CN107665712A - A kind of marine organisms recognition methods based on dynamic time warping - Google Patents

A kind of marine organisms recognition methods based on dynamic time warping Download PDF

Info

Publication number
CN107665712A
CN107665712A CN201710793186.4A CN201710793186A CN107665712A CN 107665712 A CN107665712 A CN 107665712A CN 201710793186 A CN201710793186 A CN 201710793186A CN 107665712 A CN107665712 A CN 107665712A
Authority
CN
China
Prior art keywords
dist
species
feature
mrow
distance
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.)
Withdrawn
Application number
CN201710793186.4A
Other languages
Chinese (zh)
Inventor
王栋
司纪锋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beihai Research Station Institute Of Acoustics Chinese Academy Of Sciences
Original Assignee
Beihai Research Station Institute Of Acoustics Chinese Academy Of Sciences
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beihai Research Station Institute Of Acoustics Chinese Academy Of Sciences filed Critical Beihai Research Station Institute Of Acoustics Chinese Academy Of Sciences
Priority to CN201710793186.4A priority Critical patent/CN107665712A/en
Publication of CN107665712A publication Critical patent/CN107665712A/en
Withdrawn legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L17/00—Speaker identification or verification techniques
    • 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 TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L17/00—Speaker identification or verification techniques
    • G10L17/06—Decision making techniques; Pattern matching strategies
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
    • G10L25/24—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being the cepstrum
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
    • G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Computational Linguistics (AREA)
  • Business, Economics & Management (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Game Theory and Decision Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a kind of marine organisms recognition methods based on dynamic time warping, include feature templates of the signal characteristic as the species of marine organisms species known to extraction, the feature templates of each species can include multiple features；Extract biological signal characteristic to be identified, the signal characteristic of biological signal characteristic to be identified and feature templates is subjected to pattern match using dynamic time warping algorithm, it is then the biology of this feature template that the match is successful in the threshold range of setting, otherwise it is unknown, marine organisms recognition methods disclosed in this invention is without training in advance, low volume data is only needed to carry out characteristic matching, the species belonging to signal need not be predicted and can be blended identification, without adjustment algorithm when data volume and species increase, there is good application prospect.

Description

A kind of marine organisms recognition methods based on dynamic time warping
Technical field
The present invention relates to a kind of marine organisms recognition methods, more particularly to a kind of marine organisms based on dynamic time warping Recognition methods.
Background technology
Marine organisms are identified using acoustic method, there is quick, accurate, sustainable observation.The sound of marine organisms Learn signal and be divided into actively and passively two kinds of forms, identification step includes data prediction, feature extraction, training identification model, mould Formula matches.
Current marine bioacoustics method Study of recognition is usually using the identification model for needing to train in advance in congener It is identified in kind.For example, first with training data train cetacean, fish identification model, it is cetacean signal is special during identification Sign input inputs fish signal characteristic to fish identification model, and then specifically which kind of cetacean drawn to cetacean identification model Or fish, such as blue whale or large yellow croaker.
In actual applications, there are 2 points of deficiencies in such a mode.First, it must predefine belonging to signal to be identified Species, several species mixing identification can not be accomplished.Second, the acoustic signal of collection marine organisms is more difficult, the sample data one of species As it is less, it is difficult to which logical too small amount of training data obtains preferable identification model parameter.When feature database renewal or species classification increase Added-time, need re -training or newly-built identification model.
The content of the invention
In order to solve the above technical problems, the invention provides a kind of marine organisms identification side based on dynamic time warping Method, to reach without training in advance, low volume data is only needed to carry out characteristic matching, without predicting the species belonging to signal Purpose when carrying out mixing identification, data volume and species increase without adjustment algorithm.
To reach above-mentioned purpose, technical scheme is as follows：
A kind of marine organisms recognition methods based on dynamic time warping, include the signal of marine organisms species known to extraction Feature templates of the feature as the species, the feature templates of each species can include multiple features；Extract biological letter to be identified Number feature, enters row mode using dynamic time warping algorithm by the signal characteristic of biological signal characteristic to be identified and feature templates Matching, it is then the biology of this feature template that the match is successful in the threshold range of setting, is otherwise unknown.
In such scheme, the signal characteristic is included as the envelope of active signal feature and as cpm signal feature Mel-frequency cepstrum coefficient.
In further technical scheme, the pattern matching step of the envelope is as follows：
(1) characteristic quantity of signal to be identified is denoted as F, and it is that dimension is m respectively that a certain feature of feature templates, which is denoted as M, F and M, With n vector, calculate F and M and the distance between often tie up, and the filling distance matrix dist, dist are m × n matrixes, calculation formula It is as follows：
Dist (i, j)=[F (i)-M (j)]2(1)；
(2) Cumulative Distance matrix D is calculated by dist, D is m × n matrix；
D (1,1)=dist (1,1), D the first row and first row is made to be added up respectively by dist the first row and first row Arrive：
D (i, 1)=dist (i, 1)+D (i-1,1), 2≤i≤m (2)；
D (1, j)=dist (1, j)+D (1, j-1), 2≤j≤n (3)；
Since arranging the 2nd row the 2nd, D (i, j) can only be obtained by a value in 3 points is cumulative, be D (i-1, j), D respectively (i, j-1), D (i-1, j-1), its minimum value is taken to be added the Cumulative Distance as position (i, j) with D (i, j), calculation formula is such as Under：
D (i, j)=dist (i, j)+min [D (i, j-1), D (i-1, j), D (i-1, j-1)] (4)；
(3) after filling Cumulative Distance matrix D, the path of a minimum Cumulative Distance is found from position (m, n) inverted order, will The position of path point is recorded into R as optimal path, and R is the matrix of r × 2, and r is the number of optimal path point；
(4) F and M minimum cumulative distance is D (m, n), and what is recorded in R is optimal path point, D (m, n) divided by r cans F and M distance are obtained, is designated as dFM；
(5) F and the distance d of feature in each Species Characteristics template are calculated successivelyFM, when the feature of a certain species in feature templates When quantity is more than one, to the d in same speciesFMAverage；If dFMMinimum value be less than setting threshold value, then belonging to M Species are recognition result；If dFMMinimum value be more than setting threshold value, then be determined as unknown species.
In further technical scheme, the pattern matching step of the mel-frequency cepstrum coefficient is as follows：
(1) the mel-frequency cepstrum coefficient for extracting signal to be identified is denoted as F, some mel-frequency cepstrum system of feature templates Number scale makees the matrix that M, F and M are m × d and n × d respectively, and m and n are frame numbers, and d is the dimension of each frame；Because mel-frequency falls Spectral coefficient is characterized in the form of matrix, and distance matrix dist is filled the distance between per frame by F and M, per frame between away from It is as follows from calculation formula：
(2) Cumulative Distance matrix D is calculated by dist, D is m × n matrix；
D (1,1)=dist (1,1), D the first row and first row is made to be added up respectively by dist the first row and first row Arrive：
D (i, 1)=dist (i, 1)+D (i-1,1), 2≤i≤m (2)；
D (1, j)=dist (1, j)+D (1, j-1), 2≤j≤n (3)；
Since arranging the 2nd row the 2nd, D (i, j) can only be obtained by a value in 3 points is cumulative, be D (i-1, j), D respectively (i, j-1), D (i-1, j-1), its minimum value is taken to be added the Cumulative Distance as position (i, j) with D (i, j), calculation formula is such as Under：
D (i, j)=dist (i, j)+min [D (i, j-1), D (i-1, j), D (i-1, j-1)] (4)；
(3) after filling Cumulative Distance matrix D, the path of a minimum Cumulative Distance is found from position (m, n) inverted order, will The position of path point is recorded into R as optimal path, and R is the matrix of r × 2, and r is the number of optimal path point；
(4) F and M minimum cumulative distance is D (m, n), and what is recorded in R is optimal path point, D (m, n) divided by r cans F and M distance are obtained, is designated as dFM；
(5) F and the distance d of feature in each Species Characteristics template are calculated successivelyFM, when the feature of a certain species in feature templates When quantity is more than one, to the d in same speciesFMAverage；If dFMMinimum value be less than setting threshold value, then belonging to M Species are recognition result；If dFMMinimum value be more than setting threshold value, then be determined as unknown species.
Pass through above-mentioned technical proposal, the marine organisms recognition methods provided by the invention based on dynamic time warping is for more The marine organisms of species and the application scenarios of small data quantity, the characteristics of acoustic signal passive according to marine organisms master, extract time domain Feature of the envelope of signal as active signal, extraction Mel frequency cepstral coefficients (Mel frequency cepstrum Coefficient, MFCC) feature as cpm signal, using dynamic time warping (Dynamic time warping, DTW) algorithm carries out automatic identification.The algorithm only needs low volume data to carry out characteristic matching, without precognition without training in advance Without adjustment algorithm when species belonging to signal can be blended identification, data volume and species increase, have good Application prospect.
Embodiment
The technical scheme in the embodiment of the present invention will be clearly and completely described below.
The invention provides a kind of marine organisms recognition methods based on dynamic time warping, specific embodiment are as follows：
Embodiment one：
Yellow croaker, perch, the envelope characteristic of blackhead active signal are extracted respectively, are selected 10 features of every kind of fish to be used as and are somebody's turn to do The feature templates of species, the signal characteristic of biological envelope characteristic to be identified and feature templates is entered into row mode according to following steps Matching：
(1) characteristic quantity of signal to be identified is denoted as F, and it is that dimension is m respectively that a certain feature of feature templates, which is denoted as M, F and M, With n vector, calculate F and M and the distance between often tie up, and the filling distance matrix dist, dist are m × n matrixes, calculation formula It is as follows：
Dist (i, j)=[F (i)-M (j)]2(1)；
(2) Cumulative Distance matrix D is calculated by dist, D is m × n matrix；
D (1,1)=dist (1,1), D the first row and first row is made to be added up respectively by dist the first row and first row Arrive：
D (i, 1)=dist (i, 1)+D (i-1,1), 2≤i≤m (2)；
D (1, j)=dist (1, j)+D (1, j-1), 2≤j≤n (3)；
Since arranging the 2nd row the 2nd, D (i, j) can only be obtained by a value in 3 points is cumulative, be D (i-1, j), D respectively (i, j-1), D (i-1, j-1), its minimum value is taken to be added the Cumulative Distance as position (i, j) with D (i, j), calculation formula is such as Under：
D (i, j)=dist (i, j)+min [D (i, j-1), D (i-1, j), D (i-1, j-1)] (4)；
(3) after filling Cumulative Distance matrix D, the path of a minimum Cumulative Distance is found from position (m, n) inverted order, will The position of path point is recorded into R as optimal path, and R is the matrix of r × 2, and r is the number of optimal path point；
(4) F and M minimum cumulative distance is D (m, n), and what is recorded in R is optimal path point, D (m, n) divided by r cans F and M distance are obtained, is designated as dFM；
(5) F and the distance d of feature in each Species Characteristics template are calculated successivelyFM, when the feature of a certain species in feature templates When quantity is more than one, to the d in same speciesFMAverage；If dFMMinimum value be less than setting threshold value, then belonging to M Species are recognition result；If dFMMinimum value be more than setting threshold value, then be determined as unknown species.
Recognition result is shown in Table 1：
The fish active signal recognition result of table 1
Yellow croaker Perch Blackhead
Feature templates quantity 10 10 10
Identify sample size 98 264 194
Correct identification sample size 98 262 193
Discrimination 100% 99.24% 99.48%
Embodiment two：
The mel-frequency cepstrum coefficient feature (MFCC) of 6 kinds of fish cpm signals is extracted, selects 5 features of every kind of fish As the feature templates of the species, Classification and Identification is carried out using DTW algorithms to 3 kinds, 5 kinds, 6 kinds of fish respectively, specific steps are such as Under：
(1) the mel-frequency cepstrum coefficient for extracting signal to be identified is denoted as F, some mel-frequency cepstrum system of feature templates Number scale makees the matrix that M, F and M are m × d and n × d respectively, and m and n are frame numbers, and d is the dimension of each frame；Because mel-frequency falls Spectral coefficient is characterized in the form of matrix, and distance matrix dist is filled the distance between per frame by F and M, per frame between away from It is as follows from calculation formula：
(2) Cumulative Distance matrix D is calculated by dist, D is m × n matrix；
D (1,1)=dist (1,1), D the first row and first row is made to be added up respectively by dist the first row and first row Arrive：
D (i, 1)=dist (i, 1)+D (i-1,1), 2≤i≤m (2)；
D (1, j)=dist (1, j)+D (1, j-1), 2≤j≤n (3)；
Since arranging the 2nd row the 2nd, D (i, j) can only be obtained by a value in 3 points is cumulative, be D (i-1, j), D respectively (i, j-1), D (i-1, j-1), its minimum value is taken to be added the Cumulative Distance as position (i, j) with D (i, j), calculation formula is such as Under：
D (i, j)=dist (i, j)+min [D (i, j-1), D (i-1, j), D (i-1, j-1)] (4)；
(3) after filling Cumulative Distance matrix D, the path of a minimum Cumulative Distance is found from position (m, n) inverted order, will The position of path point is recorded into R as optimal path, and R is the matrix of r × 2, and r is the number of optimal path point；
(4) F and M minimum cumulative distance is D (m, n), and what is recorded in R is optimal path point, D (m, n) divided by r cans F and M distance are obtained, is designated as dFM；
(5) F and the distance d of feature in each Species Characteristics template are calculated successivelyFM, when the feature of a certain species in feature templates When quantity is more than one, to the d in same speciesFMAverage；If dFMMinimum value be less than setting threshold value, then belonging to M Species are recognition result；If dFMMinimum value be more than setting threshold value, then be determined as unknown species.
Recognition result is shown in table 2.
The fish cpm signal recognition result of table 2
Fish species quantity 3 kinds of fish 5 kinds of fish 6 kinds of fish
Feature templates quantity 15 25 30
Identify sample size 58 80 94
Correct identification sample size 58 77 89
Discrimination 100% 96.25% 94.68%
Embodiment three：
The mel-frequency cepstrum coefficient feature (MFCC) of 3 kinds of shrimps and 12 kinds of cetacean cpm signals is extracted, with embodiment Two 6 Mesichthyes common (21 species altogether) carry out mixing identification, and specific steps are identical with embodiment two.
Recognition result is shown in table 3.
The cpm signal mixing recognition result of table 3
Mix species composition 6 kinds of fish 3 kinds of shrimps 12 kinds of cetaceans All species (21 kinds)
Feature templates quantity 30 15 60 105
Identify sample size 94 38 85 217
Mixing identifies correct sample size 81 32 77 190
Mix discrimination 86.17% 84.21% 90.59% 87.56%
Recognition result shows that DTW algorithms are respectively provided with higher discrimination to main cpm signal.
The foregoing description of the disclosed embodiments, professional and technical personnel in the field are enable to realize or using the present invention. A variety of modifications to these embodiments will be apparent for those skilled in the art, as defined herein General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, it is of the invention The embodiments shown herein is not intended to be limited to, and is to fit to and principles disclosed herein and features of novelty phase one The most wide scope caused.

Claims (4)

1. a kind of marine organisms recognition methods based on dynamic time warping, it is characterised in that including marine organisms known to extraction Feature templates of the signal characteristic of species as the species, the feature templates of each species can include multiple features；Extract and wait to know Not biological signal characteristic, it is using dynamic time warping algorithm that the signal of biological signal characteristic to be identified and feature templates is special Sign carries out pattern match, and it is then the biology of this feature template that the match is successful in the threshold range of setting, is otherwise unknown.
A kind of 2. marine organisms recognition methods based on dynamic time warping according to claim 1, it is characterised in that institute Stating signal characteristic includes the envelope as active signal feature and the mel-frequency cepstrum coefficient as cpm signal feature.
A kind of 3. marine organisms recognition methods based on dynamic time warping according to claim 2, it is characterised in that institute The pattern matching step for stating envelope is as follows：
(1) characteristic quantity of signal to be identified is denoted as F, and it is that dimension is m and n respectively that a certain feature of feature templates, which is denoted as M, F and M, Vector, calculate F and M and the distance between often tie up, and the filling distance matrix dist, dist are m × n matrixes, calculation formula is as follows：
Dist (i, j)=[F (i)-M (j)]2(1)；
(2) Cumulative Distance matrix D is calculated by dist, D is m × n matrix；
Make D (1,1)=dist (1,1), D the first row and first row be added up by dist the first row and first row respectively to obtain：
D (i, 1)=dist (i, 1)+D (i-1,1), 2≤i≤m (2)；
D (1, j)=dist (1, j)+D (1, j-1), 2≤j≤n (3)；
Since the 2nd row the 2nd arrange, D (i, j) can only by 3 points a value add up obtain, be respectively D (i-1, j), D (i, J-1), D (i-1, j-1), takes its minimum value to be added the Cumulative Distance as position (i, j) with D (i, j), and calculation formula is as follows：
D (i, j)=dist (i, j)+min [D (i, j-1), D (i-1, j), D (i-1, j-1)] (4)；
(3) after filling Cumulative Distance matrix D, the path of a minimum Cumulative Distance is found from position (m, n) inverted order, by path The position of point is recorded into R as optimal path, and R is the matrix of r × 2, and r is the number of optimal path point；
(4) F and M minimum cumulative distance is D (m, n), and what is recorded in R is optimal path point, and D (m, n) divided by r can be obtained by F and M distance, is designated as dFM；
(5) F and the distance d of feature in each Species Characteristics template are calculated successivelyFM, when the feature quantity of a certain species in feature templates When more than one, to the d in same speciesFMAverage；If dFMMinimum value be less than the threshold value of setting, then species belonging to M As recognition result；If dFMMinimum value be more than setting threshold value, then be determined as unknown species.
A kind of 4. marine organisms recognition methods based on dynamic time warping according to claim 2, it is characterised in that institute The pattern matching step for stating mel-frequency cepstrum coefficient is as follows：
(1) the mel-frequency cepstrum coefficient for extracting signal to be identified is denoted as F, some mel-frequency cepstrum coefficient note of feature templates Make the matrix that M, F and M are m × d and n × d respectively, m and n are frame numbers, and d is the dimension of each frame；Due to mel-frequency cepstrum system Number is characterized in the form of matrix, and distance matrix dist is filled by the distance between every frames of F and M, the distance between every frame meter It is as follows to calculate formula：
<mrow> <mi>d</mi> <mi>i</mi> <mi>s</mi> <mi>t</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>p</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>d</mi> </munderover> <msup> <mrow> <mo>&amp;lsqb;</mo> <mi>F</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>p</mi> <mo>)</mo> </mrow> <mo>-</mo> <mi>M</mi> <mrow> <mo>(</mo> <mi>j</mi> <mo>,</mo> <mi>p</mi> <mo>)</mo> </mrow> <mo>&amp;rsqb;</mo> </mrow> <mn>2</mn> </msup> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>5</mn> <mo>)</mo> </mrow> <mo>;</mo> </mrow>
(2) Cumulative Distance matrix D is calculated by dist, D is m × n matrix；
Make D (1,1)=dist (1,1), D the first row and first row be added up by dist the first row and first row respectively to obtain：
D (i, 1)=dist (i, 1)+D (i-1,1), 2≤i≤m (2)；
D (1, j)=dist (1, j)+D (1, j-1), 2≤j≤n (3)；
Since the 2nd row the 2nd arrange, D (i, j) can only by 3 points a value add up obtain, be respectively D (i-1, j), D (i, J-1), D (i-1, j-1), takes its minimum value to be added the Cumulative Distance as position (i, j) with D (i, j), and calculation formula is as follows：
D (i, j)=dist (i, j)+min [D (i, j-1), D (i-1, j), D (i-1, j-1)] (4)；
(3) after filling Cumulative Distance matrix D, the path of a minimum Cumulative Distance is found from position (m, n) inverted order, by path The position of point is recorded into R as optimal path, and R is the matrix of r × 2, and r is the number of optimal path point；
(4) F and M minimum cumulative distance is D (m, n), and what is recorded in R is optimal path point, and D (m, n) divided by r can be obtained by F and M distance, is designated as dFM；
(5) F and the distance d of feature in each Species Characteristics template are calculated successivelyFM, when the feature quantity of a certain species in feature templates When more than one, to the d in same speciesFMAverage；If dFMMinimum value be less than the threshold value of setting, then species belonging to M As recognition result；If dFMMinimum value be more than setting threshold value, then be determined as unknown species.
CN201710793186.4A 2017-09-06 2017-09-06 A kind of marine organisms recognition methods based on dynamic time warping Withdrawn CN107665712A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710793186.4A CN107665712A (en) 2017-09-06 2017-09-06 A kind of marine organisms recognition methods based on dynamic time warping

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710793186.4A CN107665712A (en) 2017-09-06 2017-09-06 A kind of marine organisms recognition methods based on dynamic time warping

Publications (1)

Publication Number Publication Date
CN107665712A true CN107665712A (en) 2018-02-06

Family

ID=61097313

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710793186.4A Withdrawn CN107665712A (en) 2017-09-06 2017-09-06 A kind of marine organisms recognition methods based on dynamic time warping

Country Status (1)

Country Link
CN (1) CN107665712A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109082984A (en) * 2018-06-29 2018-12-25 浙江大学城市学院 A kind of road abnormality detection model based on window division and dynamic time warping
CN112074073A (en) * 2020-08-20 2020-12-11 成都奕康真空电子技术有限责任公司 Control method and system for stabilizing front target dose rate of industrial electronic linear accelerator
WO2022034749A1 (en) * 2020-08-12 2022-02-17 日本電気株式会社 Aquatic organism observation device, aquatic organism observation system, aquatic organism observation method, and recording medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103021421A (en) * 2012-12-24 2013-04-03 天津大学 Multilevel screening detecting recognizing method for shots
CN103489446A (en) * 2013-10-10 2014-01-01 福州大学 Twitter identification method based on self-adaption energy detection under complex environment
CN104102923A (en) * 2014-07-16 2014-10-15 西安建筑科技大学 Nipponia nippon individual recognition method based on MFCC algorithm
CN105589073A (en) * 2014-10-24 2016-05-18 中国科学院声学研究所 Walsh-conversion-based fish active acoustic identification method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103021421A (en) * 2012-12-24 2013-04-03 天津大学 Multilevel screening detecting recognizing method for shots
CN103489446A (en) * 2013-10-10 2014-01-01 福州大学 Twitter identification method based on self-adaption energy detection under complex environment
CN104102923A (en) * 2014-07-16 2014-10-15 西安建筑科技大学 Nipponia nippon individual recognition method based on MFCC algorithm
CN105589073A (en) * 2014-10-24 2016-05-18 中国科学院声学研究所 Walsh-conversion-based fish active acoustic identification method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
CHE YONG YEO ET AL: "Animal Voice Recognition for Identification (ID) Detection System", 《2011 IEEE 7TH INTERNATIONAL COLLOQUIUM ON SIGNAL PROCESSING AND ITS APPLICATIONS》 *
张浩等: "基于动态时间规整的步态自动识别", 《图像图形技术研究与应用2009》 *
许枫等: "Walsh变换对鱼类特征识别的研究", 《应用声学》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109082984A (en) * 2018-06-29 2018-12-25 浙江大学城市学院 A kind of road abnormality detection model based on window division and dynamic time warping
WO2022034749A1 (en) * 2020-08-12 2022-02-17 日本電気株式会社 Aquatic organism observation device, aquatic organism observation system, aquatic organism observation method, and recording medium
JP7444264B2 (en) 2020-08-12 2024-03-06 日本電気株式会社 Aquatic organism observation equipment and aquatic organism observation method
CN112074073A (en) * 2020-08-20 2020-12-11 成都奕康真空电子技术有限责任公司 Control method and system for stabilizing front target dose rate of industrial electronic linear accelerator
CN112074073B (en) * 2020-08-20 2023-01-10 成都奕康真空电子技术有限责任公司 Control method and system for stabilizing front target dose rate of industrial electronic linear accelerator

Similar Documents

Publication Publication Date Title
JP6774551B2 (en) Speech recognition processing method and equipment
CN105469784B (en) A kind of speaker clustering method and system based on probability linear discriminant analysis model
CN108734114A (en) A kind of pet recognition methods of combination face harmony line
CN109493874A (en) A kind of live pig cough sound recognition methods based on convolutional neural networks
CN104809103A (en) Man-machine interactive semantic analysis method and system
CN107358945A (en) A kind of more people&#39;s conversation audio recognition methods and system based on machine learning
CN107564513A (en) Audio recognition method and device
CN105261367B (en) A kind of method for distinguishing speek person
CN107665712A (en) A kind of marine organisms recognition methods based on dynamic time warping
CN109871885A (en) A kind of plants identification method based on deep learning and Plant Taxonomy
CN102201236A (en) Speaker recognition method combining Gaussian mixture model and quantum neural network
CN110992988B (en) Speech emotion recognition method and device based on domain confrontation
CN104700120A (en) Method for extracting and classifying fMRI features based on adaptive entropy algorithm for projection clustering (APEC)
CN107767881B (en) Method and device for acquiring satisfaction degree of voice information
CN109256137A (en) Voice acquisition method, device, computer equipment and storage medium
WO2021233041A1 (en) Data annotation method and device, and fine granularity identification method and device
CN104063713A (en) Semi-autonomous on-line studying method based on random fern classifier
CN110019779B (en) Text classification method, model training method and device
CN104091181A (en) Injurious insect image automatic recognition method and system based on deep restricted Boltzmann machine
CN107103311A (en) A kind of recognition methods of continuous sign language and its device
CN107679231A (en) A kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System
CN102237089B (en) Method for reducing error identification rate of text irrelevant speaker identification system
CN109409182A (en) Embryo&#39;s automatic identifying method based on image procossing
CN111144462A (en) Unknown individual identification method and device for radar signals
CN111091809A (en) Regional accent recognition method and device based on depth feature fusion

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
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20180206