CN116295285A - Shallow sea water depth remote sensing inversion method based on region self-adaption - Google Patents
Shallow sea water depth remote sensing inversion method based on region self-adaption Download PDFInfo
- Publication number
- CN116295285A CN116295285A CN202310132412.XA CN202310132412A CN116295285A CN 116295285 A CN116295285 A CN 116295285A CN 202310132412 A CN202310132412 A CN 202310132412A CN 116295285 A CN116295285 A CN 116295285A
- Authority
- CN
- China
- Prior art keywords
- water depth
- model
- inversion
- remote sensing
- band
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 55
- 239000013535 sea water Substances 0.000 title claims abstract description 25
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 227
- 239000000758 substrate Substances 0.000 claims abstract description 21
- 238000012545 processing Methods 0.000 claims abstract description 19
- 238000011160 research Methods 0.000 claims abstract description 15
- 238000006243 chemical reaction Methods 0.000 claims abstract description 11
- 238000003062 neural network model Methods 0.000 claims abstract description 10
- 238000010801 machine learning Methods 0.000 claims abstract description 7
- 238000005457 optimization Methods 0.000 claims abstract description 5
- 230000001939 inductive effect Effects 0.000 claims abstract description 3
- 238000012937 correction Methods 0.000 claims description 35
- 210000002569 neuron Anatomy 0.000 claims description 18
- 230000003595 spectral effect Effects 0.000 claims description 16
- 238000010276 construction Methods 0.000 claims description 12
- 238000012795 verification Methods 0.000 claims description 12
- 238000005192 partition Methods 0.000 claims description 11
- 230000005855 radiation Effects 0.000 claims description 11
- 238000004458 analytical method Methods 0.000 claims description 9
- 210000004027 cell Anatomy 0.000 claims description 9
- 230000011218 segmentation Effects 0.000 claims description 8
- 238000000638 solvent extraction Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 4
- 241001085205 Prenanthella exigua Species 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000010219 correlation analysis Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000005286 illumination Methods 0.000 claims description 3
- 238000002310 reflectometry Methods 0.000 claims description 3
- 238000001228 spectrum Methods 0.000 claims description 2
- 230000006978 adaptation Effects 0.000 claims 8
- 230000009286 beneficial effect Effects 0.000 abstract description 3
- 238000001514 detection method Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000010521 absorption reaction Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000012876 topography Methods 0.000 description 2
- 239000000443 aerosol Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000009364 mariculture Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 239000013049 sediment Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
Images
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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
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)
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
Abstract
The invention provides a shallow sea water depth remote sensing inversion method based on regional self-adaption, which comprises the following steps: s1, collecting and processing data; s2, defining a comprehensive area according to the data collected and processed in the step S1; s3, constructing a model library, and inducing a water depth remote sensing inversion model into a single-band model, a double-band model, a multi-band model, a band ratio model, a logarithmic conversion ratio model, a neural network model and a machine learning model; s4, carrying out regional model optimization according to the model library constructed in the step S3; s5, carrying out water depth inversion according to the region model which is optimized in the step S4. The invention has the beneficial effects that: the regional self-adaptive shallow sea water depth remote sensing inversion method is established by combining the existing research foundation and comprehensively considering the water depth range, the substrate type and the water quality condition, so as to improve the remote sensing inversion precision of the offshore water depth to the greatest extent.
Description
Technical Field
The invention belongs to the technical field of remote sensing, and particularly relates to a shallow sea water depth remote sensing inversion method based on regional self-adaption.
Background
Shallow sea is the most frequent area of sea-land interaction, also is the most important area of engineering development activities such as land construction, harbour construction, mariculture, maritime traffic, and the like, carries out shallow sea water depth measurement work, accurately grasps shallow sea water depth information and seabed topography and topography, and has important significance for national maritime traffic safety, comprehensive management of coastal zone, military national defense, and the like. The traditional water depth measuring method is carried out by adopting shipborne detection equipment, but the operation cannot be carried out in shallow sea areas and sensitive sea areas, so that a large number of actually measured empty water depth areas exist in offshore shallow sea areas. The remote sensing technology has the advantages of large area, quick updating, indirect contact and the like, and can be used as a beneficial supplement for the traditional water depth measurement technology to develop the water depth information acquisition of shallow sea areas.
The invention relates to a water depth remote sensing detection method, which comprises an active water depth remote sensing detection method and a passive water depth remote sensing detection method, wherein the active water depth remote sensing detection method mainly comprises airborne Lidar measurement. The theoretical basis for developing the water depth inversion based on the remote sensing technology is as follows: sunlight passes through the atmosphere layer to enter the water body, then the sunlight and various components in the water body are subjected to absorption or scattering, part of scattered light directly passes out of the water surface, part of the scattered light reaches the bottom of the water body and is reflected and then passes out of the water surface after being absorbed and scattered by the water body components, and finally an optical signal containing the information of the water body and the water depth is received by the remote sensing sensor. And analyzing the remote sensing image containing the water depth information, and establishing a functional relation with the actually measured water depth to realize the water depth remote sensing inversion. The mainstream remote sensing water depth inversion model mainly comprises three types, namely a theoretical analysis model, a semi-theoretical semi-empirical model and a statistical analysis model. The theoretical analysis model is based on the radiation transmission process of the water body to construct the model, and has higher precision, but the model parameters are more and are difficult to acquire, so that the actual popularization and application are more difficult. At present, the most widely applied model is a half-theoretical half-experience model, the model is derived by combining the physical relationship between the water depth and the radiance value of the remote sensing image on the basis of the prior data, and a single-band model, a multiband model, a log-linear model, a log-conversion ratio model and the like are more commonly used.
Shallow sea areas have complex environments, such as non-uniformity of the type of seabed substrate, and differences exist in the degree of attenuation of light reaching the bottom; meanwhile, shallow sea is greatly influenced by land source substances, and suspended sediment, yellow substances, phytoplankton and the like cause complex water color elements of offshore water bodies. The existing model has the problem that the water depth inversion error of a shallower region (0-5 m) is higher, and meanwhile, the model is not refined and screened by fully combining the type of a near-shore substrate, the water quality condition, the imaging condition and the like, so that the portability of the model is poor.
Disclosure of Invention
In view of the above, the invention aims to provide a shallow sea water depth remote sensing inversion method based on regional self-adaption, which is established by comprehensively considering the water depth range, the substrate type and the water quality condition by combining the existing research foundation so as to improve the remote sensing inversion precision of offshore water depth to the maximum extent.
The basic idea of regional self-adaptation is to select the optimal model and parameters according to the basic conditions of depth, substrate and water quality of the water depth inversion region, instead of calculating the whole water depth inversion region by using one model, and inversion errors caused by model differences and parameters are maximized.
In order to achieve the above purpose, the technical scheme of the invention is realized as follows:
the shallow sea water depth remote sensing inversion method based on the region self-adaption comprises the following steps:
s1, collecting and processing data;
s2, defining a comprehensive area according to the data collected and processed in the step S1;
s3, constructing a model library, and inducing a water depth remote sensing inversion model into a single-band model, a double-band model, a multi-band model, a band ratio model, a logarithmic conversion ratio model, a neural network model and a machine learning model;
s4, carrying out regional model optimization according to the model library constructed in the step S3;
s5, carrying out water depth inversion according to the region model which is optimized in the step S4.
Further, in step S1, the method includes the following steps:
a1, collecting data: fully collecting various data around a research area, including remote sensing image data, measured water depth data, sea chart data, tide data and substrate type data;
a2, remote sensing image processing: performing geometric correction, radiometric calibration, flare correction, atmosphere correction and land and water segmentation processing on remote sensing image data;
a3, processing water depth data: including sea chart correction and information extraction, and water depth information tidal correction.
Further, in step A2, the method includes the following steps:
(1) Radiation calibration: through radiometric calibration processing, the DN value of the image is converted into an apparent radiance value L:
L＝Gain×DN+offset
the Gain is a Gain coefficient, the Offset is an Offset, and the Gain is obtained from the image header file;
(2) Flare correction: the solar flare of the water meter is caused by the specular reflection of the water meter radiating by the solar water, and is expressed as a bright white patch, so that the real radiation characteristic of the target water body is covered, and the flare removal is carried out by adopting a Hedley method:
wherein L' i The brightness after flare is removed for the ith wave band, L i The irradiation brightness before flare is removed for the ith wave band, theta is the inclination angle of the regression line, L NIR For the radiation brightness of the near-infrared band,
is the minimum value of the near infrared band radiance;
(3) Atmospheric correction: the method is used for eliminating the influence of atmospheric and illumination factors on the reflection of the ground object and obtaining the real reflectivity data of the ground object;
(4) Land and water segmentation: the method is used for preventing land pixels from participating in water depth inversion calculation, improves calculation efficiency, needs to divide water and land, and adopts NDWI indexes to divide water and land:
wherein p is Green And p NIR Reflectance values in the green band and the near infrared band.
Further, in step A3, the method includes the following steps:
(1) Chart correction: if the chart is a paper chart, firstly scanning, and then carrying out geometric correction according to the kilometer-outside information or the reference base chart;
(2) Extracting chart information: extracting water depth points and isodepth line thematic information based on the corrected chart;
(3) And (3) tidal correction: the sea surface height is changed at the moment influenced by the tide action, so that all the source water depth data can be applied to the same time tide level, the measured water depth data is corrected according to the instantaneous tide level in the measurement time and the instantaneous tide level difference of the remote sensing image shooting, and the water depth information acquired by the sea chart is corrected according to the depth reference plane of the sea chart and the instantaneous tide level difference of the remote sensing image shooting.
Further, in step S1, the following cases are included:
b1, partitioning based on water depth conditions: partitioning a research area according to the water depth information and the measured water depth information acquired by the chart at intervals of 5 meters;
b2, partitioning based on substrate type and water quality condition: based on the water depth, the partition is further subdivided by combining the substrate type data and specific water quality conditions of the research area;
b3, comprehensive partition: and (3) expanding the boundaries of the detailed subareas to 10 pixels in the directions of the shallow water area and the deep water area respectively, wherein the boundary of the water depth of 0 meter can not be expanded, comprehensively treating the existence of continental or other unreasonable conditions after the expansion, and finally obtaining comprehensive subarea results as subdivision areas for subsequent water depth inversion.
Further, in step S1, the specific model form is as follows:
c1, single band model
M＝A 0 +A 1 X 1
Wherein M is an inversion water depth value, A 0 And A 1 As model coefficient, X 1 A spectral value for a band;
c2 dual-band model
M＝A 0 +A 1 X 1 +A 2 X 2
Wherein M is an inversion water depth value, A 0 、A 1 And A 2 As model coefficient, X 1 And X 2 For the spectrum value of a certain wave band, two wave bands with stronger penetrability to the water body can be selected for ratio processing;
c3, multiband model
M＝A 0 +A 1 X 1 +A 2 X 2 +…+A n X n
Wherein M is an inversion water depth value, A 0 、A 1 And A n As model coefficient, X 1 、X 2 And X n A spectral value for a band;
c4, band ratio model
Wherein M is an inversion water depth value, A 0 And A 1 As model coefficient, X 1 And X 2 Spectral values of two wavebands respectively;
c5, logarithmic conversion ratio model
Wherein M is an inversion water depth value, A 0 And A 1 As model coefficient, R ω (X 1 ) And R is ω (X 2 ) Reflectance values of two wavebands respectively, a, b, m, n being a model adjustment factor;
c6 and neural network model
The method comprises the steps of BP neural network and LSTM, CNN, RNN, DBN, RBM, firstly establishing a data set matched with remote sensing images and actually measured water depth data in space, then randomly dividing the data set into training data and test data, training and testing a model, and establishing an optimal LSTM water depth inversion model, wherein the operation formula of LSTM neurons is as follows:
i t ＝Sigmoid(W Xi ·X t +W Hi ·H t-1 +b i )
f t ＝Sigmoid(W Xf ·X t +W Hf ·H t-1 +b f )
wherein,,
representing Hadamard product, X t Input for t neurons; h t-1 Output of the t-1 th neuron, W Xi 、W Xf 、W XC 、W Xo Respectively is the wave band X in different operations t Corresponding to weight, W Hi 、W Hf 、W HC 、W Ho Respectively different operation mid-wave band H t-1 Corresponding weights; b i 、b f 、b C 、b o For corresponding operation bias value, i t 、f t 、o t Input, forget and output gating are controlled respectively;
C t the cell state of the t-th neuron is combined with the cell state of the last neuron and new data input by the current neuron to form a new cell state.
Further, in step S1, the method includes the following steps:
d1, selecting water depth points: selecting water depth points according to the partition range, and randomly dividing the water depth points into two parts according to the model construction requirement, wherein one part is used for model construction, and the other part is used for model verification;
d2, sensitive wave band selection: determining a sensitive wave band of regional water depth inversion through remote sensing image spectral feature analysis or through statistical correlation analysis of spectral data and water depth data;
and D3, model construction: constructing a water depth point based on the model, sequentially constructing a single-band model, a double-band model, a multi-band model, a band ratio model, a logarithmic conversion ratio model, a neural network model and a machine learning model, and determining specific parameters of each model;
d4, model verification and optimization: carrying out water depth inversion on the regional remote sensing images one by utilizing each constructed model, verifying water depth points by utilizing the models, carrying out precision evaluation on model inversion results from two aspects of average absolute error and average error, and selecting the model with optimal precision as an inversion model of the region:
wherein Z is i For the water depth value of the ith verification point, ΔZ i Is the water depth error value of the ith verification point.
Further, in step S1, the method includes the following steps:
e1, inversion of regional water depths: carrying out regional water depth inversion by using the optimized regional water depth inversion model to obtain regional water depth inversion results which are D respectively Z1 、D Z2 、D Z3 ……；
E2, regional water depth fusion: splicing the inversion results of the water depths of the subareas, firstly unifying the results with different resolutions, adopting a 'removing-recovering' method to form constant resolution, then carrying out linear weighting on the water depth values of the overlapped areas of the boundaries obtained through external expansion, determining the weights according to the distance from the boundary of the inversion areas, wherein the weights are larger as the weights are more far from the boundary and are closer to the center of the areas, and determining the water depth values in the overlapped areas according to the following formula:
M＝M l ×f l +M r ×(1-f l )
wherein M is the inversion water depth value of a certain position in the overlapping area, M l And Mr is the water depth inversion value of the position in the overlapped area of two adjacent areas, f l For the weight of one of the regions, R is the distance between the position and the boundary of the region, and R is the total width of the overlapping region;
and E3, comprehensive precision analysis: and verifying the water depth points by using all models, and evaluating the accuracy of the fused water depth inversion result from two aspects of MAE and MRE.
Further, the scheme discloses electronic equipment, which comprises a processor and a memory which is in communication connection with the processor and is used for storing executable instructions of the processor, wherein the processor is used for executing a shallow sea water depth remote sensing inversion method based on region self-adaption.
Further, the scheme discloses a computer readable storage medium which stores a computer program, wherein the computer program realizes the shallow sea water depth remote sensing inversion method based on region self-adaption when being executed by a processor.
Compared with the prior art, the shallow sea water depth remote sensing inversion method based on the region self-adaption has the following beneficial effects:
(1) The shallow sea water depth remote sensing inversion method based on the area self-adaption is established by combining the existing research foundation and comprehensively considering the water depth range, the substrate type and the water quality condition, so as to improve the remote sensing inversion precision of the offshore water depth to the maximum extent;
(2) According to the shallow sea water depth remote sensing inversion method based on the region self-adaption, the basic idea of the region self-adaption is to select an optimal model and parameters according to the basic conditions of the depth, the substrate and the water quality of the water depth inversion region instead of calculating the whole water depth inversion region by using one model, and inversion errors caused by model differences and parameters are maximized.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention. In the drawings:
fig. 1 is a schematic diagram of a shallow sea water depth remote sensing inversion method based on region self-adaption according to an embodiment of the invention.
Detailed Description
It should be noted that, without conflict, the embodiments of the present invention and features of the embodiments may be combined with each other.
The invention will be described in detail below with reference to the drawings in connection with embodiments.
As shown in fig. 1, the scheme is designed to combine the existing research foundation, comprehensively consider the water depth range, the substrate type and the water quality condition, and establish a region self-adaptive water depth remote sensing inversion method so as to improve the remote sensing inversion precision of offshore water depth to the greatest extent.
The basic idea of regional self-adaptation is to select the optimal model and parameters according to the basic conditions of depth, substrate and water quality of the water depth inversion region, instead of calculating the whole water depth inversion region by using one model, and inversion errors caused by model differences and parameters are maximized.
The method comprises the following specific steps:
step 1: data collection and processing
1-1: and (3) collecting data: various data including remote sensing image data, measured water depth data, sea chart data, tide data, substrate type data and the like are fully collected around the research area.
1-2: remote sensing image processing: aiming at remote sensing image data, the processing of geometric correction, radiometric calibration, flare correction, atmosphere correction, land and water segmentation and the like are mainly carried out.
(1) Radiation calibration: the satellite sensor receives signals and is influenced by the actions of absorption, scattering, reflection, refraction, transmission and the like of aerosol, cloud particles, water vapor and the like in the atmosphere to cause remote sensing image information distortion, and an image DN value is converted into an apparent radiance value L through radiometric calibration processing.
L＝Gain×DN+Offset
Gain is a Gain coefficient, offset is an Offset, and the Gain is obtained from the header file.
(2) Flare correction. Water meter solar flare is caused by water meter specular reflection of solar water-entering radiation, and is expressed as a bright white patch, so that the real radiation characteristic of a target water body is covered. Flare removal can be performed using the existing Hedley process.
Wherein L' i The brightness after flare is removed for the ith wave band, L i The irradiation brightness before flare is removed for the ith wave band, theta is the inclination angle of the regression line, L NIR For the radiation brightness of the near-infrared band,
is the minimum value of the near infrared band radiance.
(3) Atmospheric correction: the atmospheric correction is mainly used for eliminating the influence of factors such as atmosphere, illumination and the like on the reflection of the ground object and obtaining the real reflectivity data of the ground object. Absolute atmosphere correction methods such as Mortran, lowtran, flaash, 6S, dark pixels and the like are included, and relative correction methods such as histogram matching and a statistical-based invariant target method and the like are also included. If the research area has only one image, the absolute atmosphere correction method is selected as much as possible. If the study area contains multiple views and multiple phases of images, a relative atmosphere correction method is recommended to be selected and normalized.
(4) And (5) land and water segmentation. In order to prevent land pixels from participating in water depth inversion calculation and improve calculation efficiency, land and water segmentation is needed. The NDWI index was used for land and water segmentation.
Wherein p is Green And p NIR Reflectance values in the green band and the near infrared band.
1-3: and (3) processing water depth data: including sea chart correction and information extraction, water depth information tidal correction, and the like.
(1) Chart correction: if the chart is a paper chart, firstly scanning is carried out, and then geometric correction is carried out according to kilometer-outside information or a reference base chart.
(2) Extracting chart information: and extracting thematic information such as water depth points, equal depth lines and the like based on the corrected chart.
(3) And (3) tidal correction: the sea level is changed at any time under the influence of tidal action, so that the water depth data of all sources are required to be unified on the same tide level at the same time for application. The measured water depth data needs to be corrected according to the instantaneous tide level in the measurement time and the tide level difference in the shooting moment of the remote sensing image, and the water depth information acquired by the sea chart needs to be corrected according to the depth reference plane of the sea chart and the tide level difference in the shooting moment of the remote sensing image.
Step 2: comprehensive region delineation
2-1: partitioning based on water depth conditions: according to the water depth information and measured water depth information obtained from the sea chart, the research area is partitioned according to 5m intervals, and the following division range is defined: z is Z d1 ∈[0,5)、Z d2 ∈[5,10)、Z d3 ∈[10,15)、Z d4 ∈[15,20)、Z d5 ∈[20,25)、Z d6 ∈[25,30)……
2-2: partitioning based on substrate type and water quality conditions: based on the water depth, the partition is further subdivided by combining the substrate type data and specific water quality conditions of the research area. For example, at Z d1 In the subareas, if the substrate types or partial areas with more obvious differences have more obvious water quality differences, Z is determined according to the substrate types or the water quality conditions d1 The partitions are subdivided. Finally, the detailed partition of the comprehensive substrate type and water quality condition is obtained and is named Z in turn 1 、Z 2 、Z 3 ……。
2-3: comprehensive partition: and (3) expanding the boundaries of the detailed subareas by 10 pixels (remote sensing images) towards the shallow water area and the deep water area respectively, wherein the 0 meter water depth boundary can not be expanded, comprehensively treating the existence of continents or other unreasonable conditions after the expansion, and finally obtaining comprehensive subarea results as subdivision areas for subsequent water depth inversion.
Step 3: model library construction
The water depth remote sensing inversion model is generalized into a single-band model, a double-band model, a multi-band model, a band ratio model, a logarithmic conversion ratio model, a neural network model, a machine learning model and the like through the current state analysis of domestic and foreign researches. The specific model form is as follows:
(1) Single-band model
M＝A 0 +A 1 X 1
M is the inversion water depth value, A 0 And A 1 As model coefficient, X 1 Is a spectral value (which may be a derivative or a logarithmically processed value) of a certain band.
(2) Dual band model
M＝A 0 +A 1 X 1 +A 2 X 2
M is the inversion water depth value, A 0 、A 1 And A 2 As model coefficient, X 1 And X 2 For spectral values of a certain band (preferably differential or pairThe value after the number processing) or can be processed by using two wave bands with stronger penetrability to the water body.
(3) Multiband model
M＝A 0 +A 1 X 1 +A 2 X 2 +…+A n X n
M is the inversion water depth value, A 0 、A 1 And A n As model coefficient, X 1 、X 2 And X n Is a spectral value (which may be a derivative or a logarithmically processed value) of a certain band.
(4) Wave band ratio model
M is the inversion water depth value, A 0 And A 1 As model coefficient, X 1 And X 2 Spectral values of the two bands, respectively.
(5) Logarithmic conversion ratio model
M is the inversion water depth value, A 0 And A 1 As model coefficient, R ω (X 1 ) And R is ω (X 2 ) Reflectance values of the two bands respectively, a, b, m, n is a model adjustment factor.
(6) Neural network model
There are BP neural networks, LSTM, CNN, RNN, DBN, RBM, etc., of which the most commonly used are LSTM networks, and LSTM models can effectively solve the problem of gradient explosion. Firstly, a data set of remote sensing images and actually measured water depth data which are matched in space is established, then the data set is randomly divided into training data and test data, training and testing of a model are carried out, and an optimal LSTM water depth inversion model is established. The operational formula of LSTM neurons is:
i t ＝Sigmoid(W Xi ·X t +W Hi ·H t-1 +b i )
f t ＝Sigmoid(W Xf ·X t +W Hf ·H t-1 +b f )
wherein,,
representing Hadamard product, X t Input for t neurons; h t-1 Output of the t-1 th neuron, W Xi 、W Xf 、W XC 、W Xo Respectively is the wave band X in different operations t Corresponding to weight, W Hi 、W Hf 、W HC 、W Ho Respectively different operation mid-wave band H t-1 Corresponding weights; b i 、b f 、b C 、b o For corresponding operation bias value, i t 、f t 、o t Input, forget and output gating are controlled respectively; c (C) t The cell state of the t-th neuron is combined with the cell state of the last neuron and new data input by the current neuron to form a new cell state.
Step 4: region model preference
4-1: and (3) selecting a water depth point: and selecting water depth points according to the partition range, and randomly dividing the water depth points into two parts according to the model construction requirement, wherein one part is used for model construction and the other part is used for model verification.
4-2: sensitive wave band selection: and determining a sensitive wave band of regional water depth inversion through remote sensing image spectral feature analysis or through statistical correlation analysis of spectral data and water depth data.
4-3: model construction: and constructing a water depth point based on the model, sequentially constructing a single-band model, a double-band model, a multi-band model, a band ratio model, a logarithmic conversion ratio model, a neural network model and a machine learning model, and determining specific parameters of each model.
4-4: model verification and preference: and carrying out water depth inversion on the regional remote sensing images one by utilizing each constructed model, verifying water depth points by utilizing the models, carrying out precision evaluation on model inversion results from two aspects of average absolute error (Mean Absolutely Error, MAE) and average error (Mean Relative Error, MRE), and selecting the model with optimal precision as an inversion model of the region.
Wherein Z is i For the water depth value of the ith verification point, ΔZ i Is the water depth error value of the ith verification point.
Step 5: depth inversion
5-1: zonal depth inversion: and carrying out regional water depth inversion by using the optimized regional water depth inversion model to obtain a regional water depth inversion result. Respectively D Z1 、D Z2 、D Z3 ……。
5-2: zonal water depth fusion: splicing the water depth inversion results of the subareas, firstly unifying the results with different resolutions, and forming constant resolution by adopting a 'removal-recovery' method. And then carrying out linear weighting on the water depth value of the boundary overlapping region obtained through the expansion, wherein the weight is determined according to the distance from the boundary of the inversion region, and the farther the weight is from the boundary, the greater the weight is, and the water depth value in the overlapping region is determined according to the following formula:
M＝M l ×f l +M r ×(1-f l )
wherein M is the inversion water depth value of a certain position in the overlapping area, M l And Mr is the water depth inversion value of the position in the overlapped area of two adjacent areas, f l And R is the distance between the position and the boundary of the region, and R is the total width of the overlapped region.
5-3: and (3) comprehensive precision analysis: and verifying the water depth points by using all models, and evaluating the accuracy of the fused water depth inversion result from two aspects of MAE and MRE.
Those of ordinary skill in the art will appreciate that the elements and method steps of each example described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the elements and steps of each example have been described generally in terms of functionality in the foregoing description to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the several embodiments provided in this application, it should be understood that the disclosed methods and systems may be implemented in other ways. For example, the above-described division of units is merely a logical function division, and there may be another division manner when actually implemented, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted or not performed. The units may or may not be physically separate, and components shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiment of the present invention.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention, and are intended to be included within the scope of the appended claims and description.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (10)
1. The shallow sea water depth remote sensing inversion method based on the area self-adaption is characterized by comprising the following steps of:
s1, collecting and processing data;
s2, defining a comprehensive area according to the data collected and processed in the step S1;
s3, constructing a model library, and inducing a water depth remote sensing inversion model into a single-band model, a double-band model, a multi-band model, a band ratio model, a logarithmic conversion ratio model, a neural network model and a machine learning model;
s4, carrying out regional model optimization according to the model library constructed in the step S3;
s5, carrying out water depth inversion according to the region model which is optimized in the step S4.
2. The shallow sea water depth remote sensing inversion method based on the area adaptation according to claim 1, wherein in step S1, the method comprises the steps of:
a1, collecting data: fully collecting various data around a research area, including remote sensing image data, measured water depth data, sea chart data, tide data and substrate type data;
a2, remote sensing image processing: performing geometric correction, radiometric calibration, flare correction, atmosphere correction and land and water segmentation processing on remote sensing image data;
a3, processing water depth data: including sea chart correction and information extraction, and water depth information tidal correction.
3. The shallow sea water depth remote sensing inversion method based on the area adaptation according to claim 2, wherein in step A2, the method comprises the following steps:
(1) Radiation calibration: through radiometric calibration processing, the DN value of the image is converted into an apparent radiance value L:
L＝Gain×DN+Offset
the Gain is a Gain coefficient, the Offset is an Offset, and the Gain is obtained from the image header file;
(2) Flare correction: the solar flare of the water meter is caused by the specular reflection of the water meter radiating by the solar water, and is expressed as a bright white patch, so that the real radiation characteristic of the target water body is covered, and the flare removal is carried out by adopting a Hedley method:
wherein L' i The brightness after flare is removed for the ith wave band, L i The irradiation brightness before flare is removed for the ith wave band, theta is the inclination angle of the regression line, L NIR For the radiation brightness of the near-infrared band,
is the minimum value of the near infrared band radiance;
(3) Atmospheric correction: the method is used for eliminating the influence of atmospheric and illumination factors on the reflection of the ground object and obtaining the real reflectivity data of the ground object;
(4) Land and water segmentation: the method is used for preventing land pixels from participating in water depth inversion calculation, improves calculation efficiency, needs to divide water and land, and adopts NDWI indexes to divide water and land:
wherein p is Green And p NIR Reflectance values in the green band and the near infrared band.
4. The shallow sea water depth remote sensing inversion method based on the area adaptation according to claim 1, wherein in step A3, the method comprises the following steps:
(1) Chart correction: if the chart is a paper chart, firstly scanning, and then carrying out geometric correction according to the kilometer-outside information or the reference base chart;
(2) Extracting chart information: extracting water depth points and isodepth line thematic information based on the corrected chart;
(3) And (3) tidal correction: the water depth data of all sources can be unified to the same time of tide level for application, the actually measured water depth data is corrected according to the instantaneous tide level in the measurement time and the instantaneous tide level difference of the remote sensing image shooting, and the water depth information obtained by the sea chart is corrected according to the depth reference plane of the sea chart and the instantaneous tide level difference of the remote sensing image shooting.
5. The shallow sea water depth remote sensing inversion method based on the area adaptation according to claim 1, wherein in step S1, the following cases are included:
b1, partitioning based on water depth conditions: partitioning a research area according to the water depth information and the measured water depth information acquired by the chart at intervals of 5 meters;
b2, partitioning based on substrate type and water quality condition: based on the water depth, the partition is further subdivided by combining the substrate type data and specific water quality conditions of the research area;
b3, comprehensive partition: and (3) expanding the boundaries of the detailed subareas to 10 pixels in the directions of the shallow water area and the deep water area respectively, wherein the boundary of the water depth of 0 meter can not be expanded, comprehensively treating the condition that the expanded water is on the land, and finally obtaining comprehensive subarea results as subdivision areas for subsequent water depth inversion.
6. The shallow sea water depth remote sensing inversion method based on area adaptation according to claim 1, wherein in step S1, a specific model form is as follows:
c1, single band model
M＝A 0 +A 1 X 1
Wherein M is an inversion water depth value, A 0 And A 1 As model coefficient, X 1 A spectral value for a band;
c2 dual-band model
M＝A 0 +A 1 X 1 +A 2 X 2
Wherein M is an inversion water depth value, A 0 、A 1 And A 2 As model coefficient, X 1 And X 2 For the spectrum value of a certain wave band, two wave bands with stronger penetrability to the water body can be selected for ratio processing;
c3, multiband model
M＝A 0 +A 1 X 1 +A 2 X 2 +…+A n X n
Wherein M is an inversion water depth value, A 0 、A 1 And A n As model coefficient, X 1 、X 2 And X n A spectral value for a band;
c4, band ratio model
Wherein M is an inversion water depth value, A 0 And A 1 As model coefficient, X 1 And X 2 Spectral values of two wavebands respectively;
c5, logarithmic conversion ratio model
Wherein M is an inversion water depth value, A 0 And A 1 As model coefficient, R ω (X 1 ) And R is ω (X 2 ) Reflectance values of two wavebands respectively, a, b, m, n being a model adjustment factor;
c6 and neural network model
The method comprises the steps of BP neural network and LSTM, CNN, RNN, DBN, RBM, firstly establishing a data set matched with remote sensing images and actually measured water depth data in space, then randomly dividing the data set into training data and test data, training and testing a model, and establishing an optimal LSTM water depth inversion model, wherein the operation formula of LSTM neurons is as follows:
i t ＝Sigmoid(W Xi ·X t +W Hi ·H t-1 +b i )
f t ＝Sigmoid(W Xf ·X t +W Hf ·H t-1 +b f )
wherein,,
representing Hadamard product, X t Input for t neurons; h t-1 Output of the t-1 th neuron, W Xi 、W xf 、W XC 、W Xo Respectively is the wave band X in different operations t Corresponding to weight, W Hi 、W Hf 、W HC 、W Ho Respectively different operation mid-wave band H t-1 Corresponding weights; b i 、b f 、b C 、b o For corresponding operation bias value, i t 、f t 、o t Input, forget and output gating are controlled respectively;
C t the cell state of the t-th neuron is combined with the cell state of the last neuron and new data input by the current neuron to form a new cell state.
7. The shallow sea water depth remote sensing inversion method based on the area adaptation according to claim 1, wherein in step S1, the method comprises the steps of:
d1, selecting water depth points: selecting water depth points according to the partition range, and randomly dividing the water depth points into two parts according to the model construction requirement, wherein one part is used for model construction, and the other part is used for model verification;
d2, sensitive wave band selection: determining a sensitive wave band of regional water depth inversion through remote sensing image spectral feature analysis or through statistical correlation analysis of spectral data and water depth data;
and D3, model construction: constructing a water depth point based on the model, sequentially constructing a single-band model, a double-band model, a multi-band model, a band ratio model, a logarithmic conversion ratio model, a neural network model and a machine learning model, and determining specific parameters of each model;
d4, model verification and optimization: carrying out water depth inversion on the regional remote sensing images one by utilizing each constructed model, verifying water depth points by utilizing the models, carrying out precision evaluation on model inversion results from two aspects of average absolute error and average error, and selecting the model with optimal precision as an inversion model of the region:
wherein Z is i For the water depth value of the ith verification point, ΔZ i Is the water depth error value of the ith verification point.
8. The shallow sea water depth remote sensing inversion method based on the area adaptation according to claim 1, wherein in step S1, the method comprises the steps of:
e1, inversion of regional water depths: carrying out regional water depth inversion by using the optimized regional water depth inversion model to obtain regional water depth inversion results which are D respectively Z1 、D Z2 、D Z3 ……；
E2, regional water depth fusion: splicing the inversion results of the water depths of the subareas, firstly unifying the results with different resolutions, adopting a 'removing-recovering' method to form constant resolution, then carrying out linear weighting on the water depth values of the overlapped areas of the boundaries obtained through external expansion, determining the weights according to the distance from the boundary of the inversion areas, wherein the weights are larger as the weights are more far from the boundary and are closer to the center of the areas, and determining the water depth values in the overlapped areas according to the following formula:
M＝M l ×f l +M r ×(1-f l )
wherein M is the inversion water depth value of a certain position in the overlapping area, M 1 And Mr is the water depth inversion value of the position in the overlapped area of two adjacent areas, f l For the weight of one of the regions, R is the distance between the position and the boundary of the region, and R is the total width of the overlapping region;
and E3, comprehensive precision analysis: and verifying the water depth points by using all models, and evaluating the accuracy of the fused water depth inversion result from two aspects of MAE and MRE.
9. An electronic device comprising a processor and a memory communicatively coupled to the processor for storing processor-executable instructions, characterized in that: the processor is used for executing the shallow sea water depth remote sensing inversion method based on the area self-adaption according to any one of claims 1-8.
10. A computer-readable storage medium storing a computer program, characterized in that: the computer program, when executed by a processor, implements the shallow sea water depth remote sensing inversion method based on regional adaptation of any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310132412.XA CN116295285A (en) | 2023-02-14 | 2023-02-14 | Shallow sea water depth remote sensing inversion method based on region self-adaption |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310132412.XA CN116295285A (en) | 2023-02-14 | 2023-02-14 | Shallow sea water depth remote sensing inversion method based on region self-adaption |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116295285A true CN116295285A (en) | 2023-06-23 |
Family
ID=86800512
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310132412.XA Pending CN116295285A (en) | 2023-02-14 | 2023-02-14 | Shallow sea water depth remote sensing inversion method based on region self-adaption |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116295285A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117496278A (en) * | 2024-01-03 | 2024-02-02 | 自然资源部第二海洋研究所 | Water depth map inversion method based on radiation transmission parameter application convolutional neural network |
| CN117523321A (en) * | 2024-01-03 | 2024-02-06 | 自然资源部第二海洋研究所 | Optical shallow water classification method based on passive remote sensing spectral image application neural network |
| CN117975255A (en) * | 2024-04-02 | 2024-05-03 | 国家海洋信息中心 | Shallow sea bottom type identification method for multispectral remote sensing image |
-
2023
- 2023-02-14 CN CN202310132412.XA patent/CN116295285A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117496278A (en) * | 2024-01-03 | 2024-02-02 | 自然资源部第二海洋研究所 | Water depth map inversion method based on radiation transmission parameter application convolutional neural network |
| CN117523321A (en) * | 2024-01-03 | 2024-02-06 | 自然资源部第二海洋研究所 | Optical shallow water classification method based on passive remote sensing spectral image application neural network |
| CN117496278B (en) * | 2024-01-03 | 2024-04-05 | 自然资源部第二海洋研究所 | Water depth map inversion method based on radiation transmission parameter application convolutional neural network |
| CN117523321B (en) * | 2024-01-03 | 2024-04-09 | 自然资源部第二海洋研究所 | Optical shallow water classification method based on passive remote sensing spectral image application neural network |
| CN117975255A (en) * | 2024-04-02 | 2024-05-03 | 国家海洋信息中心 | Shallow sea bottom type identification method for multispectral remote sensing image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20230316555A1 (en) | System and Method for Image-Based Remote Sensing of Crop Plants | |
| CN116295285A (en) | Shallow sea water depth remote sensing inversion method based on region self-adaption | |
| CN112213287B (en) | Coastal beach salinity inversion method based on remote sensing satellite image | |
| CN109376600A (en) | Multi-spectrum remote sensing image comprehensive characteristics cloud detection method of optic and device | |
| CN108596213A (en) | A kind of Classification of hyperspectral remote sensing image method and system based on convolutional neural networks | |
| CN109726705B (en) | Mangrove forest information extraction method and device and electronic equipment | |
| CN106548146A (en) | Ground mulching change algorithm and system based on space-time analysis | |
| WO2001033505A2 (en) | Multi-variable model for identifying crop response zones in a field | |
| CN112285710B (en) | Multi-source remote sensing reservoir water storage capacity estimation method and device | |
| CN110189043B (en) | Usable land resource analysis system based on high-score satellite remote sensing data | |
| CN112836725A (en) | Weak supervision LSTM recurrent neural network rice field identification method based on time sequence remote sensing data | |
| CN112052757B (en) | Method, device, equipment and storage medium for extracting fire trace information | |
| CN112013822A (en) | Multispectral remote sensing water depth inversion method based on improved GWR model | |
| Qiao et al. | Estimating maize LAI by exploring deep features of vegetation index map from UAV multispectral images | |
| Lakmal et al. | Brown planthopper damage detection using remote sensing and machine learning | |
| CN117152636A (en) | Shallow sea substrate reflectivity remote sensing monitoring method based on dual-band relation | |
| Liang et al. | Automatic remote sensing detection of floating macroalgae in the yellow and east china seas using extreme learning machine | |
| CN117115669B (en) | Object-level ground object sample self-adaptive generation method and system with double-condition quality constraint | |
| CN113534083B (en) | SAR-based corn stubble mode identification method, device and medium | |
| CN114199800A (en) | Method, system, equipment and medium for identifying rice sheath blight | |
| CN108198178B (en) | Method and device for determining atmospheric range radiation value | |
| CN104198397B (en) | The method that chamber crop nutrient content is detected under N P and K reciprocation | |
| CN117557897A (en) | Lodging monitoring method and device for target crops, electronic equipment and storage medium | |
| CN117035066A (en) | Ground surface temperature downscaling method coupling geographic weighting and random forest | |
| CN116469000A (en) | Inversion method and device for forest ground biomass and leaf area index |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |