CN102798384A - Ocean remote sensing image water color and water temperature monitoring method based on compression sampling - Google Patents
Ocean remote sensing image water color and water temperature monitoring method based on compression sampling Download PDFInfo
- Publication number
- CN102798384A CN102798384A CN2012102277857A CN201210227785A CN102798384A CN 102798384 A CN102798384 A CN 102798384A CN 2012102277857 A CN2012102277857 A CN 2012102277857A CN 201210227785 A CN201210227785 A CN 201210227785A CN 102798384 A CN102798384 A CN 102798384A
- Authority
- CN
- China
- Prior art keywords
- water temperature
- data
- remote sensing
- data set
- water
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses an ocean remote sensing image water color and water temperature monitoring method based on compression sampling. The monitoring method comprises the following steps of: obtaining original remote sensing data of ocean water color or water temperature to be detected, and carrying out compression sampling, sparse transformation and image processing on the original remote sensing data to obtain a water color or water temperature data set; sequentially conducting denoising and reconstruction processing on the water color or water temperature data set through an SOM (self-organized mapping) algorithm to obtain a processed water color or water temperature data set; conducting denoising processing on the processed water color or water temperature data set and reconstructing missing image data by utilizing the continuous interpolation property; and outputting and displaying the reconstructed missing image data. The monitoring method utilizes the compression sampling principle for reducing the amount of processed data, and utilizes the advantages of nonlinear estimation of the SOM algorithm as well as linear estimation and continuous interpolation of an improved empirical orthogonal decomposition (EOF) algorithm and the like so as to improve the efficiency of reconstructing missing images, expand the scope of the reconstructed images, and enable the precision and efficiency of ocean remote sensing water color and water temperature monitoring to be high.
Description
Technical field
The invention belongs to the marine monitoring field, particularly a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling.
Background technology
Growing along with science and technology, the ocean remote sensing technology has obtained significant progress.Application as the satellite ocean remote sensing; The use of ocean water colour water temperature remote sensing is more outstanding, but receives the influence of objective factors such as weather random variation and thin cloud detection algorithm be not ideal enough, and the disappearance of data has brought very big restriction not only for the ocean research that uses the remote sensing product and vocational work etc.; Again because seasat remotely-sensed data amount is very big; Make the application of remote sensing product receive further restriction, what is more, the disappearance of data makes the monitoring area of ocean essential significantly reduce.Though some application can be tolerated the unusual and disappearance of partial data, along with development of science and technology, many departments are increasing to the demand of complete remotely-sensed data collection, and the dynamics of marine monitoring is continued to increase.
The inventor finds to exist at least in the prior art following shortcoming and defect in realizing process of the present invention:
Monitoring method reconstruct disappearance image efficient of the prior art is low, and is harsh to the requirement of reconstruct object, has limitation.
Summary of the invention
The invention provides a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling, this method has improved the efficient of reconstruct disappearance image, has enlarged the scope of reconstructed image, sees hereinafter for details and describes:
A kind of ocean remote sensing image water colour water temperature detection method based on compression sampling said method comprising the steps of:
(1) obtains ocean water colour to be detected or the original remotely-sensed data of water temperature, original remotely-sensed data is carried out sparse conversion of compression sampling and Flame Image Process, obtain water colour water temperature data set;
(2) through the SOM method said water colour water temperature data set is carried out denoising and reconstruction processing successively, obtain and handle back water colour water temperature data set;
(3) said processing back water colour water temperature data set is carried out denoising, and utilize its continuous interpolation characteristic reconstruct disappearance view data;
(4) export and show said disappearance view data.
Said method also comprises:
When including land data and island data in the said original remotely-sensed data, reject said land data and said island data.
Saidly said water colour water temperature data set is carried out denoising and reconstruction processing successively, obtains and handle back water colour water temperature data set and be specially through the SOM method:
Nonlinear organization through said SOM method carries out non-linear interpolation reconstruct and denoising to said water colour water temperature data set, obtains said processing back water colour water temperature data set.
Said said processing back water colour water temperature data set is carried out denoising, and utilizes continuous interpolation characteristic reconstruct disappearance view data to be specially:
1) said processing back water colour water temperature data set is carried out self-adaptation EOF and decompose, obtain space mode matrix U, singular value matrix S and time modal matrix V;
2) utilize the Monte Carlo to intersect calibration set and confirm that best reconstruct mode counts P;
3) count P, said space mode matrix U, said singular value matrix S and said time modal matrix V through said best reconstruct mode and make up said disappearance view data X.
Saidly utilize the Monte Carlo to intersect calibration set to confirm that best reconstruct mode counts P and be specially:
If the intersection correction error when the mode number is P and P+1 is respectively R
PAnd R
P+1, Ratio was compared in the reduction of mode error when definition mode number was P
P=(R
P-1-R
P)/R
P, work as Ratio
PLess than assign thresholds and R
P<r
P+1The time, stop the selection of best reconstruct mode number, current mode is counted P as best reconstruct mode number, wherein, R
P-1Intersection correction error when being P-1 for the mode number.
Said method also comprises: confirm that each said best reconstruct mode counts the best iterations of P, be no more than maximum iteration time N.
The said iterations of confirming that said best reconstruct mode is counted P is specially:
Define said maximum iteration time N and minimum cross validation error, whether the difference of judging next iteration error and current iteration error is less than said minimum cross validation error, if stop iteration; Otherwise, continue to carry out iterative processing, when iterations is that said maximum iteration time N stops iteration.
Said disappearance view data X is specially:
The beneficial effect of technical scheme provided by the invention is: need handle bulk redundancy information and the low shortcoming of treatment effeciency when this method has overcome processing remote sensing water colour water temperature image in the past; Overcome and adopted traditional optimum interpolation method to need the shortcoming of the prior imformation of remotely-sensed data collection in the past; Introduced this field to the SOM algorithm; Solve the deficiency of the emerging EOF algorithm in this field, and proposed the monitoring that the SOM-EOF algorithm has better been realized ocean essential.This method can utilize the compression sampling principle to reduce the data volume of processing; Advantages such as Linear Estimation that the nonlinear estimation of comprehensive utilization self-organization mapping SOM algorithm and follow-on empirical orthogonal are decomposed the EOF algorithm and continuous interpolation; Make the objective inherent characteristics that the reconstruct effect of data set more reflected data set, improved the efficient of reconstruct disappearance image, enlarged the scope of reconstructed image; And then making that the precision of ocean remote sensing water colour water temperature key element monitoring is higher, efficient is higher.
Description of drawings
Fig. 1 is the synoptic diagram of a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling provided by the invention;
Fig. 2 is the pretreated synoptic diagram of data provided by the invention;
Fig. 3 is the synoptic diagram of SOM algorithm provided by the invention;
Fig. 4 is the synoptic diagram of modified EOF algorithm provided by the invention;
Fig. 5 a is the SST image in 2010 provided by the invention August 1;
Fig. 5 b is the SST image in 2010 provided by the invention September 21;
Fig. 5 c is the SST image in 2010 provided by the invention October 27;
Fig. 5 d is the image after Fig. 5 a reconstruct;
Fig. 5 e is the image after Fig. 5 b reconstruct;
Fig. 5 f is the image after Fig. 5 c reconstruct;
Fig. 6 a is the SST image in 2010 provided by the invention August 17;
Fig. 6 b is the SST image in 2010 provided by the invention September 17;
Fig. 6 c is the SST image in 2010 provided by the invention September 29;
Fig. 6 d is the image after Fig. 6 a reconstruct;
Fig. 6 e is the image after Fig. 6 b reconstruct;
Fig. 6 f is the image after Fig. 6 c reconstruct.
Embodiment
For making the object of the invention, technical scheme and advantage clearer, will combine accompanying drawing that embodiment of the present invention is done to describe in detail further below.
In order to improve the efficient of reconstruct disappearance image, enlarge the scope of reconstructed image, the embodiment of the invention provides a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling, referring to Fig. 1, Fig. 2, Fig. 3, Fig. 4, Fig. 5 and Fig. 6, sees hereinafter for details and describes:
101: obtain ocean water colour to be detected or the original remotely-sensed data of water temperature, original remotely-sensed data is carried out sparse conversion of compression sampling and Flame Image Process, obtain water colour water temperature data set;
Wherein, This step is specially: download ocean water colour or water temperature remote sensing images raw data to be detected from Relational database; The compressing data bag decompresses; After all decompress(ion) finishes all view data, calculate the data that comprise in each image, the satellite data of this moment still is a magnanimity.Utilize the compression sampling principle all to carry out sparse conversion to each view data again, and then make data volume significantly reduce; Because there is deviation in the latitude and longitude coordinates of the data value of image series, also need utilize image processing software that data are carried out gridding and operate calibration coordinate.
Wherein, The embodiment of the invention adopts seadas software to derive the data that each opens image, and adopts surfer software that data are carried out gridding and operate calibration coordinate, when specifically realizing; Can also adopt other disposal route, the embodiment of the invention does not limit this.
Further, in order to reduce reconstructed error, this method also comprises: when including land data and island data in the original remotely-sensed data, reject land data and island data.
Concrete elimination method is that those skilled in the art are known, and the embodiment of the invention repeats no more at this.For example: when only including the land data in the original remotely-sensed data, utilize ArcGIS software to make the land template of survey region, reject the land part in image every day, just constituted the data set of the water colour water temperature that can directly handle this moment.
102: through the SOM method water colour water temperature data set is carried out denoising and reconstruction processing successively, obtain and handle back water colour water temperature data set;
Wherein, this step is specially: the nonlinear organization through the SOM method carries out non-linear interpolation reconstruct and denoising to water colour water temperature data set, obtains and handles back water colour water temperature data set.
Consider the network research of SOM method and use maximum still two-dimensional network arrays; With and ripe good characteristic; This method has been used the network structure of a two dimension, has proposed first to be applied to the monitoring of ocean remote sensing water colour water temperature key element to the SOM method, has obtained good effect.The SOM method has according to data intrinsic characteristic reconstruct water colour water temperature remotely-sensed data effectively; Embody the nonlinear organization of remotely-sensed data collection; The advantages such as continuity that keep data structure; Can accomplish the monitoring to the ocean essential of survey region, the particular content of SOM method is referring to document 1 and 2, and the embodiment of the invention repeats no more at this.
103: carry out denoising to handling back water colour water temperature data set, and utilize continuous interpolation characteristic reconstruct disappearance view data;
This method has been utilized Alvera-Azcarate
[3-5]DINEOF, but improved this method.To the inefficient problem of svd SVD large matrix in the EOF method implementation, this method is introduced the lanczos operator
[6]Quicken the SVD decomposable process.The problem of the iterative process efficient that is directed against the EOF method again low (general predefined iterations all uses often than reality), this method is used improved iteration convergence criterion.Utilize modified EOF algorithm to the further denoising of data set, utilize its continuous linear interpolation characteristic reconstruct missing data effectively at last, as shown in Figure 4.The Linear Estimation that the EOF algorithm is to use svd SVD to carry out to data set; The nonlinear organization that can not reflect data set; And the EOF algorithm is relatively more responsive to the initialization of data set; But the space dimensionality of its estimation can be high as importing data, and this algorithm or continuous estimation.But this patent has proposed follow-on EOF algorithm on the basis of having inherited the EOF algorithm, overcome the shortcoming of original algorithm; And united the SOM algorithm; So the SOM-EOF algorithm that patent is taked can be maximized favourable factors and minimized unfavourable ones, finally make system better to the monitoring effect of ocean essential.Wherein, The nonlinear characteristic of SOM algorithm to the raw data denoising after; Its operation result during as modified EOF algorithm process to the initialization of data set; Utilize the EOF algorithm to the further denoising of data set again, utilize its continuous interpolation characteristic reconstruct missing data effectively at last, reached the purpose of ocean essential monitoring.Fig. 5 has provided the original satellite image (SST and CHL) of three secondary different disappearance ratios and the result after the processing respectively with Fig. 6.The EOF algorithm is having the spatial distribution characteristic of observation site (website) strict restriction in addition not, is the very high signal dimensionality reduction technology of a kind of efficient; Characterize the complicated original physical vector field of spatial and temporal distributions with a spot of orthogonal function uncorrelated and mutually orthogonal mutually, be very beneficial for statistical fluctuation the physical vector field; Empirical Orthogonal Function is faster than other orthogonal function convergent speed; System with a complicacy; Resolve into the linear combination of the time and space characteristic mode of limited a plurality of quadratures each other; Be very beneficial for explaining to have the advantages such as signal that the time changes the space distribution of composition from pure physical significance; Can well accomplish monitoring to the ocean essential of survey region, but the place that exists some to have much room for improvement too, and then improve the monitoring accuracy of ocean essential reconstruct; These deficiencies are embodied in the following aspects: (1) owing to be the linear interpolation mode, the EOF algorithm can not reflect the nonlinear organization of data set.(2) can not handle the high image of miss ratio.When the image pre-service; Nearly all can select to reject in advance signal to noise ratio (S/N ratio) than higher and data miss ratio than higher image; That is the data of treating reconstruct can not be very little, otherwise the just simple average of original remotely-sensed data collection of using of the reconstruction result of disallowable image substitutes, and it is objective not meet.(3) the initialization sensitive question of EOF algorithm.The EOF algorithm can not directly be used to contain the signal of missing data, and missing values is initialization at first, and what no matter initial value was selected is the average of time dimension, the average of space dimension, and linear interpolation, or the average of signal all can be brought the subjectivity problem.Because signal missing data and the data in the experience calibration set of Monte Carlo originally all use the null value that can make raw data set not have inclined to one side estimation to replace; The similar problem of issuable singular value when this also will cause singular value SVD to decompose; If the data miss ratio of raw data set is relatively low certainly; Its influence can be lower, yet if the miss ratio of data than higher, the susceptibility of the initialization problem of EOF will be than higher; And then cause the precision of reconstruct lower, directly influence the validity and the confidence level of algorithm.Be initialized as the influence of noise original in null value and the signal in addition to EOF; These all can inevitably be interpreted into useful signal, will cause singular value to increase, and main characteristic mode still less; The information of abandoning is more, and then makes the monitoring error of reconstruct increase.(4) as the Problem of Objectivity of the selection of the intersection calibration set of reconstruction accuracy standard.The optimum reconstruct number of EOFs is highstrung to the data of selecteed Monte Carlo experience calibration set, and the cloud region covered is big more, that is raw data disappearance ratio is big more, and the reconstruct number of used EOFs is just few more.This expression only has the EOF pattern of confirming of fully many large scale characteristics to be come out by reconstruct reliably, and the EOF of high-order (represent small scale and among a small circle) can not be come out by the size estimation of given specific cloud.That is to say that the mesoscale characteristics that cloud covers is sampled from beginning to hang down, reconstruct reliably, the optimum number of EOFs has included only several topmost EOFs, has abandoned most mesoscale information.And be incorporated into and intersect the data of calibration set and possibly contain contaminated data, this will cause whole data set also to be polluted.The efficiency of the program run when (5) algorithm is carried out.Before carrying out, program to set the number k of a maximum characteristic mode
Max(this number must be bigger than theoretic best reconstruct mode number), but the number k of maximum characteristic mode
MaxRelevant with the time dimension and the spatial diversity of raw data set, there is not unified standard yet, more be the experience that relies on the researcher, thus confirm that in advance a reliable numerical value is difficult to, if the k that selects
MaxBigger, can cause increasing sharply of program execution time, if the number k of the maximum characteristic mode of selecting
MaxLess, clearly do not reach the little effect of set reconstructed error.And be absorbed in infinite loop for fear of program, and before carrying out, program can set a definite maximum recurrence times N, think that promptly program had just restrained when the round-robin iterations reached maximum number N each time; Yet the selection of N is a difficult thing, and it is big more that N selects; Program expends time in many more; In fact, not that iterations is big more, reconstruction accuracy is just good more; It is more little that N selects, and is difficult to the convergence of the program that guarantees again.When confirming iterations, slow if graph of errors descends, although and error reduces smallerly, the cross validation error is big, has just stopped iteration, this can cause reconstruction accuracy to descend undoubtedly.
Wherein, this step is specially:
1) carries out self-adaptation EOF decomposition to handling back water colour water temperature data set, obtain space mode matrix U, singular value matrix S and time modal matrix V;
2) utilize the Monte Carlo to intersect calibration set and confirm that best reconstruct mode counts P;
Wherein, this step is specially: the intersection correction error when establishing the mode number and being P and P+1 is respectively R
PAnd R
P+1, Ratio was compared in the reduction of mode error when definition mode number was P
P=(R
P-1-R
P)/R
P, work as Ratio
PLess than threshold value and R
P<r
P+1The time (embodiment of the invention is that example describes with 0.01; During concrete the realization; The embodiment of the invention does not limit this), stop the selection of best reconstruct mode number, select current mode to count P as best reconstruct mode number; Just do not need to have calculated the situation of other mode numbers again, improve executing efficiency with this
[8], wherein, R
P-1Intersection correction error when being P-1 for the mode number.
And possibly bring inefficient problem for the maximum iteration time N that chooses in advance, in order to improve operational efficiency, this method also comprises: confirm that best reconstruct mode counts the iterations of P, this step is specially:
Definition maximum iteration time N and minimum cross validation error, whether the difference of judging next iteration error and current iteration error is less than minimum cross validation error, if stop iteration; Otherwise, when iterations equals maximum iteration time N, stop iteration.
Wherein, the value of maximum iteration time and minimum cross validation error is set according to the needs in the practical application, and when specifically realizing, the embodiment of the invention does not limit this.
When confirming iterations, the absolute value that differs and the error reduction that define adjacent cross validation error combine the method for advantage to control best mode than both.In addition, in the process of carrying out, introduced the lanczos operator
[6]Quicken the decomposition of matrix, further improved program efficiency.
3) count P, space mode matrix U, singular value matrix S and time modal matrix V through best mode and make up disappearance view data X.
Wherein, disappearance view data
is the mean value of the valid data in the original remotely-sensed data.Valid data are the non-missing data in the raw data matrix; The unified in advance NaN mark of using of missing data; Those do not have the data that existed of mark to be valid data, and this method is exactly to utilize the time and space correlativity of data, uses these valid data to reconstruct missing data.
104: output also shows the disappearance view data.
The feasibility of a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling of verifying that the embodiment of the invention provides with concrete test below, see hereinafter for details and describe:
Nonlinear characteristic through the SOM method to original remotely-sensed data denoising after; Its operation result during as the EOF algorithm process to the initialization of data set; Utilize the EOF algorithm to the further denoising of data set again; Utilize its continuous interpolation characteristic reconstruct missing data effectively at last, reached the purpose of ocean essential monitoring.Fig. 5 has expressed the original satellite image (SST and CHL) of three secondary different disappearance ratios and the result after the processing respectively with Fig. 6.Through above-mentioned several groups of experiments, can verify the feasibility of this method, through the pre-service of image sequence, through the data of SOM and follow-on EOF algorithm process, reconstruct efficient is high, and the reconstruct scope is wide.
Fig. 5: wherein, figure a is the SST image on August 1st, 2010, and figure b is the SST image on September 21st, 2010, and figure v is the SST image on October 27th, 2010; Figure d schemes e and the image of scheming after f is respectively figure a, figure b and the reconstruct of figure c process SOM-EOF algorithm.Annual this time period in September to October is the process that reduces gradually of entrance of Changjiang River and surrounding waters sea surface temperature thereof just; Can find out from figure e and figure f; Extra large temperature distribution plan after the reconstruct can embody the process of this extra large temperature gradual change well; Simultaneously can keep the grown form in flow field, ocean,, not it can also be seen that the influence of mankind's activity for littoral surrounding waters sea surface temperature because destroyed after the reconstruct.This 3 original images even more noteworthy, the data miss ratio of each width of cloth figure all is quite high, is respectively 42.00% (being that miss ratio is minimum in this experiment), 81.39% and 97.53%.Particularly have large tracts of land when disappearance in the image, especially scheme C, the traditional data interpolation method is difficult to the interpolation that reaches such, or even powerless.And this method can reconstruct image with higher reconstruction accuracy; At first pass through of the initialization of SOM interpolation result to EOF; Extract the physical features of original remotely-sensed data on space and time domain through Empirical Orthogonal Function again,, not only can carry out the reconstruct of effective higher reconstruction accuracy the observation data of disappearance through keeping several best reconstruct mode; The spatial-temporal distribution characteristic that has also kept the remotely-sensed data collection simultaneously is so that to the analysis of data set.And in the overall miss ratio of this experimental data collection up to 83.23%, the lower data set of miss ratio exists under the severe situation of more noises relatively, also can reach such reconstruct effect, and is not easy.
Fig. 6: wherein, figure a is the SST image on August 17th, 2010, and figure b is the SST image on September 17th, 2010, and figure c is the SST image on September 29th, 2010; Figure d schemes e and the image of scheming after f is respectively figure a, figure b and the reconstruct of figure c process SOM-EOF algorithm.The entrance of Changjiang River in annual August to September and the marine environment such as ocean temperature, the potassium that is rich in the water and P elements of surrounding waters thereof; It is the best opportunity that growths such as phytoplankton, suspension microorganism are fit to; Chlorophyll concentration distribution plan after the reconstruct of figure d and figure e can embody well; Simultaneously can keep the grown form in flow field, ocean, not because destroyed after the reconstruct.Scheme c even more noteworthy, miss ratio is up to 90.54%, and traditional interpolation method is difficult to the reconstruct effect that reaches such, or even feels simply helpless.And this method can both successful reconstruct; At first pass through of the initialization of SOM interpolation to EOF; Extract the physical features of raw data on space and time domain through Empirical Orthogonal Function again,, not only can carry out the reconstruct of effective higher reconstruction accuracy the observation data of disappearance through keeping several best reconstruct mode; The spatial-temporal distribution characteristic that has also kept the remotely-sensed data collection simultaneously is so that to the analysis of data set.
In sum; The embodiment of the invention provides a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling; When having overcome processing remote sensing water colour water temperature image in the past, this method need handle bulk redundancy information and the low shortcoming of treatment effeciency; Overcome and adopted traditional optimum interpolation method to need the shortcoming of the prior imformation of remotely-sensed data collection in the past; Introduce this field to the SOM algorithm, solved the deficiency of the emerging EOF algorithm in this field, and proposed the monitoring that the SOM-EOF algorithm has better been realized ocean essential.This method can utilize the compression sampling principle to reduce the data volume of processing; Advantages such as Linear Estimation that the nonlinear estimation of comprehensive utilization self-organization mapping SOM algorithm and follow-on empirical orthogonal are decomposed the EOF algorithm and continuous interpolation; The feasible objective inherent characteristics that the reconstruct effect of data set more reflected data set; And then making that the precision of ocean remote sensing water colour water temperature key element monitoring is higher, efficient is higher.
List of references
[1]Sorjamaa?A.,Merlin?P.,Maillet?B.et.al.SOM+EOF?for?finding?missing?values.European?Symposium?on?Artificial?Neural?Networks?Bruges(Belgium),2007.115~120
[2]Cottrell?M.,Letremy.P.Missing?values:Processing?with?the?kohonen?algorithm.Applied?Stochastic?Models?and?Data?Analysis,France,2005(5),17~20.
[3]Alvera-Azcárate?A.,Barth?A.,Rixen,et?al.Reconstruction?of?incomplete?oceanographic?data?sets?using?Empirical?Orthogonal?Functions.Application?to?the?Adriatic?Sea?surface?temperature.Ocean?Model,2005.325~346.
[4]Alvera-Azcárate?A.,Barth?A.,Beckers?J.-M.et.al.Multivariate?reconstruction?of?missing?data?in?sea?surface?temperature,chlorophyll,and?wind?satellite?fields.J.Geophys.Res.2007.112,C03008.doi:10.1029/2006JC003660.
[5]Lendasse?A.,Wertz?V.,Verleysen?M.Model?selection?with?cross-validations?and?bootstraps-application?to?time?series?prediction?with?rbfn?models.In?LNC?S,Springer-Verlag,Berlin,number2714,2003,573~580.
[6]Toumazou?V.,Cretaux?J.Using?a?Lanczos?eigensolver?in?the?computation?of?empirical?orthogonal?functions.Monthly?Weather?Review,2001,129(5):1243~1250.
[7] Sheng is towering, Shi Hanqing, and fourth is special again, utilizes the reconstruct of DINEOF method to lack the warm data in satellite remote sensing sea of survey, and Marine Sciences are made progress, and 2009,27 (2): 243 ~ 249.
[8] fourth is special again, the data reconstruct of satellite remote sensing sea surface temperature and suspension bed sediment concentration and data assimilation test: [PhD dissertation], Nanjing; Institutes Of Technology Of Nanjing, 2009.
It will be appreciated by those skilled in the art that accompanying drawing is the synoptic diagram of a preferred embodiment, the invention described above embodiment sequence number is not represented the quality of embodiment just to description.
The above is merely preferred embodiment of the present invention, and is in order to restriction the present invention, not all within spirit of the present invention and principle, any modification of being done, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (8)
1. the ocean remote sensing image water colour water temperature detection method based on compression sampling is characterized in that, said method comprising the steps of:
(1) obtains ocean water colour to be detected or the original remotely-sensed data of water temperature, original remotely-sensed data is carried out sparse conversion of compression sampling and Flame Image Process, obtain water colour water temperature data set;
(2) through the SOM method said water colour water temperature data set is carried out denoising and reconstruction processing successively, obtain and handle back water colour water temperature data set;
(3) said processing back water colour water temperature data set is carried out denoising, and utilize its continuous interpolation characteristic reconstruct disappearance view data;
(4) export and show said disappearance view data.
2. a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling according to claim 1 is characterized in that said method also comprises:
When including land data and island data in the said original remotely-sensed data, reject said land data and said island data.
3. a kind of ocean remote sensing image water colour water temperature detection method according to claim 1 based on compression sampling; It is characterized in that; Saidly said water colour water temperature data set is carried out denoising and reconstruction processing successively, obtains and handle back water colour water temperature data set and be specially through the SOM method:
Nonlinear organization through said SOM method carries out non-linear interpolation reconstruct and denoising to said water colour water temperature data set, obtains said processing back water colour water temperature data set.
4. a kind of ocean remote sensing image water colour water temperature detection method according to claim 1 based on compression sampling; It is characterized in that; Said said processing back water colour water temperature data set is carried out denoising, and utilizes continuous interpolation characteristic reconstruct disappearance view data to be specially:
1) said processing back water colour water temperature data set is carried out self-adaptation EOF and decompose, obtain space mode matrix U, singular value matrix S and time modal matrix V;
2) utilize the Monte Carlo to intersect calibration set and confirm that best reconstruct mode counts P;
3) count P, said space mode matrix U, said singular value matrix S and said time modal matrix V through said best reconstruct mode and make up said disappearance view data X.
5. a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling according to claim 4 is characterized in that, saidly utilizes the Monte Carlo to intersect calibration set to confirm that best reconstruct mode counts P and be specially:
If the intersection correction error when the mode number is P and P+1 is respectively R
PAnd R
P+1, Ratio was compared in the reduction of mode error when definition mode number was P
P=(R
P-1-R
P)/R
P, work as Ratio
PLess than assign thresholds and R
P<r
P+1The time, stop the selection of best reconstruct mode number, current mode is counted P as best reconstruct mode number, wherein, R
P-1Intersection correction error when being P-1 for the mode number.
6. a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling according to claim 4 is characterized in that said method also comprises: confirm that each said best reconstruct mode counts the best iterations of P, be no more than maximum iteration time N.
7. a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling according to claim 6 is characterized in that, the said iterations of confirming that said best reconstruct mode is counted P is specially:
Define said maximum iteration time N and minimum cross validation error, whether the difference of judging next iteration error and current iteration error is less than said minimum cross validation error, if stop iteration; Otherwise, continue to carry out iterative processing, when iterations is that said maximum iteration time N stops iteration.
8. a kind of ocean remote sensing image water colour water temperature detection method based on compression sampling according to claim 4 is characterized in that said disappearance view data X is specially:
wherein,
is the mean value of the valid data in the original remotely-sensed data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210227785.7A CN102798384B (en) | 2012-07-03 | 2012-07-03 | Ocean remote sensing image water color and water temperature monitoring method based on compression sampling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210227785.7A CN102798384B (en) | 2012-07-03 | 2012-07-03 | Ocean remote sensing image water color and water temperature monitoring method based on compression sampling |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102798384A true CN102798384A (en) | 2012-11-28 |
| CN102798384B CN102798384B (en) | 2014-09-17 |
Family
ID=47197585
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210227785.7A Expired - Fee Related CN102798384B (en) | 2012-07-03 | 2012-07-03 | Ocean remote sensing image water color and water temperature monitoring method based on compression sampling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102798384B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104101864A (en) * | 2013-04-10 | 2014-10-15 | 南京信息工程大学 | Navigation X-waveband radar ocean wave parameter inversion algorithm based on EOF decomposition |
| CN104361602A (en) * | 2014-11-26 | 2015-02-18 | 中国科学院遥感与数字地球研究所 | Water color detecting method and device based on MODIS image |
| CN106996968A (en) * | 2017-03-15 | 2017-08-01 | 淮阴工学院 | One kind is based on modified compressed sensing monitoring lake water quality system |
| CN107063195A (en) * | 2016-11-09 | 2017-08-18 | 哈尔滨工程大学 | A kind of extensive network locating method under water based on recurrence location estimation |
| CN107436954A (en) * | 2017-08-16 | 2017-12-05 | 吉林大学 | A kind of online flow data approximate processing method of quality control and device |
| CN110375719A (en) * | 2019-07-09 | 2019-10-25 | 山东科技大学 | One kind being based on the matched airborne laser sounding waterway design method in Monte Carlo |
| CN113064129A (en) * | 2021-03-03 | 2021-07-02 | 湖北中南鹏力海洋探测系统工程有限公司 | High-frequency ground wave radar ocean current synthesis method |
| CN113344805A (en) * | 2021-05-18 | 2021-09-03 | 南京信息工程大学 | Automatic meteorological station observation data restoration method based on EOF |
| JP2021533427A (en) * | 2019-07-04 | 2021-12-02 | 浙江大学Zhejiang University | Reconstruction method of geostationary sea color satellite data by empirical orthogonal function decomposition method |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107146265A (en) * | 2017-05-04 | 2017-09-08 | 安徽省气象科学研究所 | A kind of satellite cloud picture temporal interpolation method analyzed based on EOF |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101650837A (en) * | 2009-07-16 | 2010-02-17 | 浙江大学 | Virtual-reality modeling method of ocean water body |
-
2012
- 2012-07-03 CN CN201210227785.7A patent/CN102798384B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101650837A (en) * | 2009-07-16 | 2010-02-17 | 浙江大学 | Virtual-reality modeling method of ocean water body |
Non-Patent Citations (3)
| Title |
|---|
| 付东洋等: "卫星海洋遥感数据的无损压缩及解压算法", 《海洋技术》 * |
| 毛志华等: "卫星遥感业务系统海表温度误差控制方法", 《海洋学报》 * |
| 王建乐等: "基于SOM的海表温度遥感数据集的EOF算法重构", 《海洋技术》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104101864A (en) * | 2013-04-10 | 2014-10-15 | 南京信息工程大学 | Navigation X-waveband radar ocean wave parameter inversion algorithm based on EOF decomposition |
| CN104101864B (en) * | 2013-04-10 | 2016-08-03 | 南京信息工程大学 | The navigation X-band radar ocean wave parameter inversion algorithm decomposed based on EOF |
| CN104361602A (en) * | 2014-11-26 | 2015-02-18 | 中国科学院遥感与数字地球研究所 | Water color detecting method and device based on MODIS image |
| CN107063195A (en) * | 2016-11-09 | 2017-08-18 | 哈尔滨工程大学 | A kind of extensive network locating method under water based on recurrence location estimation |
| CN107063195B (en) * | 2016-11-09 | 2019-09-27 | 哈尔滨工程大学 | A kind of extensive underwater network locating method based on recurrence location estimation |
| CN106996968A (en) * | 2017-03-15 | 2017-08-01 | 淮阴工学院 | One kind is based on modified compressed sensing monitoring lake water quality system |
| CN107436954A (en) * | 2017-08-16 | 2017-12-05 | 吉林大学 | A kind of online flow data approximate processing method of quality control and device |
| CN107436954B (en) * | 2017-08-16 | 2018-10-02 | 吉林大学 | A kind of online flow data approximate processing method of quality control and device |
| JP7004844B2 (en) | 2019-07-04 | 2022-01-21 | 浙江大学 | Reconstruction method of geostationary sea color satellite data by empirical orthogonal function decomposition method |
| JP2021533427A (en) * | 2019-07-04 | 2021-12-02 | 浙江大学Zhejiang University | Reconstruction method of geostationary sea color satellite data by empirical orthogonal function decomposition method |
| US20230154081A1 (en) * | 2019-07-04 | 2023-05-18 | Zhejiang University | Method for reconstructing geostationary ocean color satellite data based on data interpolating empirical orthogonal functions |
| US11790580B2 (en) * | 2019-07-04 | 2023-10-17 | Zhejiang University | Method for reconstructing geostationary ocean color satellite data based on data interpolating empirical orthogonal functions |
| CN110375719A (en) * | 2019-07-09 | 2019-10-25 | 山东科技大学 | One kind being based on the matched airborne laser sounding waterway design method in Monte Carlo |
| CN113064129A (en) * | 2021-03-03 | 2021-07-02 | 湖北中南鹏力海洋探测系统工程有限公司 | High-frequency ground wave radar ocean current synthesis method |
| CN113344805A (en) * | 2021-05-18 | 2021-09-03 | 南京信息工程大学 | Automatic meteorological station observation data restoration method based on EOF |
| CN113344805B (en) * | 2021-05-18 | 2023-06-13 | 南京信息工程大学 | EOF-based automatic meteorological station observation data restoration method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102798384B (en) | 2014-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102798384B (en) | Ocean remote sensing image water color and water temperature monitoring method based on compression sampling | |
| Wang et al. | Deep learning for daily precipitation and temperature downscaling | |
| Najafi et al. | Understanding the changes in global crop yields through changes in climate and technology | |
| Sun et al. | Analyzing spatio-temporal changes and trade-offs to support the supply of multiple ecosystem services in Beijing, China | |
| Marlon et al. | Reconstructions of biomass burning from sediment-charcoal records to improve data–model comparisons | |
| Hamilton et al. | Ecuador’s mangrove forest carbon stocks: A spatiotemporal analysis of living carbon holdings and their depletion since the advent of commercial aquaculture | |
| Zeng et al. | Global land mapping of satellite-observed CO2 total columns using spatio-temporal geostatistics | |
| McLaughlin et al. | Rivers, runoff, and reefs | |
| Levy et al. | Freshwater availability anomalies and outbreak of internal war: results from a global spatial time series analysis | |
| Xiao et al. | The assimilation of satellite‐derived data into a one‐dimensional lower trophic level marine ecosystem model | |
| CN104484859B (en) | A kind of method that multispectral remote sensing image data remove thin cloud | |
| Li et al. | Synergistic data fusion of multimodal AOD and air quality data for near real-time full coverage air pollution assessment | |
| Li et al. | Heterogeneous effects of climate change and human activities on annual landscape change in coastal cities of mainland China | |
| Foroumandi et al. | Climate change or regional human impacts? Remote sensing tools, artificial neural networks, and wavelet approaches aim to solve the problem | |
| Chi et al. | Identifying the anthropogenic influence on the spatial distribution of plant diversity in an estuarine island through multiple gradients | |
| Knitter et al. | Land use patterns and climate change—a modeled scenario of the Late Bronze Age in Southern Greece | |
| CN115937692A (en) | Coastal wetland carbon sink effect evaluation method and system | |
| Liu et al. | The establishment of an eco-environmental evaluation model for southwest China and eastern South Africa based on the DPSIR framework | |
| Liu et al. | Unrevealing past and future vegetation restoration on the Loess Plateau and its impact on terrestrial water storage | |
| Chen et al. | iHydroSlide3D v1. 0: an advanced hydrological–geotechnical model for hydrological simulation and three-dimensional landslide prediction | |
| Povak et al. | Machine learning and hurdle models for improving regional predictions of stream water acid neutralizing capacity | |
| Yin et al. | Statistical downscaling of regional daily precipitation over southeast Australia based on self-organizing maps | |
| Lewis et al. | Identifying the multi‐scale spatial structure of plant community determinants of an important national resource | |
| Foster-Martinez et al. | Estimating wave attenuation at the coastal land margin with a GIS toolbox | |
| Roberts-Jones et al. | Estimating background error covariance parameters and assessing their impact in the OSTIA system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140917 Termination date: 20210703 |