrum enhancement process.
In some optional embodiments, the step S02 of decomposing the audio signal to obtain the high-resonance component signal may be further subdivided into steps S021-S022:
s021, performing feature enhancement on the audio signal to obtain an enhanced signal;
and S022, performing wavelet transformation on the enhanced signal, and performing morphological component analysis on the wavelet-transformed signal to obtain a high-resonance-component signal.
Specifically, in the embodiment, feature enhancement is performed on the audio signal in the offline tag data set, and the feature enhancement is mainly performed on the audio signal, and includes cleaning and enhancing data, for example, identifying missing parts in the audio signal, and deleting a data record if a certain record missing part exceeds 50% of the data record. As another example, the audio signal in the offline tag dataset is normalized or normalized.
The wavelet transform employed in the embodiment is a Q wavelet transform, i.e., a Q-factor wavelet transform. And in the embodiment, the step S022 of performing wavelet transformation on the enhanced signal, and obtaining a high-resonance component signal from the wavelet-transformed signal through morphological component analysis determines sparsity representation of a complex signal according to a wavelet basis function and a Q factor of Q wavelet transformation, and reconstructs the enhanced signal to obtain a reconstructed signal.
Specifically, in the embodiment, the RSSD algorithm uses two different wavelet basis functions, corresponding to Q factors of different sizes, obtains sparsity representation of a complex signal, and reconstructs the signal, as shown in fig. 2, the RSSD algorithm regards the resonance as a basis for sparse signal decomposition, and the Q factor is quantization of the degree of the resonance of the signal. The high-resonance signal shows higher frequency aggregation in the time domain, more oscillation waveforms exist at the same time, and the oscillation waveforms correspond to larger Q factors; a low resonant signal appears as a non-oscillating, indeterminate transient signal, corresponding to a smaller Q factor.
More specifically, in some possible embodiments, the process of reconstructing the enhanced signal in step S022 may be further subdivided into more detailed steps S022a-S022 c:
s022a, constructing a high-pass filter and a low-pass filter, and constructing a plurality of analysis filter banks according to the high-pass filter and the low-pass filter;
s022b, filtering the enhanced signal through an analysis filter bank to obtain a high-pass filter signal and a low-pass filter signal;
s022c, iteratively inputting the low-pass filtering signal to an analysis filter bank to obtain a secondary low-pass filtering signal, and constructing a reconstruction signal according to the high-pass filtering signal and the secondary low-pass filtering signal.
In particular, in the embodiment, the signal reconstruction, i.e. the process of obtaining the reconstructed signal, is performed using a TQWT toolbox. Such as where the process is primarily performed by a classification and synthesis filter bank. The analysis filter bank of each layer is composed of a high-pass filter Hhigh(w) and a low-pass filter Hlow(w) and corresponding scaling process, the definition in the embodiment is:
wherein the content of the first and second substances,
is a Daubechies filter with second-order vanishing moment, alpha and beta are scaling factors after the signal passes through a low-pass filter and a high-pass filter respectively, and 0<α<1，0<β<1, and the scaling process of the low-pass filter and the high-pass filter is defined as:
Y(w)＝X(αw),|w|≤π
wherein the Q factor quantifies the degree of signal resonance, and is defined as follows:
Q＝fcBW
wherein f iscRepresenting the center frequency of the signal and BW the bandwidth. If the sampling frequency of the original input signal is fsThen the center frequency fcAbout fsAnd the filter bank levels j and α, β are represented as:
the bandwidth BW is expressed as:
BW＝0.5βαj-1π
the Q factor can thus be derived as:
after the original signal has passed through the filter bank, the output of the low-pass channel is iteratively input into a filter bank at a deeper level up to a preset level j. At the same time, the over-sampling rate r is selected to construct the wavelet basis function phih、ΦlDeepest level JmaxAnd the over-sampling ratio r is defined with respect to α, β as follows:
therefore, in the TQWT algorithm, the three parameters Q, r, and J can be calculated by selecting α and β, and the selection of α and β is determined by the signal inherent oscillation characteristic. For the input signal, Q, r, J are set to extract its high and low resonance information, respectively.
In some possible embodiments, in the step S022 of obtaining a high resonance component signal from the wavelet-transformed signal through morphological component analysis, a process of obtaining a high resonance component signal from the wavelet-transformed signal through morphological component analysis may be performed, which further includes more detailed steps S022d-S022 e:
s022d, obtaining a wavelet coefficient of a high resonance component of a reconstructed signal through morphological component analysis;
and S022e, extracting a high-resonance component signal from the reconstructed signal according to the wavelet coefficient.
In particular, Morphological Component Analysis (MCA) is used in the examples to decompose signals with different morphological characteristics. The radiated noise is separated and extracted using the MCA algorithm to construct an optimal sparse representation of its high and low resonance components. Considering a discrete radiated noise sequence, the signal can be sparsely represented as:
x＝Φhwh+Φlwl+n
wherein, wh，wlRespectively high resonant component xhAnd a low resonant component xlCorresponding wavelet coefficient, phihAnd philAre each xh，xlThe corresponding wavelet basis function, n, represents the residual components of the signal both before removal. The aim of the MCA in the example is to obtain an optimal representation of the high and low resonance components of the signal, i.e. the resulting optimal whAnd wl. The problem can be solved by minimizing the following objective function:
wherein, JhAnd JlDenotes xhAnd xlThe number of decomposition layers of (a) is,
and
wavelet coefficient, λ, representing high and low resonance components of layer jh,jAnd λl,jAre respectively measures
And
specific value of (1) and phih、jAnd phil、jThe energy of (c) is related to:
λl,j＝kl,j||Φl、j||2,j＝1,2,3,…,Jl+1
λh,j＝kh,j||Φh、j||2,j＝1,2,3,…,Jh+1
kl,jand kh,jIs the proportionality coefficient of the energy distribution of the high resonance component and the low resonance component, and k is selected in the embodimentl,j＝kh,j0.5 to balance the energy distribution of the two components.
The augmented Lagrangian contraction algorithm through decomposition can be applied to J (w)h,wl) To obtain the optimal wavelet coefficient by solving the optimization problem
And
thus, the optimal representation for the high and low resonance components by the MCA algorithm is:
and S03, performing framing and windowing processing on the high-resonance component signal to obtain a framing signal.
Specifically, windowing and framing a high-resonance component signal, and selecting a Hanning window (Hanning) speech signal that is macroscopically unstable, microscopically stable, and has short-time stationarity (the speech signal can be considered approximately constant within 10-30 ms), allows the speech signal to be processed in short segments, each of which is called a frame (chunk). Subsequent operations require windowing, and thus, in framing, they are not truncated back-to-back, but rather overlap one another by a fraction. Windowing is to multiply a window function, and Fourier expansion is performed after windowing, so that the whole situation is more continuous, and the Gibbs effect is avoided. During the windowing process, the speech signal which is not periodic originally presents partial characteristics of the periodic function. The cost of windowing is that both end portions of a frame signal are weakened, so there needs to be overlap between frames at the time of framing.
In some alternative embodiments, the step S03 of performing frame windowing on the high-resonant component signal to obtain a frame signal may be further subdivided into steps S031 and S032:
s031, confirm the frame moves, according to the frame moves and intercepts the high resonance component signal and gets the high resonance component sub-signal;
s032, generating a framing signal according to the high-resonance component quantum signal and the window function, and removing singular sample data in the framing signal.
Specifically, the time difference between the start positions of two adjacent frames is called frame Shift (STRIDE). In one embodiment, the window size is 2048 (i.e., the number of fft points is 2048), and the frame shift is determined when the inter-frame overlap is 75%. The signals of each frame need to be normalized and de-centered, the power of the signals is uniform in time, the mean value of the samples is 0, namely the data is limited in a certain range, the adverse effect caused by singular sample data can be eliminated, meanwhile, neuron saturation can be avoided, and the convergence speed of the network is accelerated.
And S04, generating a frequency spectrum according to the framing signal, and arranging the frequency spectrum in a time domain to obtain a frequency spectrum sample.
Specifically, the time-domain framed signal is converted into the frequency-domain framed signal by fourier transform.
In some alternative embodiments, the step S04 of generating a spectrum from the framing signal and obtaining spectrum samples from the spectrum arranged in the time domain may be subdivided into more specific steps S041-S042:
s041, carrying out Fourier transform on the subframe signals to obtain frequency spectrum signals;
and S042, generating a logarithmic magnitude spectrum according to the frequency spectrum signals and arranging the logarithmic magnitude spectrum on a time domain to obtain a frequency spectrum sample.
Specifically, fourier transform is performed on each frame of signals obtained in step S03, a log-amplitude spectrum is taken from the transformed spectrum and arranged in a time domain, and 64 points on a time axis are taken as a sample, so that a LOFAR spectrum sample with a size of 1024 × 64 is obtained. The audio was sampled at 52734Hz with a duration of approximately 0.62s per sample. The number of training sets and test sets of each type of sample is collated and drawn into a table, as shown in table 1:
TABLE 1
And the ID in the table is the label of the audio in the database, and the ship type corresponding to each section of audio can be obtained according to the ID and used as a label for supervised learning of the deep neural network. And (4) performing LOFAR spectral line spectrum enhancement processing on the obtained sample based on multi-step judgment to obtain a LOFAR spectrum with enhanced line spectrum characteristics. The LOFAR spectrum is a two-dimensional matrix, can be regarded as a single-channel picture, and can be used as an input layer of a convolutional neural network
And S05, training according to the frequency spectrum sample to obtain a neural network model, obtaining second radiation noise, and identifying and classifying according to the neural network model.
Specifically, as shown in fig. 3, since the convolutional neural network is a static network belonging to a spatial structure hierarchy, and has only full connection or partial connection (interlayer connection) but no intralayer connection, it is very suitable for the computer vision field with spatial local correlation and sample independence, but for a sequence signal with temporal correlation as a main feature, data of a fixed size can only be input into the network in the form of framing, the utilization of the temporal front and back correlation of the training samples is limited, and the selection of the size of the framing also brings contradiction between the recognition accuracy and the training duration of the network. The LSTM structure is a repeating modular unit, but has memory and is more complex. The specific structure ensures the capability of the LSTM to retain the information of the earlier time step under the condition of long time step, and solves the problems of gradient explosion and gradient disappearance. However, since CNN can fully utilize the advantage of data spatial correlation, its recognition result will be better than LSTM. The embodiment thus employs the CNN model as the primary training model.
In a second aspect, a software system embodiment of the present invention, a radiated noise based target identification system, includes:
a signal acquisition unit configured to acquire an audio signal of the first radiation noise;
the signal decomposition unit is used for decomposing the audio signal to obtain a high-resonance component signal;
the framing processing unit is used for performing framing windowing processing on the high-resonance component signal to obtain a framing signal;
the time-frequency transformation unit is used for generating a frequency spectrum according to the framing signals and obtaining frequency spectrum samples according to the frequency spectrum in a time domain;
the model training unit is used for obtaining a neural network model according to the frequency spectrum sample training;
and the noise identification unit is used for acquiring the second radiation noise and carrying out identification and classification according to the neural network model.
In a third aspect, as shown in fig. 4, an embodiment of the present invention further provides a target identification apparatus based on radiated noise, which includes at least one processor; at least one memory for storing at least one program; when the at least one program is executed by the at least one processor, the at least one processor is caused to execute the method for radiated noise based object recognition as in the first aspect.
For example, in one possible embodiment, the Ubuntu 16.04 operating system, GTX 1080ti video card; the programming languages are mainly python 3.6 and Matlab R2016 b. The deep learning library and software toolbox are as follows: kerass 2.3 (tensoflow back), library for Librosa audio processing (python) and TQWT toolbox (Matlab language).
An embodiment of the present invention further provides a storage medium storing a program, where the program is executed by a processor to implement the method in the first aspect.
From the above specific implementation process, it can be concluded that the technical solution provided by the present invention has the following advantages or advantages compared to the prior art:
the method of the application decomposes the audio signal of the radiation noise, filters transient non-oscillation signals and Gaussian white noise in the radiation noise to a certain extent, is more beneficial to extracting the characteristics of the signals, and solves the problem that the sample is seriously interfered by the marine environment; in addition, windowing and framing are carried out on the signals, the signals are converted into frequency spectrum signals to be trained to obtain a neural network model, characteristic signals are identified in a targeted mode, and intelligent and accurate radiation noise identification is achieved.
In alternative embodiments, the functions/acts noted in the block diagrams may occur out of the order noted in the operational illustrations. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality/acts involved. Furthermore, the embodiments presented and described in the flow charts of the present invention are provided by way of example in order to provide a more thorough understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and in which sub-operations described as part of larger operations are performed independently.
Furthermore, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and/or features may be integrated in a single physical device and/or software module, or one or more of the functions and/or features may be implemented in a separate physical device or software module. It will also be appreciated that a detailed discussion of the actual implementation of each module is not necessary for an understanding of the present invention. Rather, the actual implementation of the various functional modules in the apparatus disclosed herein will be understood within the ordinary skill of an engineer, given the nature, function, and internal relationship of the modules. Accordingly, those skilled in the art can, using ordinary skill, practice the invention as set forth in the claims without undue experimentation. It is also to be understood that the specific concepts disclosed are merely illustrative of and not intended to limit the scope of the invention, which is defined by the appended claims and their full scope of equivalents.
Wherein the functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The logic and/or steps represented in the flowcharts or otherwise described herein, e.g., an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
It should be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
While embodiments of the invention have been shown and described, it will be understood by those of ordinary skill in the art that: various changes, modifications, substitutions and alterations can be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. The target identification method based on the radiation noise is characterized by comprising the following steps:
acquiring an audio signal of first radiation noise;
decomposing the audio signal to obtain a high-resonance component signal;
performing framing and windowing processing on the high-resonance component signal to obtain a framing signal;
generating a frequency spectrum according to the framing signal, and arranging the frequency spectrum in a time domain to obtain frequency spectrum samples;
and training according to the frequency spectrum sample to obtain a neural network model, acquiring second radiation noise, and identifying and classifying according to the neural network model.
2. The method for identifying an object based on radiated noise according to claim 1, wherein the step of decomposing the audio signal to obtain a high-resonance component signal comprises:
performing feature enhancement on the audio signal to obtain an enhanced signal;
and performing wavelet transformation on the enhanced signal, and performing morphological component analysis on the wavelet-transformed signal to obtain the high-resonance component signal.
3. The method for identifying an object based on radiation noise according to claim 1, wherein the step of performing frame windowing on the high-resonance component signal to obtain a frame signal comprises:
determining frame shift, and intercepting the high-resonance component signal according to the frame shift to obtain a high-resonance component sub-signal;
and generating the framing signal according to the high-resonance component quantum signal and a window function, and removing singular sample data in the framing signal.
4. The method of claim 1, wherein the step of generating a spectrum from the framing signal and obtaining spectral samples arranged in a time domain according to the spectrum comprises:
carrying out Fourier transform on the framing signals to obtain frequency spectrum signals;
and generating a logarithmic magnitude spectrum according to the frequency spectrum signal and arranging the logarithmic magnitude spectrum on a time domain to obtain a frequency spectrum sample.
5. The method for identifying an object based on radiation noise according to claim 2, wherein the wavelet transform is a Q wavelet transform, and the step of wavelet transforming the enhancement signal comprises:
and determining sparsity representation of the complex signal according to the wavelet basis function and the Q factor of the Q wavelet transform, and reconstructing the enhanced signal to obtain a reconstructed signal.
6. The method for identifying an object based on radiation noise according to claim 5, wherein the step of determining sparsity representation of complex signal according to wavelet basis function and Q factor of Q wavelet transform, reconstructing the enhanced signal comprises:
constructing a high-pass filter and a low-pass filter, and constructing a plurality of analysis filter groups according to the high-pass filter and the low-pass filter;
filtering the enhanced signal through the analysis filter bank to obtain a high-pass filtering signal and a low-pass filtering signal;
and iteratively inputting the low-pass filtering signal into the analysis filter bank to obtain a secondary low-pass filtering signal, and constructing the reconstruction signal according to the high-pass filtering signal and the secondary low-pass filtering signal.
7. The method for identifying an object based on radiation noise according to claim 5, wherein the step of performing morphological component analysis on the wavelet-transformed signal to obtain the high-resonance-component signal comprises:
obtaining a wavelet coefficient of the high resonance component of the reconstruction signal through morphological component analysis;
and extracting the high-resonance component signal from the reconstructed signal according to the wavelet coefficient.
8. A radiated noise based target identification system, comprising:
a signal acquisition unit configured to acquire an audio signal of the first radiation noise;
the signal decomposition unit is used for decomposing the audio signal to obtain a high-resonance component signal;
the framing processing unit is used for performing framing windowing processing on the high-resonance component signal to obtain a framing signal;
the time-frequency transformation unit is used for generating a frequency spectrum according to the framing signals and obtaining frequency spectrum samples according to the frequency spectrum in a time domain;
the model training unit is used for obtaining a neural network model according to the frequency spectrum sample training;
and the noise identification unit is used for acquiring the second radiation noise and carrying out identification and classification according to the neural network model.
9. An apparatus for identifying a target based on radiated noise, comprising:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, cause the at least one processor to perform the method of radiated noise based object identification according to any one of claims 1 to 7.
10. A storage medium having stored therein a program executable by a processor, characterized in that: the processor executable program, when executed by a processor, is for executing the method of radiated noise based object recognition according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110049764.XA CN112885362B (en) | 2021-01-14 | 2021-01-14 | Target identification method, system, device and medium based on radiation noise |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110049764.XA CN112885362B (en) | 2021-01-14 | 2021-01-14 | Target identification method, system, device and medium based on radiation noise |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112885362A true CN112885362A (en) | 2021-06-01 |
| CN112885362B CN112885362B (en) | 2024-04-09 |
Family
ID=76049103
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110049764.XA Active CN112885362B (en) | 2021-01-14 | 2021-01-14 | Target identification method, system, device and medium based on radiation noise |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112885362B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6718316B1 (en) * | 2000-10-04 | 2004-04-06 | The United States Of America As Represented By The Secretary Of The Navy | Neural network noise anomaly recognition system and method |
| CN102176312A (en) * | 2011-01-07 | 2011-09-07 | 蔡镇滨 | System and method for reducing burst noise through wavelet trapped wave |
| CN105488466A (en) * | 2015-11-26 | 2016-04-13 | 中国船舶工业系统工程研究院 | Deep neural network and underwater sound target vocal print feature extraction method |
| CN108694346A (en) * | 2017-04-05 | 2018-10-23 | 中国科学院声学研究所 | A kind of Ship Radiated-Noise signal recognition method based on two-stage CNN |
| CN110189757A (en) * | 2019-06-27 | 2019-08-30 | 电子科技大学 | A kind of giant panda individual discrimination method, equipment and computer readable storage medium |
| CN110245608A (en) * | 2019-06-14 | 2019-09-17 | 西北工业大学 | A kind of Underwater targets recognition based on semi-tensor product neural network |
| CN110246504A (en) * | 2019-05-20 | 2019-09-17 | 平安科技（深圳）有限公司 | Birds sound identification method, device, computer equipment and storage medium |
| CN110556121A (en) * | 2019-09-18 | 2019-12-10 | 腾讯科技（深圳）有限公司 | Frequency band extension method, device, electronic equipment and computer readable storage medium |
| CN110598593A (en) * | 2019-08-29 | 2019-12-20 | 东南大学 | Planetary gearbox fault diagnosis method based on resonance sparse decomposition and FastICA algorithm |
| CN110688964A (en) * | 2019-09-30 | 2020-01-14 | 哈尔滨工程大学 | Wavelet threshold and EMD combined denoising method based on sparse decomposition |
| CN111505650A (en) * | 2020-04-28 | 2020-08-07 | 西北工业大学 | HPSS-based underwater target passive detection method |
| CN111970627A (en) * | 2020-08-31 | 2020-11-20 | 广州视源电子科技股份有限公司 | Audio signal enhancement method, device, storage medium and processor |
-
2021
- 2021-01-14 CN CN202110049764.XA patent/CN112885362B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6718316B1 (en) * | 2000-10-04 | 2004-04-06 | The United States Of America As Represented By The Secretary Of The Navy | Neural network noise anomaly recognition system and method |
| CN102176312A (en) * | 2011-01-07 | 2011-09-07 | 蔡镇滨 | System and method for reducing burst noise through wavelet trapped wave |
| CN105488466A (en) * | 2015-11-26 | 2016-04-13 | 中国船舶工业系统工程研究院 | Deep neural network and underwater sound target vocal print feature extraction method |
| CN108694346A (en) * | 2017-04-05 | 2018-10-23 | 中国科学院声学研究所 | A kind of Ship Radiated-Noise signal recognition method based on two-stage CNN |
| CN110246504A (en) * | 2019-05-20 | 2019-09-17 | 平安科技（深圳）有限公司 | Birds sound identification method, device, computer equipment and storage medium |
| CN110245608A (en) * | 2019-06-14 | 2019-09-17 | 西北工业大学 | A kind of Underwater targets recognition based on semi-tensor product neural network |
| CN110189757A (en) * | 2019-06-27 | 2019-08-30 | 电子科技大学 | A kind of giant panda individual discrimination method, equipment and computer readable storage medium |
| CN110598593A (en) * | 2019-08-29 | 2019-12-20 | 东南大学 | Planetary gearbox fault diagnosis method based on resonance sparse decomposition and FastICA algorithm |
| CN110556121A (en) * | 2019-09-18 | 2019-12-10 | 腾讯科技（深圳）有限公司 | Frequency band extension method, device, electronic equipment and computer readable storage medium |
| CN110688964A (en) * | 2019-09-30 | 2020-01-14 | 哈尔滨工程大学 | Wavelet threshold and EMD combined denoising method based on sparse decomposition |
| CN111505650A (en) * | 2020-04-28 | 2020-08-07 | 西北工业大学 | HPSS-based underwater target passive detection method |
| CN111970627A (en) * | 2020-08-31 | 2020-11-20 | 广州视源电子科技股份有限公司 | Audio signal enhancement method, device, storage medium and processor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112885362B (en) | 2024-04-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wu et al. | Enhancing sound texture in CNN-based acoustic scene classification | |
| Ibrahim et al. | Automatic classification of grouper species by their sounds using deep neural networks | |
| CN110786850B (en) | Electrocardiosignal identity recognition method and system based on multi-feature sparse representation | |
| Lawgali et al. | Handwritten Arabic character recognition: Which feature extraction method? | |
| Nair et al. | Image mining applications for underwater environment management-A review and research agenda | |
| Li et al. | Learning deep models from synthetic data for extracting dolphin whistle contours | |
| CN116740387A (en) | Underwater noise identification method based on continuous wavelet transformation and improved residual neural network | |
| Vickers et al. | Robust North Atlantic right whale detection using deep learning models for denoising | |
| Zhou et al. | A denoising representation framework for underwater acoustic signal recognition | |
| CN114722812A (en) | Method and system for analyzing vulnerability of multi-mode deep learning model | |
| CN115830436A (en) | Marine organism intelligent detection method based on deep learning | |
| Ashraf et al. | Underwater ambient-noise removing GAN based on magnitude and phase spectra | |
| CN110515063A (en) | Underwater acoustic signal processing method and apparatus based on the steady wavelet transform of iteration | |
| CN112885362A (en) | Target identification method, system, device and medium based on radiation noise | |
| Han et al. | Object classification on raw radar data using convolutional neural networks | |
| Li* et al. | Ground-roll noise attenuation based on convolutional neural network | |
| Zhou et al. | An attention-based multi-scale convolution network for intelligent underwater acoustic signal recognition | |
| Lin et al. | A discrete wavelet analysis of freak waves in the ocean | |
| Zhang et al. | MSLEFC: A low-frequency focused underwater acoustic signal classification and analysis system | |
| Chen et al. | A ship-radiated noise classification method based on domain knowledge embedding and attention mechanism | |
| CN113299298A (en) | Residual error unit, network and target identification method, system, device and medium | |
| Wang et al. | Underwater Acoustic Target Recognition Combining Multi-scale Features and Attention Mechanism | |
| Divyabarathi et al. | Object classification in underwater SONAR images using transfer learning based ensemble model | |
| Lu et al. | Transfer Learning of Image Classification Networks in Application to Dolphin Whistle Detection | |
| Li et al. | The learned multi-scale deep filters for underwater acoustic target modeling and recognition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |