CN1527198A - Automatic fixed-star and galaxy galaxy distinguishing method based on optical spectrum - Google Patents
Automatic fixed-star and galaxy galaxy distinguishing method based on optical spectrum Download PDFInfo
- Publication number
- CN1527198A CN1527198A CNA031201059A CN03120105A CN1527198A CN 1527198 A CN1527198 A CN 1527198A CN A031201059 A CNA031201059 A CN A031201059A CN 03120105 A CN03120105 A CN 03120105A CN 1527198 A CN1527198 A CN 1527198A
- Authority
- CN
- China
- Prior art keywords
- galaxy
- fixed star
- spectrum
- normal
- active
- 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
- 238000001228 spectrum Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 26
- 230000003287 optical effect Effects 0.000 title abstract 2
- 230000003595 spectral effect Effects 0.000 claims abstract description 10
- 230000003068 static effect Effects 0.000 claims abstract description 9
- 230000000694 effects Effects 0.000 claims description 8
- 230000001537 neural effect Effects 0.000 claims description 7
- 238000012847 principal component analysis method Methods 0.000 claims 1
- 238000004088 simulation Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 7
- 239000011159 matrix material Substances 0.000 description 6
- 235000019892 Stellar Nutrition 0.000 description 3
- 238000010606 normalization Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 241000284156 Clerodendrum quadriloculare Species 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
Images
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The system for automatic distinction between fixed star and galaxy based on optical spectrum consists of static spectrum template of fixed stars, normal galaxies and active galaxies, computer and relevant software. The automatic distinction process includes the following steps: automatically distinguishing non-active celestial bodies including fixed stars and normal galaxies from active celestial bodies including active galaxies; and automatically distinguishing fixed star from normal galaxies. The automatic spectral celestial body distinguishing technology of the present invention has correct rate up to 96% for fixed stars and 94%for normal galaxies.
Description
Technical field
The present invention relates to computer assisted celestial body recognition methods.
Background technology
The automatic identification of fixed star and galaxy has significance to the large-scale telescope project of touring the heavens.Be that PSF (Point SpreadFunction) detection by photographic data is finished to the automatic identification of fixed star and galaxy in the world at present.Can allow the large-scale telescope spectrum project of touring the heavens carry out camera observation based on the fixed star of spectrum and the automatic identification technology of galaxy to target, therefore significant.In the research of spectrum automatic identification technology, because the normally very dark unknown celestial body of scientific goal of the plan of touring the heavens, therefore so low and also the unknown of red shift parameter of the spectral signal-noise ratio that obtains is to the technical certain degree of difficulty that has of being identified in automatically of fixed star under spectral red shift the unknown, the low signal-to-noise ratio situation and galaxy.Most in the world automatic identification technologies that relate to galaxy spectrum all are to launch research under the known prerequisite of red shift value, so are not suitable for red shift value condition of unknown.
Summary of the invention
The object of the present invention is to provide a kind of being applicable under the unknown of spectral red shift value, the low signal-to-noise ratio situation, have high accuracy based on the fixed star of spectrum and the automatic identifying method of galaxy.
For achieving the above object, a kind of automatic identifying method based on spectrum fixed star and galaxy is made up of static spectrum template, computing machine and the related software of fixed star, normal galaxy and active galaxy, and described method comprises step:
Automatically discern non-active object and active objects, described non-active object comprises fixed star and normal galaxy, and described active objects comprises active galaxy;
Automatically discern fixed star and normal galaxy, thereby realize the automatic identification of fixed star and galaxy.
The classification accuracy rate of celestial body spectrum automatic identification technology of the present invention can reach 96% to fixed star, can reach 94% to normal galaxy, can reach 94% to active galaxy.
Description of drawings
Fig. 1 is based on the fixed star of spectrum and the automatic identification process synoptic diagram of galaxy;
The automatic identification process figure of activity of Fig. 2 right and wrong and active objects;
Fig. 3 is the automatic identification process figure of fixed star and normal galaxy;
Fig. 4 is the structural drawing of combined radial base neural net.
Embodiment
The present invention is made up of static spectrum template, computing machine and the related software of fixed star, normal galaxy, active galaxy.Standard fixed star template includes seven big spectral types: i.e. the spectrum of O, B, A, F, G, K, M; The static template of normal galaxy should comprise E, S0, Sa, Sb type; The static template of active galaxy should comprise: starburst galaxy SB1, SB2, SB3, SB4, SB5, SB6, Sc.Comprise two big steps based on the fixed star of spectrum and the automatic identification technology of galaxy (comprising normal galaxy, active galaxy), as shown in Figure 1.The first step is finished the automatic identification of ground floor among Fig. 1, the identification (as shown in Figure 2) between promptly non-activity and the active objects, and non-active object comprises fixed star and normal galaxy, active objects comprises active galaxy; Second step was finished the automatic identification of second layer left part, promptly fixed star and normal galaxy was discerned (as shown in Figure 3), thereby realized the automatic identification of fixed star and galaxy.The implementation method of the first step mainly comprises data normalization, adopts principal component method structure three-dimensional feature space, selects training sample, training sample projection, adopts the optimal classification face of supporting vector machine searching optimal classification face, sample data standardization to be identified, projection, application supporting vector machine to be identified as non-activity or active objects.The implementation method in second step mainly comprises data normalization, adopts principal component method structure dimensionality reduction feature space, selects training sample, training sample projection, the sorter that adopts the combined radial base neural net to carry out sorter training, sample data standardization to be identified, projection, application combination radial base neural net are identified as fixed star or normal galaxy.Specifying of each step is as follows:
1, standardization of data
The dimension or the amplitude of variation difference of M the row variable of data sample matrix X to be identified, its order of magnitude may differ manyfold.Therefore, in order to eliminate the different influences that bring with amplitude of variation of dimension, raw data should be done standardization.
2, adopt the three-dimensional feature space of principal component method structure celestial body spectrum
Select the spectrum template of fixed star template and galaxy to carry out principal component analysis.Step is as follows: suppose the spectrum samples composition data matrix X that chooses, X calculates covariance matrix D=X after the 1st step standardization
TX, eigenvalue and the eigenvector matrix of calculating covariance matrix D; Its eigenvalue is arranged from big to small, chooses the eigenvector of first three eigenvalue correspondence, i.e. principal component, structural attitude matrix E
M * 3, obtain three-dimensional feature space.Spectrum samples x to be identified
1 * MAfter the 1st step standardization, project to this three-dimensional feature space by xE, obtain three-dimensional unique point.
3, select training sample
Training sample has: the stellar spectrum sample, comprising the fixed star of seven big spectral types; The galaxy spectrum samples, the simulated spectra sample comprising normal galaxy and the static template of active galaxy through obtaining after the red shift.
4, adopt supporting vector machine to seek the optimal classification face.
The training sample that the 3rd step was chosen goes on foot the resulting three-dimensional feature point of projection through the 1st step standardization and the 2nd, adopts supporting vector machine, seeks the optimal classification surface function of non-active object and active objects.The three-dimensional feature point of sample to be identified can be identified as non-activity or active objects by this classifying face function.
5, adopt principal component method, the dimensionality reduction feature space of structure fixed star and normal galaxy
The static template of stellar spectrum that comprises seven big spectral types and normal galaxy is carried out principal component analysis, and step and the 2nd step are similar, different is choose here before n dimension (the dimensionality reduction eigenmatrix H of the eigenvector structure n dimension of the eigenvalue correspondence of n＜M)
M * nAnd feature space.Sample to be identified projects in this dimensionality reduction feature space through the 1st step standardization, obtains the characteristic variable of dimensionality reduction.
6, select training sample
Training sample has: the stellar spectrum sample, comprising the fixed star of seven big spectral types; The normal galaxy spectrum samples is comprising the simulated spectra sample of four kinds of static templates through obtaining after the red shift.
7, adopt combined radial base neural net training classifier
The training sample that the 6th step was chosen goes on foot standardization, projection on the dimensionality reduction feature space in the 5th step through the 1st, obtain the characteristic variable of dimensionality reduction, then they inputs as the combined radial base neural net, carry out the training of sorter, the output bidimensional, expression belongs to the probability of fixed star and normal galaxy respectively.
8, the classification of active objects and non-active object
Sample to be identified through the standardization in the 1st step, project on the three-dimensional feature space in the 2nd step, obtain three-dimensional unique point, as the input of the optimal classification surface function in the 4th step, if the output that obtains is greater than 0, then being non-active object, if less than 0, then is active objects.
9, the classification of fixed star and normal galaxy
Go on foot the spectrum of the non-active object that identifies through standardization, projection on the dimensionality reduction feature space in the 5th step to the 8th, obtain of the input of the characteristic variable of dimensionality reduction as the 7th step sorter, according to the probable value of sorter output, select the big one dimension of probability, differentiation is fixed star or normal galaxy.
Claims (9)
1. automatic identifying method based on spectrum fixed star and galaxy, by fixed star, normal galaxy,
Form with static spectrum template, computing machine and the related software of active galaxy, described method comprises step:
Automatically discern non-active object and active objects, described non-active object comprises fixed star and normal galaxy, and described active objects comprises active galaxy;
Automatically discern fixed star and normal galaxy, thereby realize the automatic identification of fixed star and galaxy.
2. by the described method of claim 1, it is characterized in that described automatic identification non-active object and active objects comprise step:
1) standardization of sample data;
2) the three-dimensional principal component feature space of structure celestial body spectrum, and sample carried out projection;
3) select training sample;
4) adopt supporting vector machine, training sample through the 1st step standardization, project on the three-dimensional principal component space in the 2nd step, the three-dimensional feature point that projection is obtained is trained the optimal classification face between non-activity and the active objects as importing again.
3. by the described method of claim 1, it is characterized in that described automatic identification fixed star and normal galaxy comprise step:
1) standardization of sample data;
2) adopt principal component analysis method, fixed star and normal galaxy spectrum are carried out dimensionality reduction, obtain the feature space of dimensionality reduction;
3) select training sample;
4) adopt the combined radial base neural net, training sample is projected on the feature space that the 2nd step obtained as input, carry out the sorter of fixed star and normal galaxy and train;
5) spectrum samples to be identified through the 1st step standardization, project on the 2nd three-dimensional principal component space that obtains of step, again the unique point after the projection as input, and utilize the classifying face function to be identified as non-activity or active objects;
6) to the spectrum samples of the 5th non-active object of step in the classification results through the 1st step standardization, project on the dimensionality reduction feature space that the 2nd step obtained, the characteristic variable after the projection is input to is identified as fixed star or normal galaxy in the sorter.
4, described by claim 1 based on the fixed star of spectrum and the automatic identifying method of galaxy, it is characterized in that: carry out the identification of non-active object and active objects (active galaxy) earlier, carry out the identification of fixed star and normal galaxy again.
5, described by claim 2 based on the fixed star of spectrum and the automatic identifying method of galaxy, it is characterized in that: in the three-dimensional principal component feature space of described celestial body spectrum, adopt the supporting vector machine algorithm to discern the spectrum of non-activity and active objects.
6, described by claim 3 based on the fixed star of spectrum and the automatic identifying method of galaxy, it is characterized in that:, the sample point in the feature space as characteristic variable, is trained the sorter of fixed star and normal galaxy at the dimensionality reduction feature space of non-active object spectrum.
7, described by claim 3 based on the fixed star of spectrum and the automatic identifying method of galaxy, it is characterized in that: adopt the combined radial base neural net, the sorter of training fixed star and normal galaxy.
8, described by claim 3 based on the fixed star of spectrum and the automatic identifying method of galaxy, it is characterized in that: the simulated spectra of the galaxy after the training sample of the optimal classification face of non-activity and active objects comprises the fixed star of various spectral types and carries out the red shift simulation.
9, by claim 6 or 7 described, it is characterized in that: the simulated spectra that the training sample of the sorter of fixed star and normal galaxy comprises the fixed star of various spectral types and carries out the normal galaxy after red shift is simulated based on the fixed star of spectrum and the automatic identifying method of galaxy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA031201059A CN1527198A (en) | 2003-03-07 | 2003-03-07 | Automatic fixed-star and galaxy galaxy distinguishing method based on optical spectrum |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA031201059A CN1527198A (en) | 2003-03-07 | 2003-03-07 | Automatic fixed-star and galaxy galaxy distinguishing method based on optical spectrum |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1527198A true CN1527198A (en) | 2004-09-08 |
Family
ID=34285322
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA031201059A Pending CN1527198A (en) | 2003-03-07 | 2003-03-07 | Automatic fixed-star and galaxy galaxy distinguishing method based on optical spectrum |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1527198A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100412889C (en) * | 2006-04-17 | 2008-08-20 | 中国科学院自动化研究所 | Special star automatic recognition method based on wavelet character |
| CN100487395C (en) * | 2005-05-26 | 2009-05-13 | 中国科学院自动化研究所 | Astronomical spectrum automatic sorting and red shift measuring method based on similarity measure |
| CN109272027A (en) * | 2018-08-30 | 2019-01-25 | 昆明理工大学 | A kind of pulsar candidate's body recognition methods based on artificial neural network ensembles |
-
2003
- 2003-03-07 CN CNA031201059A patent/CN1527198A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100487395C (en) * | 2005-05-26 | 2009-05-13 | 中国科学院自动化研究所 | Astronomical spectrum automatic sorting and red shift measuring method based on similarity measure |
| CN100412889C (en) * | 2006-04-17 | 2008-08-20 | 中国科学院自动化研究所 | Special star automatic recognition method based on wavelet character |
| CN109272027A (en) * | 2018-08-30 | 2019-01-25 | 昆明理工大学 | A kind of pulsar candidate's body recognition methods based on artificial neural network ensembles |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108647585B (en) | Traffic identifier detection method based on multi-scale circulation attention network | |
| CN112446388A (en) | Multi-category vegetable seedling identification method and system based on lightweight two-stage detection model | |
| CN108108751B (en) | Scene recognition method based on convolution multi-feature and deep random forest | |
| CN111062282A (en) | Transformer substation pointer type instrument identification method based on improved YOLOV3 model | |
| CN102982349B (en) | A kind of image-recognizing method and device | |
| CN110490174A (en) | Multiple dimensioned pedestrian detection method based on Fusion Features | |
| Younis et al. | Detection and annotation of plant organs from digitised herbarium scans using deep learning | |
| CN112232371A (en) | American license plate recognition method based on YOLOv3 and text recognition | |
| CN114170512A (en) | Remote sensing SAR target detection method based on combination of network pruning and parameter quantification | |
| CN110503098A (en) | A kind of object detection method and equipment of quick real-time lightweight | |
| CN112784487B (en) | Flight action recognition method and device | |
| Dharwadkar et al. | Floriculture classification using simple neural network and deep learning | |
| US20040091144A1 (en) | Automatic encoding of a complex system architecture in a pattern recognition classifier | |
| CN1527198A (en) | Automatic fixed-star and galaxy galaxy distinguishing method based on optical spectrum | |
| CN117710841A (en) | Small target detection method and device for aerial image of unmanned aerial vehicle | |
| US7113636B2 (en) | Method and computer program product for generating training data for a new class in a pattern recognition classifier | |
| CN107941210A (en) | A kind of method for recognising star map of combination nerual network technique and triangle algorithm | |
| CN115206455B (en) | Deep neural network-based rare earth element component content prediction method and system | |
| US20040042665A1 (en) | Method and computer program product for automatically establishing a classifiction system architecture | |
| CN113392695B (en) | Highway truck and wheel axle identification method thereof | |
| CN114782709A (en) | Image small target detection method and system based on Gaussian distribution strategy | |
| CN116758363A (en) | Weight self-adaption and task decoupling rotary target detector | |
| CN109887005B (en) | TLD target tracking method based on visual attention mechanism | |
| Hussain et al. | Intra-class recognition of fruits using dcnn for commercial trace back-system | |
| Wei et al. | A visual positioning method for wedge support robot based on Pruned-Ghost-YOLOv3 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |