CN105578373B - A kind of loudspeaker simple tone detecting method dividing shape based on image - Google Patents
A kind of loudspeaker simple tone detecting method dividing shape based on image Download PDFInfo
- Publication number
- CN105578373B CN105578373B CN201410537518.9A CN201410537518A CN105578373B CN 105578373 B CN105578373 B CN 105578373B CN 201410537518 A CN201410537518 A CN 201410537518A CN 105578373 B CN105578373 B CN 105578373B
- Authority
- CN
- China
- Prior art keywords
- image
- time
- frequency
- loudspeaker
- fractal
- 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.)
- Expired - Fee Related
Links
Landscapes
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of malfunctioning speaker simple tone detecting methods for dividing shape based on image, including Time-Frequency Analysis Method and image fractal method.Time-Frequency Analysis Method includes: to carry out wavelet package transforms to the response signal of loudspeaker using the orthogonality and time-frequency locality of wavelet function, it is translated into two dimensional image signal, again to two dimensional image binaryzation and edge extracting, the response signal time-frequency image of differing complexity is obtained.Image fractal method includes: the calculating to time-frequency image after edge extracting by box-counting dimension, compares the complexity of time-frequency image.The present invention is based on image fractal theories, and in conjunction with modern advanced signal processing mode, the precision of discrimination is high, and efficiency is big, has great importance to the on-line monitoring of loudspeaker.
Description
Technical field
The present invention relates to one kind to be used for the online simple tone detecting method of loudspeaker.
Background technique
In loudspeaker as modern necessity, industrial technology product, to its performance carry out strictly, to accurately measure be it
Indispensable link in trial-production, production.It is pure tone detection that first mass, which surveys inspection process, i.e., application sine sweep signal with
Motivate loudspeaker, the whether pure qualification of tonal signal of its sending of detection.So far produced worldwide producer be still all according to
This work is completed by human ear, testing result is because of the age of people, gender, mood and prolonged detection, the generated sense of hearing
Fatigue etc. all can bring difference to result.
Currently, most of in terms of loudspeaker faults diagnosis both at home and abroad detected from the design and electro mechanical parameter of loudspeaker
Angle study.Less using Time-Frequency Analysis Method extraction loudspeaker faults characteristic aspect research, only Cambridge in 2005 is big
Manuel doctor Davy has carried out loudspeaker response signal using three kinds of time-frequency distributions that the kernel function of Cohen class generates
Analysis, but be not known and propose fault signature classification method;The Pascal Brunet of Tufts university utilizes short time discrete Fourier transform
The time-frequency distributions situation of qualified loudspeaker and malfunctioning speaker is analyzed, is divided whether how also not providing to loudspeaker faults
Class.It can effectively improve the fault signature of loudspeaker response signal using Time-Frequency Analysis Method herein, fractal dimension can specify table
Up to time-frequency image complexity the characteristics of loudspeaker is analyzed.Therefore using Wavelet Packet Transform Method to loudspeaker response
Signal carries out time-frequency conversion, keeps the fault signature of loudspeaker obvious;The meter box dimension of time-frequency image is obtained with image fractal method
Number, and pass through whether clustering method carries out failure to loudspeaker in this, as fault signature and identify.
Summary of the invention
The present invention provides a kind of malfunctioning speaker simple tone detecting method based on image point shape, in loudspeaker event online
In barrier detection, including Time-Frequency Analysis Method and image fractal method, mainly by becoming the response signal of loudspeaker through wavelet packet
Change be converted into two dimensional image signal with improve the time-frequency of fault message characterization, through image binaryzation and edge extracting pretreatment after,
It as fault signature and is identified by the box-counting dimension that image fractal method extracts time-frequency image again.
Present invention provide the technical scheme that a kind of malfunctioning speaker simple tone detecting method for dividing shape based on image, special
Sign is: motivating loudspeaker with the logarithmic sine swept-frequency signal of 20Hz-1500Hz-20Hz, acquires the response signal of loudspeaker, obtain
Response signal time-domain image is obtained, obtains the time-frequency image of a frequency sweep cycle by wavelet package transforms;Then to time-frequency image into
Row binaryzation and edge extracting, the calculating for dividing the method for shape to carry out box counting dimension to loudspeaker treated image using image, most
The complexity for determining image according to calculated result afterwards, that is, whether judging the failure of loudspeaker.This method includes time frequency analysis side
Method and image fractal method;Time-Frequency Analysis Method is completed when motivating loudspeaker acquisition response signal, and image fractal method is right
It is completed when after the image progress binaryzation and edge processing after time frequency analysis.Loudspeaker time-frequency image box-counting dimension algorithm flow chart
As shown in Figure 1.
Wherein the Time-Frequency Analysis Method the following steps are included:
Step 1, step 1 motivate loudspeaker with the logarithmic sine swept-frequency signal of 20Hz-1500Hz-20Hz, acquire loudspeaker
Response signal, obtain response signal time-domain image;
Step 1.1, using microphone pick response signal;
Step 1.2, when time-frequency image obtains, for it is subsequent based on image divide shape fault signature extract it is convenient, in figure
Gray value carries out inverse pretreatment, and the biggish part of energy value in image is rendered as black or Dark grey, and energy value is close to 0
Part be rendered as white.
Step 2 is handled time-domain image by time frequency analysis, obtains the time-frequency image of a frequency sweep cycle;
Wherein the image fractal method the following steps are included:
Step 1, on time-frequency image, binaryzation and edge extracting are carried out to image;
Step 2 calculates the complexity of image using the definition of box-counting dimension by image fractal principle;
Step 2.1, using grid cladding process, used side length of element maximum value depends on the side length of image overall area,
Side length of element minimum value is always 1;
Step 2.2, time-frequency image are calculated in the way of bisection;
Time-frequency image is established as nonsingular matrix by step 2.3, calculates the number of non-zero submatrices;
Step 2.4 calculates meter box size and number, and it is quasi- that Least squares approach is then carried out under logarithmic coordinates system
It closes, obtaining its slope is box-counting dimension;
Step 3, whether judge the failure of loudspeaker by the result of calculating.
The present invention is since using the malfunctioning speaker simple tone detecting method for dividing shape based on image, key technology is to design
Judgement whether based on image fractal method come to loudspeaker faults;For adapt to on-line checking needs, using time frequency analysis
Method extracts feature, identifies whether using image fractal method to the failure of loudspeaker.
Detailed description of the invention
Fig. 1 is loudspeaker time-frequency image box-counting dimension algorithm flow chart of the invention;
Fig. 2 is that cover of the invention is bonded malfunctioning speaker wavelet packet time-frequency image grayscale image;
Fig. 3 is that cover of the invention is bonded malfunctioning speaker wavelet packet time-frequency image edge extracting figure;
Fig. 4 is loudspeaker time-frequency image box-counting dimension algorithm flow chart of the invention.
Specific embodiment
In conjunction with attached drawing 1, specific embodiments of the present invention are illustrated:
Pass through microphone pick loudspeaker response signal;
In application process, it is divided into time frequency analysis and image is divided to shape two processes.
The process of time frequency analysis mainly comprises the steps that
A, loudspeaker is motivated with the logarithmic sine swept-frequency signal of 20Hz-1500Hz-20Hz, acquires the response letter of loudspeaker
Number, obtain response signal time-domain image;
B, time-domain image is handled by time frequency analysis, obtains the time-frequency image of a frequency sweep cycle.;
Its specific steps please refers to shown in Fig. 1 are as follows:
Into time frequency analysis process 1, the response signal 2 of loudspeaker is acquired, by signal processing analysis, obtains response signal
Time-domain image 3 carries out processing 4 to time-domain image by time frequency analysis, obtains the time-frequency image of a frequency sweep cycle.
Image divides shape process, mainly comprises the steps that
A, on time-frequency image, binaryzation and edge extracting are carried out to image;
B, the complexity of image is calculated using the definition of box-counting dimension by image fractal principle;
C, whether judging the failure of loudspeaker by the result of calculating.
Its specific steps is as shown in Figure 1 are as follows:
Divide shape process 5 into image, binaryzation and edge extracting 6 is carried out to image, meter is utilized by image fractal principle
The definition of box counting dimension carries out calculating 7 to the complexity of image, 8 whether judging the failure of loudspeaker by the result of calculating.
Concrete principle is as follows with implementation method.
1, loudspeaker generates sound wave using non linear mechanical mode of vibration, therefore its response signal belongs to non-stationary signal,
The fault signature of loudspeaker response signal has different manifestations under different time different frequency, in the high-order frequency of its response signal
Particularly evident in rate, the response signal by acquiring loudspeaker obtains time-frequency image.A frequency sweep is obtained by Time-frequency Analysis
The time-frequency image in period.Time-frequency Analysis mainly passes through Wavelet Packet Transform Method and carries out time-frequency conversion to loudspeaker response signal.
2, the method for dividing the malfunctioning speaker pure tone of shape to detect based on image.
Image fractal method can characterize under certain condition or in the process, show in one aspect to it is whole similar
Property.In the present invention, using image fractal method for the calculating of the loudspeaker time-frequency image complexity after edge extracting, pass through
Comparison of computational results finds out malfunctioning speaker.
Box-counting dimension DBThere are the definition of a series of equivalent, usual approximate calculation box-counting dimension are as follows:
D in formulaBFor box-counting dimension;F is set of the Hausdorff dimension strictly larger than its topological dimension;For grid division
Side length;For the number for covering image Region Of Interest grid.
For a breadth w pixel, time-frequency image after the edge extracting of high h pixel is represented by a h*w matrix, square
The position of each element corresponds to a pixel position in battle array.Black pixel point indicates that white is 0 in corresponding matrix with 1.
Using grid cladding process, used side length of element maximum value depends on the side length of image overall area, and side length of element minimum value begins
It is eventually 1.From each fault signature of vehicle-mounted loudspeaker time-frequency image it is found that meter box region shared by energy major part is got in image
More, then box-counting dimension is bigger, and failure performance is more obvious.Therefore, image fractal dimension can be used as the typical fault table of time-frequency image
Sign.As shown in Figure 1, being divided into such a way that each width is pre-processed rear speaker time-frequency image according to bisectionPart,
K meter box size and number, are then carried out best square under logarithmic coordinates system and forced by the calculating for carrying out meter box size and number
Nearly fitting, obtaining its slope is box-counting dimension.
As shown in Figure 2, it is stretched due to image in the case where not changing pixel number, does not influence the table of fault message
It is existing, therefore h=w can be set, time-frequency image is established as nonsingular square matrix.So establishing it halves time-frequency image box-counting dimension number
Model is learned to be shown below.
In formula,For fractal dimension, nkThe Region Of Interest lattice number obtained after as halving every time, that is, characterize
The energy point number of failure.
Loudspeaker response signal is easy to extract, and image fractal dimension algorithm is easily achieved.Guaranteeing loudspeaker equal conditions
Under, provide a kind of method of loudspeaker simple tone detection.Finally illustrate: above embodiments are only to illustrate the present invention rather than limit
Technical solution described in the invention;Therefore, although this specification referring to above-mentioned each embodiment to present invention has been
Detailed description, but still can modify to the present invention or equivalent replacement;Its technical solution and its improvement should all be covered in this hair
In bright scope of the claims.
Claims (4)
1. a kind of malfunctioning speaker simple tone detecting method for dividing shape based on image, it is characterised in that: use 20Hz-1500Hz-20Hz
Logarithmic sine swept-frequency signal motivate loudspeaker, acquire the response signal of loudspeaker, response signal time-domain image obtained, by small
Wave packet transform obtains the time-frequency image of a frequency sweep cycle;Then binaryzation and edge extracting are carried out to time-frequency image, utilizes figure
As the calculating for dividing the method for shape to carry out box counting dimension to loudspeaker treated image, answering for image is finally determined according to calculated result
Miscellaneous degree, that is, whether judging the failure of loudspeaker, wherein with method include: Time-Frequency Analysis Method and image fractal method;
Wherein the Time-Frequency Analysis Method the following steps are included:
Step 1 motivates loudspeaker with the logarithmic sine swept-frequency signal of 20Hz-1500Hz-20Hz, acquires the response letter of loudspeaker
Number, obtain response signal time-domain image;
Step 2 is handled time-domain image by time frequency analysis, obtains the time-frequency image of a frequency sweep cycle;
Wherein the image fractal method the following steps are included:
Step 1, on time-frequency image, binaryzation and edge extracting are carried out to image;
Step 2 calculates the complexity of image using the definition of box-counting dimension by image fractal principle;
Step 3, box-counting dimension are bigger, and failure performance is more obvious, whether judging the failure of loudspeaker by the result of calculating.
2. the simple tone detecting method of the malfunctioning speaker according to claim 1 for dividing shape based on image, it is characterized in that: time-frequency
The step 1 of analysis method and image fractal method further include:
Step 1.1, using microphone pick response signal;
Step 1.2, when time-frequency image obtains, for it is subsequent based on image divide shape fault signature extract it is convenient, to gray scale in figure
Value carries out inverse pretreatment, and the biggish part of energy value in image is rendered as black or Dark grey, portion of the energy value close to 0
Divide and is rendered as white.
3. the simple tone detecting method of the malfunctioning speaker according to claim 1 for dividing shape based on image, it is characterized in that: time-frequency
Step 2 described in analysis method further include: the Time-Frequency Information that wavelet packet basis is corresponded in analysis signal takes analysis signal and wavelet packet
The inner product of base obtains the corresponding wavelet packet coefficient of analysis signal.
4. the simple tone detecting method of the malfunctioning speaker according to claim 1 for dividing shape based on image, it is characterized in that: image
Step 2 described in fractal method further include:
Step 4.1, using grid cladding process, used side length of element maximum value depends on the side length of image overall area, grid
Side length minimum value is always 1;
Step 4.2, time-frequency image are calculated in the way of bisection;
Time-frequency image is established as nonsingular matrix by step 4.3, calculates the number of non-zero submatrices;
Step 4.4 calculates meter box size and number, and Least squares approach fitting is then carried out under logarithmic coordinates system,
Obtaining its slope is box-counting dimension.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410537518.9A CN105578373B (en) | 2014-10-14 | 2014-10-14 | A kind of loudspeaker simple tone detecting method dividing shape based on image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410537518.9A CN105578373B (en) | 2014-10-14 | 2014-10-14 | A kind of loudspeaker simple tone detecting method dividing shape based on image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105578373A CN105578373A (en) | 2016-05-11 |
CN105578373B true CN105578373B (en) | 2019-01-15 |
Family
ID=55887948
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410537518.9A Expired - Fee Related CN105578373B (en) | 2014-10-14 | 2014-10-14 | A kind of loudspeaker simple tone detecting method dividing shape based on image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105578373B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107657610B (en) * | 2017-09-29 | 2020-10-16 | 哈尔滨工业大学 | CT scanning result analysis method based on box counting method |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6178261B1 (en) * | 1997-08-05 | 2001-01-23 | The Regents Of The University Of Michigan | Method and system for extracting features in a pattern recognition system |
CN101170843A (en) * | 2007-11-30 | 2008-04-30 | 清华大学 | Speaker online pure voice failure diagnosis method |
CN201438762U (en) * | 2009-04-15 | 2010-04-14 | 天津科技大学 | Sound attenuating chamber for loudspeaker pure tone online detection |
CN101900789A (en) * | 2010-07-07 | 2010-12-01 | 湖南大学 | Tolerance analog circuit fault diagnosing method based on wavelet transform and fractal dimension |
CN102474683A (en) * | 2009-08-03 | 2012-05-23 | 图象公司 | Systems and method for monitoring cinema loudspeakers and compensating for quality problems |
-
2014
- 2014-10-14 CN CN201410537518.9A patent/CN105578373B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6178261B1 (en) * | 1997-08-05 | 2001-01-23 | The Regents Of The University Of Michigan | Method and system for extracting features in a pattern recognition system |
CN101170843A (en) * | 2007-11-30 | 2008-04-30 | 清华大学 | Speaker online pure voice failure diagnosis method |
CN201438762U (en) * | 2009-04-15 | 2010-04-14 | 天津科技大学 | Sound attenuating chamber for loudspeaker pure tone online detection |
CN102474683A (en) * | 2009-08-03 | 2012-05-23 | 图象公司 | Systems and method for monitoring cinema loudspeakers and compensating for quality problems |
CN101900789A (en) * | 2010-07-07 | 2010-12-01 | 湖南大学 | Tolerance analog circuit fault diagnosing method based on wavelet transform and fractal dimension |
Also Published As
Publication number | Publication date |
---|---|
CN105578373A (en) | 2016-05-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103871053B (en) | Vision conspicuousness-based cloth flaw detection method | |
US8630962B2 (en) | Error detection method and its system for early detection of errors in a planar or facilities | |
JP6289400B2 (en) | Method and system for detecting events in a signal subject to periodic stationary background noise | |
CN102445640B (en) | GIS device intelligent recognition method based on vector machine and artificial fish swarm optimization | |
CN111161756B (en) | Method for extracting and identifying abnormal whistle contour in wind sweeping sound signal of fan blade | |
JP6061693B2 (en) | Abnormality diagnosis apparatus and abnormality diagnosis method using the same | |
CN110398647B (en) | Transformer state monitoring method | |
JP2020519909A (en) | Water leak detection method for water pipes | |
CN104849633A (en) | Switchgear partial discharge mode recognition method | |
CN105277567A (en) | Fabric flaw detection method | |
CN112727710B (en) | Wind field thunderbolt density statistical method and system based on audio signals | |
RU2010149800A (en) | METHOD AND DEVICE FOR CLASSIFICATION OF GENERATING SOUND PROCESSES | |
CN110795843A (en) | Method and device for identifying faults of rolling bearing | |
Du et al. | Weighted low-rank sparse model via nuclear norm minimization for bearing fault detection | |
CN103955922A (en) | Method for detecting flaws of printed fabric based on Gabor filter | |
CN104766099A (en) | Distributed optical fiber circumference vibration signal processing and recognizing method based on image | |
JP2012018066A (en) | Device for inspecting abnormality | |
CN105578373B (en) | A kind of loudspeaker simple tone detecting method dividing shape based on image | |
CN106955097A (en) | A kind of fetal heart frequency state classification method | |
KR101526100B1 (en) | Defect characteristic extracting method and apparatus of induction motor using 3 dimensional color model | |
Hua et al. | Mechanical fault diagnosis using color image recognition of vibration spectrogram based on quaternion invariable moment | |
CN102346171B (en) | Flaw detecting system and method | |
CN116309493A (en) | Method and system for detecting defects of textile products | |
CN116541771A (en) | Unbalanced sample bearing fault diagnosis method based on multi-scale feature fusion | |
CN101996316A (en) | Fuzzy identification method for identifying cracks of deep drawn part |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190115 Termination date: 20201014 |
|
CF01 | Termination of patent right due to non-payment of annual fee |