CN105651263B - Shallow water depth multi-source remote sensing merges inversion method - Google Patents
Shallow water depth multi-source remote sensing merges inversion method Download PDFInfo
- Publication number
- CN105651263B CN105651263B CN201510975396.6A CN201510975396A CN105651263B CN 105651263 B CN105651263 B CN 105651263B CN 201510975396 A CN201510975396 A CN 201510975396A CN 105651263 B CN105651263 B CN 105651263B
- Authority
- CN
- China
- Prior art keywords
- depth
- water
- mrow
- msub
- image
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C13/00—Surveying specially adapted to open water, e.g. sea, lake, river or canal
- G01C13/008—Surveying specially adapted to open water, e.g. sea, lake, river or canal measuring depth of open water
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Hydrology & Water Resources (AREA)
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Image Processing (AREA)
Abstract
Shallow water depth multi-source remote sensing fusion inversion method includes：The first step：Multi-spectrum remote sensing image is pre-processed, obtains extra large table reflectivity；Second step：Field measurement water depth value obtains and processing；3rd step：Single source Depth extraction and depth of water segment identification；4th step：Multi-source Depth extraction merges；5th step：Depth extraction precision test；Using the Depth extraction result in n kind lists source and its corresponding depth of water segment identification image and fusion parameters as input, carry out the fusion of multi-source Depth extraction by pixel；After the completion of Depth extraction precision test, using final water depth value as the actual water depth value output data of remote sensing images.Compared with existing inversion method, this method can comprehensively utilize a variety of remotely-sensed data sources and the difference of Water Depth Information is responded, and excavate bathymetric data therein, improve inversion accuracy, handled by Decision fusion, the marine sounding of shallow water area under the complex situations that are particularly suitable for use in.
Description
Technical field
The present invention relates to a kind of marine sounding method, belongs to space remote sensing technical field, more particularly to one kind can profit
The deep penetrating shallow water depth multi-source remote sensing fusion inversion method of ocean water is carried out with a variety of remote sensors.
Background technology
Marine sounding is to ensure ship's navigation, development port and pier and ocean engineering construction, formulate seashore and island
The necessary basis data of Correlative plan.Compared with depth of water in-site measurement means, remote sensing technology is with covering is wide, the cycle is short, expense
Low, many-sided advantage such as spatial resolution is high.Since the 1970s, it is anti-the various passive remote sensing depth of waters have been carried out both at home and abroad
The research of model is drilled, conventional visible ray Depth extraction model mainly includes analysis model, half analysis semiempirical model and statistics
Model.Using different models, carried out in recent years in water-depth measurement fields such as river, lake, reservoir, island and littoral zone peripheries
Inverting application.
Depth of water visual remote sensing inverting is to obtain effective solution of the shallow sea complicated landform depth of water, it is particularly possible to which inverting obtains
Ship is taken can not be close and to be difficult to the water depth information into region.But because model is difficult to take into account physical mechanism and parametrization, because
The space that this existing visible ray RS Fathoming inverse model precision improves again is limited.
The limitation of environmental condition when the inverting of depth of water multi-source remote sensing can overcome single source video imaging, and multi-source Remote Sensing Images carry
The more rich band class information and its spectral resolution being not quite similar that supply are also beneficial to the extraction of Water Depth Information, currently will
Multi-source is applied to the research work of RS Fathoming inverting, but is the interpolation for spatial information mostly, not in Decision fusion aspect
Develop and apply.And Decision fusion can make full use of existing remote sensing image resource and information, to improve the optical remote sensing depth of water
Inversion accuracy provides new way.
Chinese patent (application number 201310188829.4, data of publication of application CN 104181515A) discloses that " one kind is based on
The shallow water depth inversion method of blue-yellow wave band high-spectral data ".It is mainly used in solving to carry out clean water using optical remote sensing means
The model of body Depth extraction is mostly established for multispectral data, and such algorithm is few by multispectral data wide waveband, spectral information
Restriction, the invention proposed based on high-spectral data and a kind of utilizes blue-yellow wave band (450-610 according to water body optical attenuation mechanism
Nanometer) high-spectral data inverting cleaning water body shallow water depth new method, this method can accurately extract shallow water depth within 30 meters
Distributed intelligence, and it is directed to a kind of remote sensor, it is only necessary to an algorithm coefficient demarcation is carried out, algorithm universality is substantially changed
It is kind.But the party obtains image as detection data source using the remote sensor in single source, available remote sensing image spectroscopic data wave band,
Spectral information is limited in scope, and is unfavorable for improving the accuracy that shallow water depth inverting is used for water-depth measurement, especially in complex situations
Under to the Effect on Detecting of neritic province domain depth of water deficiency.
The content of the invention
The invention provides a kind of shallow water depth multi-source remote sensing based on Decision fusion to merge inversion method, existing for solving
Have in technology and only to use the image of single source remote sensor as data source, its remote sensing image spectroscopic data wave band, spectral information make
It is limited with scope, the problem of water-depth measurement precision and poor accuracy.
Shallow water depth multi-source remote sensing merges inversion method, comprises the following steps：
The first step：Multi-spectrum remote sensing image is pre-processed, obtains extra large table reflectivity；
The pretreatment includes radiance conversion, atmospheric correction and solar flare and removed；
Second step：Field measurement water depth value obtains and processing；
The bathymetric data of test block and corresponding latitude and longitude coordinates are obtained, the tidal height at measurement moment is confirmed by tide table
Value, bathymetric data is corrected to the depth of water for obtaining theoretical depth reference plane, further according to the acquisition moment of multi-spectrum remote sensing image, to reason
The tidal correction of the instantaneous depth of water is carried out by the bathymetric data of depth datum to obtain the instantaneous depth of water；
3rd step：Single source Depth extraction and depth of water segment identification；
According to the relation between the depth of water at depth of water control point and corresponding image picture element reflectivity, carried out using multiband model
Statistical regression, the input that the parameter exported in the source image Depth extraction merges as multi-source inverting, and to multiband mould
Type carries out parameter calibration, and multiband model formation is as follows,
Xi=Ln (ρi-ρsi) (2)
Wherein, Z is the depth of water, and n is the wave band number for participating in inverting, A0And AiFor undetermined coefficient；ρiIt is the i-th wave band reflectivity
Data, ρsiIt is the reflectivity at the wave band deep water；
Depth of water control point is divided into multiple depth of water sections as inputting, the average relative error of each depth of water section is exported, as more
Another input, i.e. fusion parameters of source Depth extraction fusion；As the fusion parameters of multi-source Depth extraction fusion input, also wrap
Include the Kappa coefficients of single source image of output and the segmental averaging precision of each depth of water section；
Wherein, n is depth of water control point number, and k represents depth of water section, in formula 3, δkFor average relative error, ziIt is i-th of water
The measured value at deep control point, zi' it is its inverting value, in formula 4,For Kappa coefficients, xiiRepresent the control point correctly classified
Number, xi+、x+iIt is the ranks boundary value of error matrix when segmentation statistics is carried out to depth of water control point, in formula 5, δma_kIt is that segmentation is flat
Equal precision, PAkIt is producer's precision of k-th of depth of water section, UAkIt is user's precision of k-th of depth of water section；
Using fusion parameters and whole scape remote sensing image, single source Depth extraction result is calculated, and be corrected reason
By being segmented after depth datum to result, depth of water segment identification image is obtained；
4th step：Multi-source Depth extraction merges；
Using the Depth extraction result in n kind lists source and its corresponding depth of water segment identification image and fusion parameters as input, by
Pixel carries out the fusion of multi-source Depth extraction, specifically includes；
A) when the poll of some depth of water section is t, andExplanation hasKind or more kind image
Inversion result be in same depth of water section, whereinExpression rounds downwards, now, if 2 kinds or image of more than two kinds
Equal Depth extraction value is obtained, then directly assigns this value for current pixel, otherwise, compares this several image in the flat of the depth of water section
Equal relative error and mean accuracy, using depth of water section mean accuracy maximum as final pixel value；Only when the average essence of the depth of water section
Spend depth of water section average relative error corresponding to maximum image it is also maximum when, the image that selects mean accuracy to take second place；
B) when maximum number of votes obtained t meetsAnd it is t to have x (x >=2) numbers of votes obtained, now contrasts Kappa
Coefficient and n grader respective corresponding depth section mean accuracy, if the maximum image of Kappa coefficients and mean accuracy are maximum
Be all determined as same depth of water section, and be homologous image, by the water depth value of the image picture element as a result；If not same scape
Image, it is determined that this two scape average relative error in the depth of water section is less；If the maximum image of Kappa coefficients is judged
The depth of water section difference maximum with mean accuracy, then take the former water depth value；It is t in votes if only 1 number of votes obtained is t
Depth of water section in, using depth of water section mean accuracy it is maximum as final pixel value；When the image that the depth of water section mean accuracy is maximum
When corresponding depth of water section average relative error is also maximum, the image that selects mean accuracy to take second place；
C) as maximum votes t=1, the water depth value corresponding to the maximum image of Kappa coefficients is selected；
5th step：Depth extraction precision test；
The precision test is to utilize multi-source inverting knot after single source inversion result before the development fusion of depth of water checkpoint and fusion
The comparison of fruit, after the completion of Depth extraction precision test, using final water depth value as the actual water depth value output data of remote sensing images.
Shallow water depth multi-source remote sensing fusion inversion method as described above, the spoke brightness transition in the first step is will be distant
Sense image DN value is converted into spoke brightness value；The solar flare, which removes, can use median method, averaging method or wavelet method；It is described big
Gas correction can use FLAASH, dark pixel or 6S atmospheric correction methods.
The reference images that Decision fusion selects in the present invention, depending on the engineer's scale of required depth of water image and resolution ratio, nothing
Particular/special requirement.If the maximum image of selection spatial resolution makees benchmark, although the speed of service of fusion has a certain upgrade,
Spatial match uses represents whole pixel with the Decision fusion numerical value at pixel centre coordinate, and the information content of loss is larger.So
Consider from treatment effeciency and inverting fusion accuracy angle, it is preferred to use anti-with the depth of water that resolution ratio highest image is generated
Result image is drilled as benchmark, while needs the position by reference images pixel centre coordinate and other remote sensing source depth of water images to enter
Row matching, obtains all single source inverting water depth values and other information at the coordinate, Decision fusion is carried out, to reduce possible loss
Information content, ensure the precision of inverting.
Beneficial effects of the present invention：
Compared with existing inversion method, this method can comprehensively utilize a variety of remotely-sensed data sources and the difference of Water Depth Information is rung
Should, remote sensing image spectroscopic data wave band, the use range of spectral information are expanded, excavates bathymetric data therein, improves inverting
Precision, handled by Decision fusion, the marine sounding of shallow water area under the complex situations that are particularly suitable for use in.
Brief description of the drawings
Fig. 1 is the flow chart of the present invention；
Fig. 2 is the depth of water multi-source inverting fusion flow chart of the present invention；
Fig. 3 a are depth of water multi-source fusion result scatter diagrams；
Fig. 3 b are single source WorldView-2 Depth extractions result scatter diagrams；
Fig. 3 c are single source Pleiades Depth extractions result scatter diagrams；
Fig. 3 d are single source QuickBird Depth extractions result scatter diagrams；
Fig. 3 e are single source SPOT-6 Depth extractions result scatter diagrams；
Fig. 4 is depth of water multi-source remote sensing inverting fusion results of the present invention；
Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described.
The specific embodiment of the invention is described in further detail with reference to accompanying drawing 1, the fusion inverting of shallow water depth multi-source remote sensing
Method, specifically include following steps：
The first step：Multi-spectral Remote Sensing Data pre-processes：
First have to locate the multi-spectrum remote sensing image that the participation Depth extraction obtained by Multiple Source Sensor merges in advance
Reason, including spoke brightness transition, atmospheric correction and solar flare remove.Spoke brightness transition is that image DN value is converted into spoke brightness
Process, spoke brightness transition formula corresponding to different remotely-sensed data products is different, typically using following two：
Corresponding parameters can obtain in the meta data file of image in formula.Obtain multispectral spoke brightness image
Afterwards, atmospheric correction is carried out using the methods of FLAASH or dark pixel, 6S, obtains extra large table reflectivity data；To remove Hai Biaotai
The interference that positive solar flare and floating object etc. are brought, then using progress solar flare removal the methods of intermediate value, average or small echo.
Second：Field measurement water depth value obtains and processing：
The bathymetric data of test block is obtained using multi-beam fathometer or other water-depth measurement means, while corresponding to acquisition
Latitude and longitude coordinates.The tidal height value for confirming the measurement moment with tide table is corrected to bathymetric data.
3rd step：Single source Depth extraction and depth of water segment identification：
According to the relation between the depth of water at depth of water control point and corresponding image picture element reflectivity, carried out using multiband model
Statistical regression, the input that the parameter exported in the source image Depth extraction merges as multi-source inverting, and to multiband mould
Type carries out parameter calibration, and multiband model formation is as follows,
Xi=Ln (ρi-ρsi) (2)
Wherein, Z is the depth of water, and n is the wave band number for participating in inverting, A0And AiFor undetermined coefficient；ρiIt is the i-th wave band reflectivity
Data, ρsiIt is the reflectivity at the wave band deep water.
Depth of water control point is divided into multiple depth of water sections as inputting, the average relative error of each depth of water section is exported, as more
Another input, i.e. fusion parameters of source Depth extraction fusion；
The fusion parameters of input, in addition to the Kappa coefficients of single source image of output and the segmental averaging of each depth of water section
Precision；
Wherein, n is depth of water control point number, and k represents depth of water section, in formula 3, δkFor average relative error, ziIt is i-th of water
The measured value at deep control point, zi' it is its inverting value, in formula 4,For Kappa coefficients, xiiRepresent the control point correctly classified
Number, xi+、x+iBe to depth of water control point carry out segmentation statistics when, the ranks boundary value of error matrix, δma_kIt is segmental averaging precision,
PAkIt is producer's precision of k-th of depth of water section, UAkIt is user's precision of k-th of depth of water section；
Using obtained parameter and whole scape remote sensing image, single source Depth extraction result is calculated to obtain, and by its instantaneous depth of water school
Theoretical depth reference plane is just arrived, result is being segmented afterwards, is obtaining depth of water segment identification image；
This sentences error matrix of the WorldView-2 images used in multi-source inverting fusion method at depth of water control point
Exemplified by, shown in table 1.
Table 1：The error matrix at WorldView-2 depth of waters control point
In error matrix, row represent ground reference checking information, the classification that behavior remotely-sensed data obtains, main diagonal element
(such as x11, x is expressed as in formulaii) it is the correct pixel of classification, the outer element of diagonal is that remotely-sensed data classification is joined relative to ground
Examine the pixel number of mistake.So in this experiment, 1~4 represents 4 depth of water sections being divided into using 2m, 5m, 10m as interval, z respectively
To survey the depth of water, z' is the inverting depth of water.Row represent the quantity at control point in 4 actual measurement depth of water sections, and row represents and utilizes remote sensing image
Inverting obtains the quantity at control point in 4 depth of water sections, and leading diagonal is that the pixel inverting depth of water is assigned to the section that correctly sounds the depth of the water
Point number, conversely, line outside be mistake segmentation point number.Producer's precision (PA) in error matrix is it is assumed that 1 water
Deep control point in kth class, during the remote sensing image inverting depth of water by this put corresponding to pixel be classified as k probability, by kth class just
The summation of true classification number divided by kth row (is expressed as x in formula+i) try to achieve.User's precision (UA) is if be that the image inverting depth of water will
When certain dominating pair of vertices answers the pixel to be grouped into kth class, true sound the depth of the water at the depth of water control point belongs to the percentage of kth class, its
Calculate the number by being correctly categorized as kth class divided by be categorized as k summation (in the namely summation of row k, i.e. formula
xi+)。
Kappa coefficients are that uniformity or precision are measured between Classification in Remote Sensing Image figure and reference data, by leading diagonal and ranks
Probabilistic consistency that sum provides is expressed.Kappa coefficients in example can be construed to utilize this scape for 0.7686
The water depth distribution obtained after the WorldView-2 image inverting depth of waters is better than the depth of water section of random division with 76.86% degree.
Producer's precision and user's precision are better closer to 1, optimal situation be producer's precision and user's precision all
For 1.So considered for balance do not lose it is biased, and also to simplify the final number of parameters for participating in Decision fusion, this implementation
The average of the two is taken as fusion parameters, i.e. segmental averaging precision in example.
Using fusion parameters and whole scape remote sensing image, single source Depth extraction result is calculated, and be corrected reason
By being segmented after depth datum to result, depth of water segment identification image is obtained；
4th step：Multi-source Depth extraction merges：
Merged using single source Depth extraction result, depth of water segment identification image and fusion parameters as multi-source Depth extraction defeated
Enter, carry out by pixel and merge.Determine finally to take in the depth of water section institute votes of current pixel using 4 single sources (i.e. remote sensing image)
Value, it is specific as follows：
A) when the poll of some depth of water section is more than or equal to 3, the image inversion result for illustrating to have 3 kinds or more than 3 kinds is same
Then directly it is current pixel if the image of 2 and the above obtain equal Depth extraction value now in one depth of water section
This value is assigned, otherwise, compares average relative error and mean accuracy of this several image in the depth of water section, selects mean accuracy as far as possible
The big and water depth value as current pixel that average relative error is small.On the premise of mean accuracy is larger, investigates the image and exist
The average relative error of this depth of water section, if the image average relative error of mean accuracy maximum is also maximum, the selection abandoned is flat
Second largest image of equal precision；
B) when maximum number of votes obtained be equal to 2, and win the vote situation be 2,2, it is meant that the inverting depth of water for having two width images respectively falls
In same depth of water section.The mean accuracy of Kappa coefficients and 4 graders in respective corresponding depth section is now investigated, if Kappa systems
Several maximum images and mean accuracy maximums are all determined as same depth of water section, and are homologous images, just select the image picture
The water depth value of member is as a result, if not same scape image, then select this two scape average relative error in the depth of water section smaller
's.If the difference of the depth of water section that the maximum image of Kappa coefficients is judged and mean accuracy maximum, selects the former depth of water
Value.When maximum number of votes obtained be equal to 2, and win the vote situation be 2,1,1, votes be 2 depth of water section in, select mean accuracy as far as possible
The big and water depth value as current pixel that average relative error is small；
C) when maximum votes are 1, i.e., the classification results in 4 single sources differ, that is to say, that the inverting of 4 scape images
Water depth distribution now believes the maximum image of Kappa coefficients in 4 depth of water sections.
5th step：Depth extraction precision test：
Carry out the precision test of single source inversion result and multi-source inversion result using checkpoint, calculate overall and point different water
The average relative error and mean absolute error of deep section, so as to be verified to the precision of multi-source Depth extraction fusion.
(1) depth of water multi-source remote sensing inverting fusion parameters and Fusion Model implementation status
The present embodiment to choose the QuickBird on January 10th, 2008, the WorldView-2 of on 2 7th, 2010,
The process that the SPOT-6 in 5, Pleiades and 2013 on April on March 9th, 2012 carries out depth of water multi-source inverting fusion is carried out pair
Than checking.The inverted parameters obtained by blue, green, the red three wave bands log-linear model inverting depth of water, and control are illustrated in table 2
Segmental averaging relative error at system point.Multi-source Remote Sensing Images Depth extraction fusion using WorldView-2 Depth extractions image as
Basis, Decision fusion is carried out with the Depth extraction result of Pleiades, QuickBird and SPOT-6 image.Image modality in table 2
Order according to the spatial resolution of image, gradually increase arranges from left to right, segmental averaging precision and segmental averaging relative error
1-4 represent 4 depth of water sections using 2m, 5m, 10m by spaced points point successively.
Table 2：Depth of water list source remote-sensing inversion parameter and multi-source Decision fusion parameter
By comparative analysis as can be seen that overall segmentation precision highest is SPOT-6 images, that worst is QuickBird
Image.Comprehensive segmentation mean accuracy and segmental averaging relative error are considered, the most preferably Pleiades images in the 1st section, it
Segmentation precision highest, it is and also smaller in the average relative error of this section, be secondly SPOT-6 images, its average relative error
It is small 8 percentage points compared with the former, but segmental averaging precision is inferior to the former.Although segmental averaging of the Pleiades images in the 2nd section
Precision is best, but it is maximum in the average relative error of this depth of water section is 4 scape images, so ensure that higher point
On the premise of section mean accuracy, average relative error is it is also preferable that SPOT-6 images.In 3rd and 4 section, no matter SPOT-6 images
It is best in segmental averaging precision or segmental averaging relative error.
As shown in Fig. 2 in the result of depth of water multi-source inverting fusion, the inverting depth of water fusion evaluation of generation has 1002 rows,
1054 row, that is, share 1056108 pixels.By statistics, determine that the pixel number of water depth value is most according to the 2nd rule, be
860835, the 81.51% of all pixel numbers are accounted for, when illustrating to carry out Decision fusion by pixel, the maximum ballot of most of pixels
Number account for sum more than half, that is to say, that at least 3 scape images are same depth of water section in the inverting water depth value of the pixel, and
And final water depth value is depended in the maximum Depth extraction image of this depth of water section mean accuracy.Secondly, it is more to perform number
Be the 6th rule, be 72132, percentage 6.83%, minimum is the 9th rule, only 128 pixels.Only when 4
The water depth value of image inverting carries out the 9th rule in different water depth Duan Zhongcai, although this means that the Depth extraction of 4 scape images
Ability is different, but the result difference in depth of water segmentation is not too large, and only very small part has obvious difference, institute
It can be played a role with this 4 scape image in Decision fusion.
(2) the overall precision checking analysis of depth of water multi-source remote sensing inverting fusion
Ratio of precision is made compared with obtained each precision evaluation refers to single source result before merging to depth of water multi-source inverting fusion results
Mark as shown in table 3 below.
Table 3：The overall precision of depth of water multi-source inverting fusion compares
Three evaluation indexes all show for the result of the more former image inverting of result after depth of water multi-source Decision fusion
Improve more notable.Average relative error be followed successively by from small to large Decision fusion image, SPOT-6 images, QuickBird images,
Pleiades images and WorldView-2 images, compared to worst WorldView-2 images, image controls in the depth of water after fusion
The average relative error at point place reduces more than 40 percentage points, and when being initialized before fusion to result image, it is exactly with this scape shadow
The inversion result of picture illustrates that Decision fusion largely improves the inversion result of former image really as benchmark.Even if with 4 scapes
The best SPOT-6 images of inversion accuracy are compared in image, and fusion evaluation also has subtracting for 12.7 percentage in relative error
It is small.The minimum value of mean absolute error is by fusion evaluation or SPOT-6 images and Pleiades with differing 1.4m between maximum
For image compared to obtaining, the mean absolute error of QuickBird images and WorldView-2 images is larger, value be respectively 1.6m and
1.8m, as many as 0.8m and 1m are differed between minimum value.Kappa coefficients for evaluating segmentation precision also indicate that：Fusion evaluation
Pixel is more accurate in the differentiation that depth of water section belongs to, next to that Pleiades images and SPOT-6 images, by QuickBird shadows
As inverting, to obtain depth of water segment identification image precision poor.It is generally understood that Kappa values, when more than 0.80, classification chart and ground are joined
Examine that uniformity between information is very big or precision is very high, Kappa values both greater than critical value of this 4 images, illustrate that its is consistent
Property is all relatively good.Worst is WorldView-2 images, ranks most end with 0.6139 Kappa values.
As shown in Fig. 3 a, 3b, 3c, 3d, 3e, the front and rear actual measurement depth of water of depth of water multi-source inverting fusion and the inverting depth of water are given
Scatter diagram.By scatter diagram it can be found that in addition to Pleiades images, another 3 scape image to below 2m depth of water point efficiency of inverse process all
It is undesirable.In WorldView-2 image scatter diagrams, concentration is compared in the distribution of data point, and average relative error should be received greatly
The influence of the data point of phytal zone.WorldView-2 images and the maximum water depth value of Pleiades image invertings are beyond actual measurement
The scope of depth of water checkpoint, this does not occur in the scatter diagram of QuickBird images and SPOT-6 images.
As shown in figure 4, single source inversion result of different resolution is combined together by the present embodiment by Decision fusion, enclose
The 20m that island one is enclosed is fine and smooth with shallow water area texture, and water depth ratio is smaller, the reef disk being clear that where northern island；In island west
The profundal zone texture of the bigger depth of south and northeastward is then more coarse；It is about at 20m in depth, depth of water gradient is larger, by
Shallow to deep transition is more apparent.
(3) the segmentation precision checking analysis of depth of water multi-source remote sensing inverting fusion
The segmentation error distribution of depth of water multi-source inverting fusion is observed, as shown in table 4, is increased with depth, average relative error
With mean absolute error without the trend of regular increase or reduction.
Table 4：The segmentation error of depth of water multi-source inverting fusion compares
In 0-2m depth of water sections, although the average relative error of inversion result is generally relatively low, but still have it is very significant poor
Away from.Precision highest is depth of water multi-source inverting fusion evaluation, and average relative error is 39.1%, mean absolute error 0.3m.
Next to that Pleiades images, differ 3.9%, mean absolute error is equal with the former average relative error.It is afterwards
QuickBird images, SPOT-6 images and WorldView-2 images, its average relative error and mean absolute error all be in by
Cumulative big gesture, especially WorldView-2 images, its average relative error in the depth of water section is the 2 of SPOT-6 images
Times, compared with best inverting fusion evaluation, gap up to 210.1%, and mean absolute error is also almost inverting fusion
4 times of image, it is 1.1m.In 2-5m depth of water sections, precision highest is inverting fusion evaluation and SPOT-6 images, and both is flat
Equal relative error and mean absolute error are equal, respectively 5.3% and 0.2m.WorldView-2 images are anti-the depth of water section
Drill ability to be substantially improved compared to shallow water section, average relative error 28.8%, mean absolute error 1.0m, but with the depth of water
Inverting fusion evaluation best Duan Jingdu is compared with SPOT-6 images, and gap is quite obvious.QuickBird images with
32.8% average relative error and 1.4m mean absolute error come the 4th, and inversion accuracy is best in 0-2m depth of water sections
Inversion accuracy of the Pleiades images in this depth of water section is worst.In 5-10m depth of water sections, by average relative error and averagely
The order arrangement of absolute error from small to large, be followed successively by inverting fusion evaluation and SPOT-6 images, QuickBird images,
Pleiades images, WorldView-2 images.8 percentages, average absolute are differed between the average relative error of minimum and maximum
Error at most differs 0.6m.In the depth of water section of 10-20m scopes, minimum average relative error and mean absolute error are come
It is maximum for Pleiades images, value respectively 6.3%, 22.5% and 3.4m, 0.9m from SPOT-6 images.Shadow is merged in inverting
The inversion accuracy of picture is preferable, average relative error 6.4%, mean absolute error 1.0m.
The inverting of depth of water multi-source merge inverting in, SPOT-6 images in addition to inverting ability is performed poor in shallow water section,
Other depth of water Duan Jun are the scapes of precision highest 1 in all remote sensing Depth extraction images.Pleiades images effectively compensate for SPOT-
6 images are in the deficiency of phytal zone, but it is worst in 2-5m and 10-20m precision is 4 scapes.WorldView-2 images are in 0-
Inversion accuracy is worst in this 2 depth of water sections of 2m and 5-10m, general in other 2 depth of water section precision.And QuickBird images exist
Inversion accuracy in each depth of water section is all hovered moderate.Except the precision in 10-20m depth of water sections is slightly below SPOT-6 shadows
Picture, multi-source inverting fusion evaluation are best in the inversion accuracy of other depth of water sections.
It is of the invention compared with existing inversion method, this method can comprehensively utilize a variety of remotely-sensed data sources to Water Depth Information
Difference response, expands remote sensing image spectroscopic data wave band, the use range of spectral information, excavates bathymetric data therein, carry
High inversion accuracy, is handled by Decision fusion, the marine sounding of shallow water area under the complex situations that are particularly suitable for use in.
The technology contents of the not detailed description of the present invention are known technology.
Claims (2)
1. shallow water depth multi-source Remote Sensing Images inversion method, it is characterised in that comprise the following steps：
The first step：Multi-spectrum remote sensing image is pre-processed, obtains extra large table reflectivity；
The pretreatment includes radiance conversion, atmospheric correction and solar flare and removed；
Second step：Field measurement water depth value obtains and processing；
The bathymetric data of test block and corresponding latitude and longitude coordinates are obtained, the tidal height value at measurement moment is confirmed by tide table, will
Bathymetric data correction obtains the depth of water of theoretical depth reference plane, further according to the acquisition moment of multi-spectrum remote sensing image, to theoretical deep
The bathymetric data of degree reference plane carries out the tidal correction of the instantaneous depth of water to obtain the instantaneous depth of water；
3rd step：Single source Depth extraction and depth of water segment identification；
According to the relation between the depth of water at depth of water control point and corresponding image picture element reflectivity, counted using multiband model
Return, the input that the parameter exported in the source image Depth extraction merges as multi-source inverting, and multiband model is entered
Row parameter calibration, multiband model formation is as follows,
<mrow>
<mi>Z</mi>
<mo>=</mo>
<msub>
<mi>A</mi>
<mn>0</mn>
</msub>
<mo>+</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msub>
<mi>A</mi>
<mi>i</mi>
</msub>
<msub>
<mi>X</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
Xi=Ln (ρi-ρsi) (2)
Wherein, Z is the depth of water, and n is the wave band number for participating in inverting, A0And AiFor undetermined coefficient；ρiIt is the i-th wave band reflectivity data,
ρsiIt is the reflectivity at the i-th wave band deep water；
Depth of water control point is divided into multiple depth of water sections as inputting, the average relative error of each depth of water section is exported, as multi-source water
Another input, i.e. fusion parameters of deep inverting fusion；The fusion parameters inputted as the fusion of multi-source Depth extraction, in addition to it is defeated
The Kappa coefficients of the single source image gone out and the segmental averaging precision of each depth of water section；
<mrow>
<msub>
<mi>&delta;</mi>
<mi>k</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>n</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<mfrac>
<mrow>
<mo>|</mo>
<msub>
<mi>z</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<msup>
<msub>
<mi>z</mi>
<mi>i</mi>
</msub>
<mo>&prime;</mo>
</msup>
<mo>|</mo>
</mrow>
<msub>
<mi>z</mi>
<mi>i</mi>
</msub>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mover>
<mi>K</mi>
<mo>^</mo>
</mover>
<mo>=</mo>
<mfrac>
<mrow>
<mi>n</mi>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>k</mi>
</munderover>
<msub>
<mi>x</mi>
<mrow>
<mi>i</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>-</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>k</mi>
</munderover>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mrow>
<mi>i</mi>
<mo>+</mo>
</mrow>
</msub>
<mo>&times;</mo>
<msub>
<mi>x</mi>
<mrow>
<mo>+</mo>
<mi>i</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msup>
<mi>n</mi>
<mn>2</mn>
</msup>
<mo>-</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>k</mi>
</munderover>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mrow>
<mi>i</mi>
<mo>+</mo>
</mrow>
</msub>
<mo>&times;</mo>
<msub>
<mi>x</mi>
<mrow>
<mo>+</mo>
<mi>i</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>&delta;</mi>
<mrow>
<mi>m</mi>
<mi>a</mi>
<mo>_</mo>
<mi>k</mi>
</mrow>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>PA</mi>
<mi>k</mi>
</msub>
<mo>+</mo>
<msub>
<mi>UA</mi>
<mi>k</mi>
</msub>
</mrow>
<mn>2</mn>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>5</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, n is depth of water control point number, and k represents depth of water section, in formula 3, δkFor average relative error, ziIt is i-th of depth of water control
Make the measured value of point, zi' it is its inverting value, in formula 4,For Kappa coefficients, xiiRepresent the control point number correctly classified, xi+、
x+iIt is the ranks boundary value of error matrix when segmentation statistics is carried out to depth of water control point, in formula 5, δma_kIt is segmental averaging precision,
PAkIt is producer's precision of k-th of depth of water section, UAkIt is user's precision of k-th of depth of water section；
Using fusion parameters and whole scape remote sensing image, single source Depth extraction result is calculated, and is corrected theoretical deep
Result is segmented after degree reference plane, obtains depth of water segment identification image；
4th step：Multi-source Depth extraction merges；
Using the Depth extraction result in n kind lists source and its corresponding depth of water segment identification image and fusion parameters as input, by pixel
Carry out the fusion of multi-source Depth extraction, specifically include；
A) when the poll of some depth of water section is t, andExplanation hasThe inverting knot of kind or more kind image
Fruit be in same depth of water section, whereinExpression rounds downwards, now, if 2 kinds or image of more than two kinds obtain phase
Deng Depth extraction value, then directly assign this value for current pixel, otherwise, compare this several image in the average relative of the depth of water section
Error and mean accuracy, using depth of water section mean accuracy maximum as final pixel value；Only when the depth of water section mean accuracy is maximum
Image corresponding to depth of water section average relative error it is also maximum when, select the image that takes second place of mean accuracy；
B) when maximum number of votes obtained t meetsAnd it is t to have x (x >=2) numbers of votes obtained, now contrasts Kappa coefficients
With n grader respective corresponding depth section mean accuracy, if the maximum image of Kappa coefficients and mean accuracy are maximum all
It is determined as same depth of water section, and is homologous image, by the water depth value of the image picture element as a result；If not same scape shadow
Picture, it is determined that this two scape average relative error in the depth of water section is less；If the water that the maximum image of Kappa coefficients is judged
The deep section difference maximum with mean accuracy, then take the former water depth value；It is t's in votes if only 1 number of votes obtained is t
In depth of water section, using depth of water section mean accuracy maximum as final pixel value；When the image pair that the depth of water section mean accuracy is maximum
When the depth of water section average relative error answered is also maximum, the image that takes second place of mean accuracy is selected；
C) as maximum votes t=1, the water depth value corresponding to the maximum image of Kappa coefficients is selected；
5th step：Depth extraction precision test；
The precision test utilizes multi-source inversion result after single source inversion result before the development fusion of depth of water checkpoint and fusion
Compare, after the completion of Depth extraction precision test, using final water depth value as the actual water depth value output data of remote sensing images.
2. shallow water depth multi-source Remote Sensing Images inversion method according to claim 1, it is characterised in that in the first step
Radiance conversion be that remote sensing image DN values are converted into spoke brightness value；The solar flare, which removes, can use median method,
Value method or wavelet method；The atmospheric correction can use FLAASH, dark pixel or 6S atmospheric correction methods.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510975396.6A CN105651263B (en) | 2015-12-23 | 2015-12-23 | Shallow water depth multi-source remote sensing merges inversion method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510975396.6A CN105651263B (en) | 2015-12-23 | 2015-12-23 | Shallow water depth multi-source remote sensing merges inversion method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105651263A CN105651263A (en) | 2016-06-08 |
| CN105651263B true CN105651263B (en) | 2018-02-23 |
Family
ID=56476649
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510975396.6A Expired - Fee Related CN105651263B (en) | 2015-12-23 | 2015-12-23 | Shallow water depth multi-source remote sensing merges inversion method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105651263B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109059796B (en) * | 2018-07-20 | 2020-07-31 | 自然资源部第三海洋研究所 | Shallow sea water depth multispectral satellite remote sensing inversion method for water depth control point-free area |
| CN109657392A (en) * | 2018-12-28 | 2019-04-19 | 北京航空航天大学 | A kind of high-spectrum remote-sensing inversion method based on deep learning |
| CN109631862A (en) * | 2019-01-22 | 2019-04-16 | 青岛秀山移动测量有限公司 | A kind of multi-Sensor Information Fusion Approach of intertidal zone integration mapping |
| CN111561916B (en) * | 2020-01-19 | 2021-09-28 | 自然资源部第二海洋研究所 | Shallow sea water depth uncontrolled extraction method based on four-waveband multispectral remote sensing image |
| CN111651707B (en) * | 2020-05-28 | 2023-04-25 | 广西大学 | Tidal level inversion method based on optical shallow water region satellite remote sensing image |
| CN112013822A (en) * | 2020-07-22 | 2020-12-01 | 武汉智图云起科技有限公司 | Multispectral remote sensing water depth inversion method based on improved GWR model |
| CN111947628B (en) * | 2020-08-25 | 2022-05-27 | 自然资源部第一海洋研究所 | Linear water depth inversion method based on inherent optical parameters |
| CN113326470B (en) * | 2021-04-11 | 2022-08-16 | 桂林理工大学 | Remote sensing water depth inversion tidal height correction method |
| CN113255144B (en) * | 2021-06-02 | 2021-09-07 | 中国地质大学（武汉） | Shallow sea remote sensing water depth inversion method based on FUI partition and Randac |
| CN113639716A (en) * | 2021-07-29 | 2021-11-12 | 北京航空航天大学 | Depth residual shrinkage network-based water depth remote sensing inversion method |
| CN113793374B (en) * | 2021-09-01 | 2023-12-22 | 自然资源部第二海洋研究所 | Method for inverting water depth based on water quality inversion result by improved four-band remote sensing image QAA algorithm |
| CN114943161B (en) * | 2022-07-27 | 2022-09-27 | 中国水利水电科学研究院 | Inland lake terrain inversion method based on multi-source remote sensing data |
| CN117514148B (en) * | 2024-01-05 | 2024-03-26 | 贵州航天凯山石油仪器有限公司 | Oil-gas well working fluid level identification and diagnosis method based on multidimensional credibility fusion |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9046363B2 (en) * | 2012-04-27 | 2015-06-02 | SATOP GmbH | Using multispectral satellite data to determine littoral water depths despite varying water turbidity |
| CN104457901B (en) * | 2014-11-28 | 2018-01-05 | 南京信息工程大学 | A kind of method and system for determining the depth of water |
-
2015
- 2015-12-23 CN CN201510975396.6A patent/CN105651263B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN105651263A (en) | 2016-06-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105651263B (en) | Shallow water depth multi-source remote sensing merges inversion method | |
| CN102436652B (en) | Automatic registering method of multisource remote sensing images | |
| CN102708369B (en) | Sea ice parameter extraction method on basis of satellite image | |
| CN103148842B (en) | Shallow sea sand wave area multi-beam sounding terrain reconstruction method based on remote sensing image features | |
| CN102254319A (en) | Method for carrying out change detection on multi-level segmented remote sensing image | |
| CN102750696B (en) | Affine invariant feature and coastline constraint-based automatic coastal zone remote-sensing image registration method | |
| CN109059796A (en) | The multispectral satellite remote sensing inversion method of shallow water depth without depth of water control point region | |
| CN105469393A (en) | Shallow water depth multi-temporal remote sensing image inversion method based on decision fusion | |
| CN111008664B (en) | Hyperspectral sea ice detection method based on space-spectrum combined characteristics | |
| CN103226832B (en) | Based on the multi-spectrum remote sensing image change detecting method of spectral reflectivity mutation analysis | |
| CN112013822A (en) | Multispectral remote sensing water depth inversion method based on improved GWR model | |
| CN105627997A (en) | Multi-angle remote sensing water depth decision fusion inversion method | |
| CN103729846A (en) | LiDAR point cloud data edge detection method based on triangular irregular network | |
| CN103745453B (en) | Urban residential areas method based on Google Earth remote sensing image | |
| CN104680151B (en) | A kind of panchromatic remote sensing image variation detection method of high-resolution for taking snow covering influence into account | |
| CN113221813B (en) | Coastline remote sensing extraction method | |
| CN103065320A (en) | Synthetic aperture radar (SAR) image change detection method based on constant false alarm threshold value | |
| CN117274831A (en) | Offshore turbid water body depth inversion method based on machine learning and hyperspectral satellite remote sensing image | |
| Zhang et al. | A study on coastline extraction and its trend based on remote sensing image data mining | |
| CN111561916B (en) | Shallow sea water depth uncontrolled extraction method based on four-waveband multispectral remote sensing image | |
| CN101383046A (en) | Three-dimensional reconstruction method on basis of image | |
| CN114387531B (en) | Surface temperature downscaling method based on improved geographic weighted regression model | |
| CN112633123B (en) | Heterogeneous remote sensing image change detection method and device based on deep learning | |
| CN102760293A (en) | Image quality evaluation method based on distance matrix | |
| CN106210696A (en) | A kind of method and device of real-time virtual View Synthesis |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180223 Termination date: 20181223 |