CN109165545A - A kind of new signal characteristic extracting methods based on fractal dimension - Google Patents
A kind of new signal characteristic extracting methods based on fractal dimension Download PDFInfo
- Publication number
- CN109165545A CN109165545A CN201810734013.XA CN201810734013A CN109165545A CN 109165545 A CN109165545 A CN 109165545A CN 201810734013 A CN201810734013 A CN 201810734013A CN 109165545 A CN109165545 A CN 109165545A
- Authority
- CN
- China
- Prior art keywords
- dimension
- signal
- box counting
- fractal dimension
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
- G06F2218/10—Feature extraction by analysing the shape of a waveform, e.g. extracting parameters relating to peaks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Complex Calculations (AREA)
Abstract
The invention proposes a kind of new signal characteristic extracting methods based on fractal dimension, which is characterized in that extracts box counting dimension, Higuchi fractal dimension, Katz fractal dimension and the improvement box counting dimension of signal respectively.The present invention is in order to improve feature extraction algorithm to the anti-interference ability of noise, propose a kind of new modulated signal feature extraction algorithm based on box counting dimension, it is proposed two new parameters: kurtosis reconciliation parameter and harmonic average box counting dimension, and with traditional box counting dimension, Higuchi fractal dimension, Katz fractal dimension, four-dimensional Cancers Fractional Dimension Feature vector is collectively formed, and then realizes and the multidimensional fractal characteristic of signal is extracted.The present invention calculates simple, can identify to complicated modulated signal, this is for fractal dimension in field of signal identification using providing certain theoretical foundation.
Description
Technical field
The present invention relates to a kind of signal characteristic extracting methods, belong to field of digital information processing.
Background technique
Signal characteristic abstraction is to analyze the correlated characteristic information of signal, and then extraction effectively can distinguish and identify letter
Number process.It is widely used in, and image recognition, noise processed, fault detection and medical diagnosis etc. are nearly all to be related to letter
Cease the field of processing.Signal characteristic abstraction is basis and the key of digital information processing.Signal by feature extraction is special
Whether sign improves and correctly directly affects the recognition effect of whole system.Since the signal of physical presence has very strong complexity
And therefore how erratic behavior efficiently extracts the difficult point that signal characteristic is always the research.
Up to now, feature extraction mode and algorithm emerge one after another, but common signal recognition method still remain it is following
Two o'clock is insufficient: being difficult to non-linear and signal the non-stationary signal feature of accurate description system；It is difficult to solve determining for signal characteristic
Measure evaluation problem.And the representative property of point shape is can be defined with simple method with self-similarity, and generated by iterating,
The wave character of signal can be described effectively.Common fractal dimension mainly has box counting dimension, Higuchi dimension, Katz FRACTAL DIMENSION
Several and Multifractal Dimension etc..These algorithms largely advance the development of digital processing field.With science and technology
Be constantly progressive, environment becomes increasingly complex, and the complexity of signal of communication is increasing, how to efficiently extract the feature of signal
It is the hot spot studied now.
Summary of the invention
The purpose of the present invention is: realize the validity feature extraction under low signal-to-noise ratio environment to signal.
In order to achieve the above object, the technical solution of the present invention is to provide a kind of new signal based on fractal dimension is special
Levy extracting method, which is characterized in that extract box counting dimension, Higuchi fractal dimension, Katz fractal dimension and the improvement of signal respectively
Box counting dimension, in which: improve the extracting method of box counting dimension the following steps are included:
Original signal x (i) is converted to obtain new signal y (i) by Hilbert by step 1；
Step 2, the instantaneous amplitude A (i) that original signal x (i) is found out using the real and imaginary parts of new signal y (i), to instantaneous
Amplitude A (i) is normalized to obtain instant amplitude value a (i)；
Step 3, kurtosis reconciliation parameter Q (i) for calculating original signal x (i):
In formula, N indicates the data length of original signal x (i)；
Step 4 calculates harmonic average box counting dimension K:
In formula, Db(i) box counting dimension for being instant amplitude value a (i).
Preferably, the expression formula of instantaneous amplitude A (i) described in step 2 are as follows:
In formula, Re y (i) indicates the real part of new signal y (i), Im y (i)
Indicate the imaginary part of new signal y (i).
Preferably, the expression formula of instant amplitude value a (i) described in step 2 are as follows:
In formula, MAX (A (i)) indicates the maximum value in instantaneous amplitude A (i).
In existing one dimensional fractal algorithm, each algorithm all respectively has shortcoming, such as Higuchi Cancers Fractional Dimension Feature algorithm
Calculated value is more accurate, but stability is very poor.Katz algorithm stability is slightly good, but and does not have good recognition capability.
Generally, the classifying quality of box counting dimension is preferable, and calculation amount is simple, simulation time is short.But since signal is increasingly complicated, tradition
Box counting dimension face sophisticated signal when still will appear part signal overlapping the case where.Technical solution proposed by the invention exists
It is improved on original box counting dimension, proposes two new parameters, and combine traditional algorithm, form four dimensions feature, reach one
Dimension divides the purpose for carrying out feature extraction to complex modulated signal under shape and identifying.
The present invention proposes a kind of new based on box counting dimension to improve feature extraction algorithm to the anti-interference ability of noise
Modulated signal feature extraction algorithm, propose two new parameters: kurtosis reconciles parameter and harmonic average box counting dimension, and with tradition
Box counting dimension, Higuchi fractal dimension, Katz fractal dimension collectively form four-dimensional Cancers Fractional Dimension Feature vector, and then realize to letter
Number multidimensional fractal characteristic extract.The present invention calculates simply, can identify that this is fractal dimension to complicated modulated signal
Certain theoretical foundation is provided in the application of field of signal identification.
Detailed description of the invention
Fig. 1 is the lower 6 kinds of signals box counting dimension characteristic profile of -10dB~20dB signal-to-noise ratio；
Fig. 2 is the lower 6 kinds of signals Higuchi Dimension Characteristics distribution map of -10dB~20dB signal-to-noise ratio；
Fig. 3 is the lower 6 kinds of signals Katz Dimension Characteristics distribution map of -10dB~20dB signal-to-noise ratio；
Fig. 4 is that the lower 6 kinds of signals of -10dB~20dB signal-to-noise ratio improve back box Dimension Characteristics distribution map.
Specific embodiment
Present invention will be further explained below with reference to specific examples.It should be understood that these embodiments are merely to illustrate the present invention
Rather than it limits the scope of the invention.In addition, it should also be understood that, after reading the content taught by the present invention, those skilled in the art
Member can make various changes or modifications the present invention, and such equivalent forms equally fall within the application the appended claims and limited
Range.
The present invention provides a kind of new signal characteristic extracting methods based on fractal dimension, extract the box dimension of signal respectively
Number, Higuchi fractal dimension, Katz fractal dimension and improvement box counting dimension.
Wherein, extract the box counting dimension of signal the following steps are included:
1. couple original signal x (i) carries out discrete processes, minimum sampling interval is set as ε.
2. calculating the range scale s (k ε) of signal longitudinal coordinate:
In above formula, N0For sampling number, k ε indicates the side length of the box of the different scale of covering signal, S1It indicates to intercept certain
The maximum value of one segment signal sampled point, S2Indicate the minimum value of a certain segment signal sampled point of interception.
All sampling interval collection comprising signal are set as Nkε:
3. selecting matched curve lgk ε~lgNkεThe middle linearity good one section is used as non-scaling section, then: lgNkε=-dBlgkε
+ b, wherein dBIndicate the slope of signal；B indicates the numerical value of matched curve on the y axis；k1≤k≤k2, k1、k2Respectively above-mentioned nothing
The starting point and maximal end point of scaling interval.
4. finally, finding out the slope of this straight line according to the measurement method of least square method, so that it may identification needed for calculating
The box counting dimension D of signal:
Extract signal Higuchi fractal dimension the following steps are included:
1. couple original signal x (i) carries out discrete processes, discrete signal x (n) is obtained.Discrete signal x (n) is recombinated,
Construct k new time seriesesIt is as follows:
In formula, m=1,2,3 ..., k indicate some starting point for the number of winning the confidence, and N indicates the total points of signal.
2. pair curve constituted or time series seek its average length Lm(k):
3. ask overall average length and L (k):
4. total average length L (k) due to discrete-time signal sequence is proportional to scale k, both sides are taken into logarithm simultaneously,
:Wherein, least square method matched curveSlope D be exactly original signal x
(i) Higuchi fractal dimension.
Extract signal Katz fractal dimension the following steps are included:
If original signal x (i) is by series of points (xi, yi) composition, signal length N, then the Katz of original signal x (i) points
Shape dimension D can be obtained by following formula:
In above formula, L is the length of original signal x (i), then L are as follows:
D is initial point (x1, y1) to the maximum distance of other points, then d are as follows:
Improve box counting dimension extracting method the following steps are included:
1. original signal x (i) is converted to obtain new signal y (i) by Hilbert；
2. the instantaneous amplitude A (i) of original signal x (i) is found out using the real and imaginary parts of new signal y (i), to instantaneous amplitude
A (i) is normalized to obtain instant amplitude value a (i).
The expression formula of instantaneous amplitude A (i) are as follows:
In formula, Re y (i) indicates that the real part of new signal y (i), Imy (i) indicate the imaginary part of new signal y (i)
The expression formula of instant amplitude value a (i) are as follows:
In formula, MAX (A (i)) indicates the maximum value in instantaneous amplitude A (i).
3. calculating kurtosis reconciliation parameter Q (i) of original signal x (i):
In formula, N indicates the data length of original signal x (i)；
4. calculating harmonic average box counting dimension K:
In formula, Db(i) box counting dimension for being instant amplitude value a (i).
Claims (3)
1. a kind of new signal characteristic extracting methods based on fractal dimension, which is characterized in that extract respectively signal box counting dimension,
Higuchi fractal dimension, Katz fractal dimension and improvement box counting dimension, in which: the extracting method for improving box counting dimension includes following step
It is rapid:
Original signal x (i) is converted to obtain new signal y (i) by Hilbert by step 1；
Step 2, the instantaneous amplitude A (i) that original signal x (i) is found out using the real and imaginary parts of new signal y (i), to instantaneous amplitude
A (i) is normalized to obtain instant amplitude value a (i)；
Step 3, kurtosis reconciliation parameter Q (i) for calculating original signal x (i):
In formula, N indicates the data length of original signal x (i)；
Step 4 calculates harmonic average box counting dimension K:
In formula, Db(i) box counting dimension for being instant amplitude value a (i).
2. a kind of new signal characteristic extracting methods based on fractal dimension as described in claim 1, which is characterized in that step
The expression formula of instantaneous amplitude A (i) described in 2 are as follows:
In formula, Re y (i) indicates that the real part of new signal y (i), Im y (i) indicate
The imaginary part of new signal y (i).
3. a kind of new signal characteristic extracting methods based on fractal dimension as described in claim 1, which is characterized in that step
The expression formula of instant amplitude value a (i) described in 2 are as follows:
In formula, MAX (A (i)) indicates the maximum value in instantaneous amplitude A (i).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810734013.XA CN109165545A (en) | 2018-07-05 | 2018-07-05 | A kind of new signal characteristic extracting methods based on fractal dimension |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810734013.XA CN109165545A (en) | 2018-07-05 | 2018-07-05 | A kind of new signal characteristic extracting methods based on fractal dimension |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109165545A true CN109165545A (en) | 2019-01-08 |
Family
ID=64897347
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810734013.XA Pending CN109165545A (en) | 2018-07-05 | 2018-07-05 | A kind of new signal characteristic extracting methods based on fractal dimension |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109165545A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110364187A (en) * | 2019-07-03 | 2019-10-22 | 深圳华海尖兵科技有限公司 | A kind of endpoint recognition methods of voice signal and device |
| CN112152731A (en) * | 2020-09-08 | 2020-12-29 | 重庆邮电大学 | Fractal dimension-based unmanned aerial vehicle detection and identification method |
| CN112230270A (en) * | 2020-12-14 | 2021-01-15 | 西南交通大学 | Earthquake early warning method, device, equipment and readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101976442A (en) * | 2010-11-09 | 2011-02-16 | 东华大学 | Method for extracting fractal profile for representing fabric texture and Sobel operator filtering detail mixed characteristic vector |
| JP2011115393A (en) * | 2009-12-03 | 2011-06-16 | Utsunomiya Univ | Skin feature determination system, skin feature determination method, and skin feature determination program |
| CN102930172A (en) * | 2012-11-15 | 2013-02-13 | 江苏科技大学 | Extraction method of multi-scale characteristic and fluctuation parameter of sea wave based on EMD |
| CN103457890A (en) * | 2013-09-03 | 2013-12-18 | 西安电子科技大学 | Method for effectively recognizing digital modulating signals in non-Gaussian noise |
| CN103997475A (en) * | 2014-05-29 | 2014-08-20 | 西安电子科技大学 | Method for recognizing digital modulation signals under Alpha stable distribution noise |
| CN104052702A (en) * | 2014-06-20 | 2014-09-17 | 西安电子科技大学 | Method for identifying digital modulation signals in presence of complicated noise |
-
2018
- 2018-07-05 CN CN201810734013.XA patent/CN109165545A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011115393A (en) * | 2009-12-03 | 2011-06-16 | Utsunomiya Univ | Skin feature determination system, skin feature determination method, and skin feature determination program |
| CN101976442A (en) * | 2010-11-09 | 2011-02-16 | 东华大学 | Method for extracting fractal profile for representing fabric texture and Sobel operator filtering detail mixed characteristic vector |
| CN102930172A (en) * | 2012-11-15 | 2013-02-13 | 江苏科技大学 | Extraction method of multi-scale characteristic and fluctuation parameter of sea wave based on EMD |
| CN103457890A (en) * | 2013-09-03 | 2013-12-18 | 西安电子科技大学 | Method for effectively recognizing digital modulating signals in non-Gaussian noise |
| CN103997475A (en) * | 2014-05-29 | 2014-08-20 | 西安电子科技大学 | Method for recognizing digital modulation signals under Alpha stable distribution noise |
| CN104052702A (en) * | 2014-06-20 | 2014-09-17 | 西安电子科技大学 | Method for identifying digital modulation signals in presence of complicated noise |
Non-Patent Citations (2)
| Title |
|---|
| 龙哓红: ""典型通信信号及生物医学信号的识别与认知技术研究"", 《中国优秀硕士学位论文全文数据库（电子期刊）医药卫生科技辑》 * |
| 龙晓红等: ""基于调和平均分形盒维数的无线通信信号调制识别算法"", 《江苏大学学报（自然科学版）》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110364187A (en) * | 2019-07-03 | 2019-10-22 | 深圳华海尖兵科技有限公司 | A kind of endpoint recognition methods of voice signal and device |
| CN112152731A (en) * | 2020-09-08 | 2020-12-29 | 重庆邮电大学 | Fractal dimension-based unmanned aerial vehicle detection and identification method |
| CN112152731B (en) * | 2020-09-08 | 2023-01-20 | 重庆邮电大学 | Fractal dimension-based unmanned aerial vehicle detection and identification method |
| CN112230270A (en) * | 2020-12-14 | 2021-01-15 | 西南交通大学 | Earthquake early warning method, device, equipment and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Qu et al. | Radar signal intra-pulse modulation recognition based on convolutional neural network | |
| Lee et al. | An implementation of leaf recognition system using leaf vein and shape | |
| CN105809693B (en) | SAR image registration method based on deep neural network | |
| CN109165545A (en) | A kind of new signal characteristic extracting methods based on fractal dimension | |
| CN103942540A (en) | False fingerprint detection algorithm based on curvelet texture analysis and SVM-KNN classification | |
| CN110110738A (en) | A kind of Recognition Method of Radar Emitters based on multi-feature fusion | |
| CN110175560A (en) | A kind of radar signal intra-pulse modulation recognition methods | |
| CN105844279A (en) | Depth learning and SIFT feature-based SAR image change detection method | |
| CN103116763A (en) | Vivo-face detection method based on HSV (hue, saturation, value) color space statistical characteristics | |
| CN109977191A (en) | Problem map detection method, device, electronic equipment and medium | |
| CN112287796B (en) | Radiation source identification method based on VMD-Teager energy operator | |
| CN105405148B (en) | A kind of remote sensing image Chinese white poplar recognition methods of combination trees shadow character | |
| CN109461132A (en) | SAR image automatic registration method based on feature point geometric topological relation | |
| CN107729926A (en) | A kind of data amplification method based on higher dimensional space conversion, mechanical recognition system | |
| CN102194208A (en) | Image falsification detecting and falsification positioning method based on image signature | |
| Jiang | On orientation and anisotropy estimation for online fingerprint authentication | |
| CN105956544A (en) | Remote sensing image road intersection extraction method based on structural index characteristic | |
| CN111104872A (en) | GF-2 image integrity authentication method applying SIFT and SVD perceptual hashing | |
| CN108492288A (en) | The high score satellite image change detecting method of multiple dimensioned stratified sampling based on random forest | |
| CN101814146B (en) | Method for extracting action potential feature based on multiple wavelet transformation | |
| Lv et al. | An algorithm of Iris feature-extracting based on 2D Log-Gabor | |
| CN107526064A (en) | Adaptive LFM modulated parameter estimating methods based on two dimensional character | |
| CN108768563A (en) | A kind of cooperative frequency spectrum sensing method and relevant apparatus | |
| Fu et al. | Recognition of bridge over water in high-resolution remote sensing images | |
| Peiqin et al. | Finger vein recognition algorithm based on optimized GHT |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190108 |
|
| RJ01 | Rejection of invention patent application after publication |