CN105738293A - Remote sensing quantitative reversion method and system of crop physical and chemical parameters - Google Patents

Abstract

The invention discloses a remote sensing quantitative reversion method and system of crop physical and chemical parameters.According to the method, LUTs is segmented into LUTsi of the same number according to the spectral band number, all the LUTsi are ranked to obtain LUTsi_sort, for each LUTsi_sort, a bisection method searching method is utilized for searching for simulating vegetation canopy reflectivity data with similarity to real vegetation canopy reflectivity data meeting the preset requirement, all the simulating vegetation canopy reflectivity data and content of corresponding items of the data in the LUTs are used for constructing a form in a corresponding relation form to obtain LUTs_new, each simulating vegetation canopy reflectivity data and TD of the real vegetation canopy reflectivity data corresponding to the simulating vegetation canopy reflectivity data in the LUTs_new are calculated, and crop physical and chemical parameter data corresponding to the minimum value in the TD serves as the optimal solution of the lookup table algorithm.According to the method and the system, the size and the direction character of the vegetation canopy reflectivity data are considered comprehensively, and the calculation complexity is reduced while the crop physical and chemical parameter inversion numerical precision is improved.

Description

The remote sensing quantitative inversion method of a kind of crop physical and chemical parameter and system

Technical field

The present invention relates to crop physical and chemical parameter technical field, in particular, relate to remote sensing quantitative inversion method and the system of a kind of crop physical and chemical parameter.

Background technology

Crop physical and chemical parameter is the class key index characterizing crop pattern structure and physiologically active, including plant physiology parameter (such as population density, plant height, leaf area index etc.) and crop biochemical parameter (such as chlorophyll content, nitrogen content, carotenoid content etc.), crop physical and chemical parameter, in the crop groups Growing state survey in precision agriculture field, quality estimation, production forecast, pestforecasting and crop field management policy development and enforcement etc., all has extensive and important theoretical significance and application value realistic.

At present, the remote sensing quantitative inversion method of crop physical and chemical parameter is broadly divided into statistics class method and mechanism class method, statistical relationship between statistics class method Main Basis crop physical and chemical parameter and remote sensing observations data builds quantitative inversion model, and the method is easy to use, but universality is poor, portability is weak；Mechanism class method is from physical mechanisms such as crop spectral responses, choose the mechanism class models such as vegetation radiative transfer model or crop growth model, the morphosis of quantitative description crop and physiologically active, this type of method explicit physical meaning, clear mathematical logic, universality are strong.Mechanism class method is most commonly used that vegetation radiative transfer model PROSAIL reverse mode, and the classical numerical algorithm the most easy to use realizing vegetation radiative transfer model PROSAIL reverse mode is to look for table algorithm, lookup table algorithm is from proposing to enjoy so far the concern of domestic and international researcher.

The principle of employing lookup table algorithm remote sensing quantitative inversion crop physical and chemical parameter is: first, build bigger crop physical and chemical parameter look-up table；Then, based on the forward mode of PROSAIL model, quantitative simulation obtains Vegetation canopy reflectance look-up table, wherein, and canopy reflectance spectrum data and crop physical and chemical parameter one_to_one corresponding；Finally, for real Vegetation canopy reflectivity data, search simulation Vegetation canopy reflectivity data the most similar with it, and using the crop physical and chemical parameter of its correspondence as quantitative inversion result, this result is also referred to as the optimal solution of crop physical and chemical parameter quantitative inversion.The final goal of lookup table algorithm is to find the optimal solution of crop physical and chemical parameter quantitative inversion, its key point and difficult point are in that the lookup precision of optimal solution and search speed, wherein, lookup precision is affected by the similarity quantitative assessing index of Vegetation canopy reflectivity data, and lookup speed is affected by the data search method of optimal solution.In existing lookup table algorithm, the most frequently used data similarity quantitative assessing index is root-mean-square error.This index quantification describes the difference in size between true and simulation Vegetation canopy reflectivity data, but ignores the direction difference between Vegetation canopy reflectivity data, therefore searches precision and is not as high；The most frequently used data search method is full traversal search method.Full traversal search method can find globally optimal solution, but the computation complexity of the method is higher, and required time is longer, so the demand of the timely fast quantification inverting of crop physical and chemical parameter cannot be met, is subject to bigger limitation in real world applications.

To sum up, how to provide a kind of inverting numerical precision high and the remote sensing quantitative inversion method of crop physical and chemical parameter that complexity is low and system are those skilled in the art's technical problems urgently to be resolved hurrily.

Summary of the invention

In view of this, the present invention provides remote sensing quantitative inversion method and the system of a kind of crop physical and chemical parameter, to realize reducing computation complexity while improving inverting numerical precision, thus shortening search time, improves inverting efficiency.

A kind of remote sensing quantitative inversion method of crop physical and chemical parameter, including:

Will simulation Vegetation canopy reflectance look-up table LUT_sIt is divided into N according to spectral band_bIndividual Vegetation canopy reflectance look-up table LUT_si, wherein, N_bRepresent described LUT_sThe spectral band number of middle Vegetation canopy reflectivity data, described LUT_sFor according to PROSAIL model forward mode simulate obtain with crop physical and chemical parameter look-up table LUT_pCorresponding simulation Vegetation canopy reflectance look-up table；

According to each described LUT_siIn the size of spectral band reflectivity values corresponding to the Vegetation canopy reflectance that comprises, require each described LUT according to predetermined order_siIt is ranked up, the Vegetation canopy reflectance look-up table LUT after being sorted_{si_sort}；

For LUT each described_{si_sort}, utilize binary search method, find out NL_iIndividual and true Vegetation canopy reflectivity data R_mI the similarity of () meets the simulation Vegetation canopy reflectivity data of preset requirement, wherein, and described NL_iAccording to the simulation Vegetation canopy reflectivity data sum TNL and weight W utilizing binary search method to find out_iCalculating obtains, 0≤W_i≤ 1,1≤i≤N_b；

By each described simulation Vegetation canopy reflectivity data and described simulation Vegetation canopy reflectivity data at described LUT_sThe content of middle corresponding entry builds form with the form of corresponding relation, obtains dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}；

Calculate described LUT_{s_new}In each simulation Vegetation canopy reflectivity data and the reflectance comprehensive differences index TD of the true Vegetation canopy reflectivity data corresponding with this simulation Vegetation canopy reflectivity data；

Minima is found out from calculated all described TD, and using the crop physical and chemical parameter data corresponding for the described minima optimal solution as lookup table algorithm.

Preferably, described NL_iCalculating process include formula (1) and formula (2):

NL_i=round (W_i×TNL)(1)；

In formula, described W_iRepresent weight, 0≤W_i≤ 1,1≤i≤N_b, described TNL represents the simulation Vegetation canopy reflectivity data sum utilizing binary search method to find out and described TNL < N, and described N represents described LUT_sIn entry number, round represents bracket function；

$W_i=\frac{{\left(\frac{R_m\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{{\mathrm{\&Sigma;}}_{k=1}^{N_b}{\left(\frac{R_m\left(k\right)}{{\mathrm{\&sigma;}}_s\left(k\right)}\right)}^2}\&times;100\%---\left(2\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_mK () represents R_mAt the reflectance of kth spectral band, σ_sI () represents R_sThe standard deviation of (i), σ_sK () represents R_sThe standard deviation of (k), 1≤i≤N_b, 1≤k≤N_b。

Preferably, the calculating process of described TD includes formula (3), formula (4) and formula (5):

TD=α × NRMSE+ (1-α) × DCC (3)；

In formula, α represents weight, 0≤α≤1, and NRMSE represents normalization root-mean-square error；DCC represents direction variation coefficient；

$NRMSE=\sqrt{\frac{{\mathrm{\&Sigma;}}_{i=1}^{N_b}{\left(\frac{R_m\left(i\right)-R_s\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{N_b}}---\left(4\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_sI () represents R_sAt the reflectance of i-th spectral band, σ_sI () represents R_sThe standard deviation of (i), 1≤i≤N_b；

$DCC=\frac{sum\left(abs\right(\mathrm{sgn}\left({\mathrm{RC}}_m\right)-\mathrm{sgn}\left({\mathrm{RC}}_s\right)\left)\right)}{\frac{N_b}{2}}---\left(5\right);$

In formula, RC_m=(R_m(2)-R_m(1), R_m(3)-R_m(2) ..., R_m(N_b)-R_m(N_b-1)), RC_s=(R_s(2)-R_s(1), R_s(3)-R_s(2) ..., R_s(N_b)-R_s(N_b-1)), 1≤i≤N_b, sum represents that summing function, abs represent the function that takes absolute value, and sgn represents sign function.

A kind of remote sensing quantitative inversion system of crop physical and chemical parameter, including:

Cutting unit, for simulating Vegetation canopy reflectance look-up table LUT_sIt is divided into N according to spectral band_bIndividual Vegetation canopy reflectance look-up table LUT_si, wherein, N_bRepresent described LUT_sThe spectral band number of middle Vegetation canopy reflectivity data, described LUT_sFor according to PROSAIL model forward mode simulate obtain with crop physical and chemical parameter look-up table LUT_pCorresponding simulation Vegetation canopy reflectance look-up table；

Sequencing unit, for according to each described LUT_siIn the size of spectral band reflectivity values corresponding to the Vegetation canopy reflectance that comprises, require each described LUT according to predetermined order_siIt is ranked up, the Vegetation canopy reflectance look-up table LUT after being sorted_{si_sort}；

First searches unit, for for LUT each described_{si_sort}, utilize binary search method, find out NL_iIndividual and true Vegetation canopy reflectivity data R_mI the similarity of () meets the simulation Vegetation canopy reflectivity data of preset requirement, wherein, and described NL_iAccording to the simulation Vegetation canopy reflectivity data sum TNL and weight W utilizing binary search method to find out_iCalculating obtains, 0≤W_i≤ 1,1≤i≤N_b；

Form construction unit, for by each described simulation Vegetation canopy reflectivity data and described simulation Vegetation canopy reflectivity data at described LUT_sThe content of middle corresponding entry builds form with the form of corresponding relation, obtains dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}；

Computing unit, is used for calculating described LUT_{s_new}In each simulation Vegetation canopy reflectivity data and the reflectance comprehensive differences index TD of the true Vegetation canopy reflectivity data corresponding with this simulation Vegetation canopy reflectivity data；

Second searches unit, for finding out minima from calculated all described TD, and using the crop physical and chemical parameter data corresponding for the described minima optimal solution as lookup table algorithm.

Preferably, described NL_iCalculating process include formula (1) and formula (2):

NL_i=round (W_i×TNL)(1)；

In formula, described W_iRepresent weight, 0≤W_i≤ 1,1≤i≤N_b, described TNL represents the simulation Vegetation canopy reflectivity data sum utilizing binary search method to find out and described TNL < N, and described N represents described LUT_sIn entry number, round represents bracket function；

$W_i=\frac{{\left(\frac{R_m\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{{\mathrm{\&Sigma;}}_{k=1}^{N_b}{\left(\frac{R_m\left(k\right)}{{\mathrm{\&sigma;}}_s\left(k\right)}\right)}^2}\&times;100\%---\left(2\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_mK () represents R_mAt the reflectance of kth spectral band, σ_sI () represents R_sThe standard deviation of (i), σ_sK () represents R_sThe standard deviation of (k), 1≤i≤N_b, 1≤k≤N_b。

Preferably, the calculating process of described TD includes formula (3), formula (4) and formula (5):

TD=α × NRMSE+ (1-α) × DCC (3)；

In formula, α represents weight, 0≤α≤1, and NRMSE represents normalization root-mean-square error；DCC represents direction variation coefficient；

$NRMSE=\sqrt{\frac{{\mathrm{\&Sigma;}}_{i=1}^{N_b}{\left(\frac{R_m\left(i\right)-R_s\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{N_b}}---\left(4\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_sI () represents R_sAt the reflectance of i-th spectral band, σ_sI () represents R_sThe standard deviation of (i), 1≤i≤N_b；

$DCC=\frac{sum\left(abs\right(\mathrm{sgn}\left({\mathrm{RC}}_m\right)-\mathrm{sgn}\left({\mathrm{RC}}_s\right)\left)\right)}{\frac{N_b}{2}}---\left(5\right);$

In formula, RC_m=(R_m(2)-R_m(1), R_m(3)-R_m(2) ..., R_m(N_b)-R_m(N_b-1)), RC_s=(R_s(2)-R_s(1), R_s(3)-R_s(2) ..., R_s(N_b)-R_s(N_b-1)), 1≤i≤N_b, sum represents that summing function, abs represent the function that takes absolute value, and sgn represents sign function.

From above-mentioned technical scheme it can be seen that the invention provides the remote sensing quantitative inversion method of a kind of crop physical and chemical parameter and system, will simulation Vegetation canopy reflectance look-up table LUT_sSpectral band number according to its Vegetation canopy reflectivity data comprised, is divided into and spectral band number equal number of Vegetation canopy reflectance look-up table LUT_si, by each LUT_siRequire that sequence obtains LUT according to predetermined order_{si_sort}, for each LUT_{si_sort}Utilize binary search method, find out and meet the simulation Vegetation canopy reflectivity data of preset requirement with true Vegetation canopy reflectivity data similarity, by each simulation Vegetation canopy reflectivity data and these data at LUT_sThe content of middle corresponding entry builds form with the form of corresponding relation, obtains dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}, calculate LUT_{s_new}In each simulation Vegetation canopy reflectivity data and the reflectance comprehensive differences index TD of the true Vegetation canopy reflectivity data corresponding with this reflectivity data, and using the crop physical and chemical parameter data corresponding for minima in the TD optimal solution as lookup table algorithm.Can be seen that, the present invention makes full use of remote sensing, mathematics, the multidisciplinary advantage such as computer, consider the size and Orientation characteristic of Vegetation canopy reflectivity data, build high-precision true Vegetation canopy reflectivity data and simulation Vegetation canopy reflectivity data similarity quantitative assessing index, from data search angle, select suitable mathematical model optimizing data search strategy, introduce feasible computerized algorithm and complete data search process, computation complexity is reduced while improving inverting numerical precision, thus shortening search time, improve inverting efficiency, meet the remote sensing quantitative inversion high accuracy of crop physical and chemical parameter and quick demand.

Accompanying drawing explanation

In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, the accompanying drawing used required in embodiment or description of the prior art will be briefly described below, apparently, accompanying drawing in the following describes is only embodiments of the invention, for those of ordinary skill in the art, under the premise not paying creative work, it is also possible to obtain other accompanying drawing according to the accompanying drawing provided.

Fig. 1 is the remote sensing quantitative inversion method flow chart of a kind of crop physical and chemical parameter disclosed in the embodiment of the present invention；

Fig. 2 is the structural representation of the remote sensing quantitative inversion system of a kind of crop physical and chemical parameter disclosed in the embodiment of the present invention.

Detailed description of the invention

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is only a part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, the every other embodiment that those of ordinary skill in the art obtain under not making creative work premise, broadly fall into the scope of protection of the invention.

For meeting the application demands such as precision agriculture Growing state survey, quality estimation, production forecast, pestforecasting, there is provided data support and theoretical method to instruct for crop field management policy development and enforcement etc., build a kind of high accuracy, quickly and the remote sensing quantitative inversion method of the crop physical and chemical parameter of universality necessary.

In crop physical and chemical parameter quantitative inversion field, classical lookup table algorithm is easy to use, universality is strong, but in the lookup precision and lookup speed of crop physical and chemical parameter optimal solution, still need to further improvement, to agree with crop physical and chemical parameter remote sensing quantitative inversion to high accuracy and high efficiency real world applications demand.The present invention is dissecting on the basis of the true Vegetation canopy reflectivity data of existing quantitative description and simulation Vegetation canopy reflectivity data similarity evaluation index, proposition can the evaluation index of quantitative description Vegetation canopy reflectivity data direction difference and direction variation coefficient, and it is combined with the evaluation index describing Vegetation canopy reflectivity data difference in size and normalization root-mean-square error, build comprehensive and quantitative and describe the new types of data similarity evaluation index of Vegetation canopy reflectivity data size and Orientation difference；Optimize data search strategy, for the rapidly and efficiently data search method of subordinate ordered array and binary search in Import computer subject, quantitatively search for the optimal solution of crop physical and chemical parameter.The present invention, from theory analysis and numerical value practical term, discloses remote sensing quantitative inversion method and the system of a kind of crop physical and chemical parameter, to reduce computation complexity while realizing improving inverting numerical precision, thus shortening search time, improves inverting efficiency.

It is specific as follows that technical scheme comprises content:

Referring to Fig. 1, the remote sensing quantitative inversion method flow chart of a kind of crop physical and chemical parameter that the embodiment of the present invention provides, including step:

Step S11, general simulation Vegetation canopy reflectance look-up table LUT_sIt is divided into N according to spectral band_bIndividual Vegetation canopy reflectance look-up table LUT_si；

Wherein, N_bRepresent described LUT_sThe spectral band number of middle Vegetation canopy reflectivity data, described LUT_sFor according to PROSAIL model forward mode simulate obtain with crop physical and chemical parameter look-up table LUT_pCorresponding simulation Vegetation canopy reflectance look-up table.

PROSAIL model is vegetation radiative transfer model comparatively classical in the world, is mainly used in simulation road radiation transmission process between soil, blade, canopy.PROSAIL model includes forward mode and reverse mode, the forward mode of this model: by inputting spectral reflectance data, vegetation blade and canopy physical and chemical parameter data, observation geometrical condition data etc., it is possible to quantitative simulation Vegetation canopy reflectivity data.The reverse mode of this model: by inputting Vegetation canopy reflectivity data, it is possible to the relatively accurately crop physical and chemical parameter such as quantitative inversion leaf area index, chlorophyll content.

It should be noted that LUT_pIn the crop physical and chemical parameter entry number that comprises and LUT_sIn comprise simulation Vegetation canopy reflectance entry number identical, be positive integer N.

Step S12, according to each described LUT_siIn the size of spectral band reflectivity values corresponding to the Vegetation canopy reflectance that comprises, require each described LUT according to predetermined order_siIt is ranked up, the Vegetation canopy reflectance look-up table LUT after being sorted_{si_sort}；

Wherein, predetermined order requires can be from big to small or from small to large, concrete foundation is actually needed and determines.

Step S13, for LUT each described_{si_sort}, utilize binary search method, find out NL_iIndividual and true Vegetation canopy reflectivity data R_mI the similarity of () meets the simulation Vegetation canopy reflectivity data of preset requirement；

Binary search refers to: to seek the root (solution of x) of known function f (x)=0, then: 1) first find out an interval [a, b] so that f (a) and f (b) contrary sign.According to intermediate value theorem, this interval necessarily comprises equation root；2) midpoint in this interval is askedAnd find out the value of f (m)；3) if f (m) and f (a) sign is identical, then take [m, b] for new interval, otherwise take [a, m]；4) 2 are repeated) and 3) to perfect precision.

Wherein, in this step, described NL_iAccording to the simulation Vegetation canopy reflectivity data sum TNL and weight W utilizing binary search method to find out_iCalculating obtains, 0≤W_i≤ 1,1≤i≤N_b。

Concrete, described NL_iCalculating process include formula (1) and formula (2):

NL_i=round (W_i×TNL)(1)；

In formula, described W_iRepresent weight, 0≤W_i≤ 1,1≤i≤N_b, described TNL represents the simulation Vegetation canopy reflectivity data sum utilizing binary search method to find out and described TNL < N, and described N represents described LUT_sIn entry number, round represents bracket function；

$W_i=\frac{{\left(\frac{R_m\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{{\mathrm{\&Sigma;}}_{k=1}^{N_b}{\left(\frac{R_m\left(k\right)}{{\mathrm{\&sigma;}}_s\left(k\right)}\right)}^2}\&times;100\%---\left(2\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_mK () represents R_mAt the reflectance of kth spectral band, σ_sI () represents R_sThe standard deviation of (i), σ_sK () represents R_sThe standard deviation of (k), 1≤i≤N_b, 1≤k≤N_b。

Step S14, by each described simulation Vegetation canopy reflectivity data and described simulation Vegetation canopy reflectivity data at described LUT_sThe content of middle corresponding entry builds form with the form of corresponding relation, obtains dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}；

Step S15, calculate described LUT_{s_new}In each simulation Vegetation canopy reflectivity data and the reflectance comprehensive differences index TD with this true Vegetation canopy reflectivity data of simulation Vegetation canopy reflectivity data；

Step S16, from calculated all described TD, find out minima, and using the crop physical and chemical parameter data corresponding for the described minima optimal solution as lookup table algorithm.

In summary it can be seen, the present invention will simulate Vegetation canopy reflectance look-up table LUT_sSpectral band number according to its Vegetation canopy reflectivity data comprised, is divided into and spectral band number equal number of Vegetation canopy reflectance look-up table LUT_si, by each LUT_siRequire that sequence obtains LUT according to predetermined order_{si_sort}, for each LUT_{si_sort}Utilize binary search method, find out and meet the simulation Vegetation canopy reflectivity data of preset requirement with true Vegetation canopy reflectivity data similarity, by each simulation Vegetation canopy reflectivity data and these data at LUT_sThe content of middle corresponding entry builds form with the form of corresponding relation, obtains dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}, calculate LUT_{s_new}In each simulation Vegetation canopy reflectivity data and the reflectance comprehensive differences index TD of the true Vegetation canopy reflectivity data corresponding with this reflectivity data, and using the crop physical and chemical parameter data corresponding for minima in the TD optimal solution as lookup table algorithm.Can be seen that, the present invention makes full use of remote sensing, mathematics, the multidisciplinary advantage such as computer, consider the size and Orientation characteristic of Vegetation canopy reflectivity data, build high-precision true Vegetation canopy reflectivity data and simulation Vegetation canopy reflectivity data similarity quantitative assessing index, from data search angle, select suitable mathematical model optimizing data search strategy, introduce feasible computerized algorithm and complete data search process, computation complexity is reduced while improving inverting numerical precision, thus shortening search time, improve inverting efficiency, meet the remote sensing quantitative inversion high accuracy of crop physical and chemical parameter and quick demand.

For the remote sensing quantitative inversion method of crop physical and chemical parameter, present invention also offers a specific embodiment, specific as follows:

For target in hyperspectral remotely sensed image EnMAP, carry out the quantitative inversion numerical experiment of crop physical and chemical parameter and leaf area index and chlorophyll content.

Wherein, 200 row × 200 that are sized to of EnMAP remote sensing image arrange × 244 spectral bands；Simulation Vegetation canopy reflectance look-up table LUT based on PROSAIL model construction_sIn comprise simulation Vegetation canopy reflectance entry number be 100000, the value of the α in reflectance comprehensive differences index TD is 0.5, dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}In comprise simulation Vegetation canopy reflectance entry number TNL be 1000；It is Windows764 position system, IntelCoreI5CPU, 8G internal memory for carrying out the computer running environment of numerical experiment.

Experimental result is:

Prior art, namely based on root-mean-square error for similarity quantitative assessing index and be data search method with full traversal search the numerical experiment results of classical lookup table algorithm be: crop physical and chemical parameter quantitative inversion required time is 63350.2 seconds, the root-mean-square error of leaf area index and Quantitative inversion of chlorophyll content respectively 1.2 and 18.8g/cm²。

The numerical experiment results of technical scheme is: crop physical and chemical parameter quantitative inversion required time is 466.7 seconds, and the root-mean-square error of leaf area index and Quantitative inversion of chlorophyll content is 1.1 and 16.5 μ g/cm respectively²。

From experimental result it can be seen that the remote sensing quantitative inversion method of crop physical and chemical parameter provided by the invention improves numerical precision and the inverting efficiency of crop physical and chemical parameter quantitative inversion.

Based on the difference in size quantitative assessing index NRMSE and direction difference quantitative assessing index DCC of Vegetation canopy reflectivity data, the present invention constructs new types of data similarity quantitative assessing index and reflectance comprehensive differences index TD.The calculating process of reflectance comprehensive differences index TD is referring to formula (3):

TD=α × NRMSE+ (1-α) × DCC (3)；

In formula, α represents weight, and its vegetation pattern concrete with study area is relevant with kind, landform and landforms, observation geometrical condition etc., and 0≤α≤1, NRMSE represents normalization root-mean-square error；DCC represents direction variation coefficient.

On the basis of reflectance comprehensive differences index TD, for asking for the optimal solution of crop physical and chemical parameter, Develop Data is needed to search for.In lookup table algorithm, the most frequently used data search method is full traversal search, the computation complexity of the method is O (N × log (N)), N represents the entry number in look-up table, higher computation complexity and bigger memory space take so that the method need to pay higher time cost when computer realizes.For meeting high accuracy and the high efficiency real world applications demand of crop physical and chemical parameter remote sensing quantitative inversion, the present invention constructs the new types of data searching method based on binary search.In Computer Subject, binary search is the rapidly and efficiently data search method for subordinate ordered array, the computation complexity of the method is O (log (N)), compared to full traversal search method, has more significant reduction in computation complexity and time complexity.

In prior art, root-mean-square error RMSE is the evaluation index of quantitative description Vegetation canopy reflectivity data difference in size, which characterizes the Vegetation canopy reflectivity data antipode at each spectral band, but owing to not considering the data span of each spectral band reflectance in calculating process, therefore the result of calculation that may result in RMSE biases toward the spectral band that data span is bigger.The contribution to Vegetation canopy reflectivity data difference in size quantitative assessment of each spectral band is considered for equilibrium, the present invention have chosen the normalization root-mean-square error NRMSE quantitative assessing index as Vegetation canopy reflectivity data difference in size, the computing formula of normalization root-mean-square error NRMSE such as formula (4):

$NRMSE=\sqrt{\frac{{\mathrm{\&Sigma;}}_{i=1}^{N_b}{\left(\frac{R_m\left(i\right)-R_s\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{N_b}}---\left(4\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_sI () represents R_sAt the reflectance of i-th spectral band, σ_sI () represents R_sThe standard deviation of (i), 1≤i≤N_b。

In addition, Vegetation canopy reflectivity data has size and Orientation characteristic simultaneously, so when quantitative assessment is truly with simulation Vegetation canopy reflectivity data difference, size of data difference and direction difference should be considered, therefore propose can the computing formula such as formula (5) of the evaluation index of quantitative description Vegetation canopy reflectivity data direction difference and direction variation coefficient DCC, direction variation coefficient DCC for the present invention:

$DCC=\frac{sum\left(abs\right(\mathrm{sgn}\left({\mathrm{RC}}_m\right)-\mathrm{sgn}\left({\mathrm{RC}}_s\right)\left)\right)}{\frac{N_b}{2}}---\left(5\right);$

In formula, RC_m=(R_m(2)-R_m(1), R_m(3)-R_m(2) ..., R_m(N_b)-R_m(N_b-1)), RC_s=(R_s(2)-R_s(1), R_s(3)-R_s(2) ..., R_s(N_b)-R_s(N_b-1)), 1≤i≤N_b, sum represents that summing function, abs represent the function that takes absolute value, and sgn represents sign function.

Corresponding with said method embodiment, present invention also offers the remote sensing quantitative inversion system of a kind of crop physical and chemical parameter.

Referring to Fig. 2, the structural representation of the remote sensing quantitative inversion system of a kind of crop physical and chemical parameter that the embodiment of the present invention provides, including:

Cutting unit 21, for simulating Vegetation canopy reflectance look-up table LUT_sIt is divided into N according to spectral band_bIndividual Vegetation canopy reflectance look-up table LUT_si, wherein, N_bRepresent described LUT_sThe spectral band number of middle Vegetation canopy reflectivity data, described LUT_sFor according to PROSAIL model forward mode simulate obtain with crop physical and chemical parameter look-up table LUT_pCorresponding simulation Vegetation canopy reflectance look-up table；

Sequencing unit 22, for according to each described LUT_siIn the size of spectral band reflectivity values corresponding to the Vegetation canopy reflectance that comprises, require each described LUT according to predetermined order_siIt is ranked up, the Vegetation canopy reflectance look-up table LUT after being sorted_{si_sort}；

First searches unit 23, for for LUT each described_{si_sort}, utilize binary search method, find out NL_iIndividual and true Vegetation canopy reflectivity data R_mI the similarity of () meets the simulation Vegetation canopy reflectivity data of preset requirement；

Wherein, described NL_iAccording to the simulation Vegetation canopy reflectivity data sum TNL and weight W utilizing binary search method to find out_iCalculating obtains, 0≤W_i≤ 1,1≤i≤N_b。

Concrete, described NL_iCalculating process include formula (1) and formula (2):

NL_i=round (W_i×TNL)(1)；

In formula, described W_iRepresent weight, 0≤W_i≤ 1,1≤i≤N_b, described TNL represents the simulation Vegetation canopy reflectivity data sum utilizing binary search method to find out and described TNL < N, and described N represents described LUT_sIn entry number, round represents bracket function；

$W_i=\frac{{\left(\frac{R_m\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{{\mathrm{\&Sigma;}}_{k=1}^{N_b}{\left(\frac{R_m\left(k\right)}{{\mathrm{\&sigma;}}_s\left(k\right)}\right)}^2}\&times;100\%---\left(2\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_mK () represents R_mAt the reflectance of kth spectral band, σ_sI () represents R_sThe standard deviation of (i), σ_sK () represents R_sThe standard deviation of (k), 1≤i≤N_b, 1≤k≤N_b。

Form construction unit 24, for by each described simulation Vegetation canopy reflectivity data and described simulation Vegetation canopy reflectivity data at described LUT_sThe content of middle corresponding entry builds form with the form of corresponding relation, obtains dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}；

Computing unit 25, is used for calculating described LUT_{s_new}In each simulation Vegetation canopy reflectivity data and the reflectance comprehensive differences index TD of the true Vegetation canopy reflectivity data corresponding with this simulation Vegetation canopy reflectivity data；

Second searches unit 26, for finding out minima from calculated all described TD, and using the crop physical and chemical parameter data corresponding for the described minima optimal solution as lookup table algorithm.

In summary it can be seen, the present invention will simulate Vegetation canopy reflectance look-up table LUT_sSpectral band number according to its Vegetation canopy reflectivity data comprised, is divided into and spectral band number equal number of Vegetation canopy reflectance look-up table LUT_si, by each LUT_siRequire that sequence obtains LUT according to predetermined order_{si_sort}, for each LUT_{si_sort}Utilize binary search method, find out and meet the simulation Vegetation canopy reflectivity data of preset requirement with true Vegetation canopy reflectivity data similarity, by each simulation Vegetation canopy reflectivity data and these data at LUT_sThe content of middle corresponding entry builds form with the form of corresponding relation, obtains dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}, calculate LUT_{s_new}In each simulation Vegetation canopy reflectivity data and the reflectance comprehensive differences index TD of the true Vegetation canopy reflectivity data corresponding with this reflectivity data, and using the crop physical and chemical parameter data corresponding for minima in the TD optimal solution as lookup table algorithm.Can be seen that, the present invention makes full use of remote sensing, mathematics, the multidisciplinary advantage such as computer, consider the size and Orientation characteristic of Vegetation canopy reflectivity data, build high-precision true Vegetation canopy reflectivity data and simulation Vegetation canopy reflectivity data similarity quantitative assessing index, from data search angle, select suitable mathematical model optimizing data search strategy, introduce feasible computerized algorithm and complete data search process, computation complexity is reduced while improving inverting numerical precision, thus shortening search time, improve inverting efficiency, meet the remote sensing quantitative inversion high accuracy of crop physical and chemical parameter and quick demand.

Based on the difference in size quantitative assessing index NRMSE and direction difference quantitative assessing index DCC of Vegetation canopy reflectivity data, the present invention constructs new types of data similarity quantitative assessing index and reflectance comprehensive differences index TD.The calculating process of reflectance comprehensive differences index TD is referring to formula (3):

TD=α × NRMSE+ (1-α) × DCC (3)；

In formula, α represents weight, and its vegetation pattern concrete with study area is relevant with kind, landform and landforms, observation geometrical condition etc., and 0≤α≤1, NRMSE represents normalization root-mean-square error；DCC represents direction variation coefficient.

On the basis of reflectance comprehensive differences index TD, for asking for the optimal solution of crop physical and chemical parameter, Develop Data is needed to search for.In lookup table algorithm, the most frequently used data search method is full traversal search, the computation complexity of the method is O (N × log (N)), N represents the entry number in look-up table, higher computation complexity and bigger memory space take so that the method need to pay higher time cost when computer realizes.For meeting high accuracy and the high efficiency real world applications demand of crop physical and chemical parameter remote sensing quantitative inversion, the present invention constructs the new types of data searching method based on binary search.In Computer Subject, binary search is the rapidly and efficiently data search method for subordinate ordered array, the computation complexity of the method is O (log (N)), compared to full traversal search method, has more significant reduction in computation complexity and time complexity.

In prior art, root-mean-square error RMSE is the evaluation index of quantitative description Vegetation canopy reflectivity data difference in size, which characterizes the Vegetation canopy reflectivity data antipode at each spectral band, but owing to not considering the data span of each spectral band reflectance in calculating process, therefore the result of calculation that may result in RMSE biases toward the spectral band that data span is bigger.The contribution to Vegetation canopy reflectivity data difference in size quantitative assessment of each spectral band is considered for equilibrium, the present invention have chosen the normalization root-mean-square error NRMSE quantitative assessing index as Vegetation canopy reflectivity data difference in size, the computing formula of normalization root-mean-square error NRMSE such as formula (4):

$NRMSE=\sqrt{\frac{{\mathrm{\&Sigma;}}_{i=1}^{N_b}{\left(\frac{R_m\left(i\right)-R_s\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{N_b}}---\left(4\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_sI () represents R_sAt the reflectance of i-th spectral band, σ_sI () represents R_sThe standard deviation of (i), 1≤i≤N_b。

In addition, Vegetation canopy reflectivity data has size and Orientation characteristic simultaneously, so when quantitative assessment is truly with simulation Vegetation canopy reflectivity data difference, size of data difference and direction difference should be considered, therefore propose can the computing formula such as formula (5) of the evaluation index of quantitative description Vegetation canopy reflectivity data direction difference and direction variation coefficient DCC, direction variation coefficient DCC for the present invention:

$DCC=\frac{sum\left(abs\right(\mathrm{sgn}\left({\mathrm{RC}}_m\right)-\mathrm{sgn}\left({\mathrm{RC}}_s\right)\left)\right)}{\frac{N_b}{2}}---\left(5\right);$

In formula, RC_m=(R_m(2)-R_m(1), R_m(3)-R_m(2) ..., R_m(N_b)-R_m(N_b-1)), RC_s=(R_s(2)-R_s(1), R_s(3)-R_s(2) ..., R_s(N_b)-R_s(N_b-1)), 1≤i≤N_b, sum represents that summing function, abs represent the function that takes absolute value, and sgn represents sign function.

It should be noted that the operation principle of each ingredient refers to embodiment of the method corresponding part in system embodiment, repeat no more herein.

Finally, it can further be stated that, in this article, the relational terms of such as first and second or the like is used merely to separate an entity or operation with another entity or operating space, and not necessarily requires or imply the relation that there is any this reality between these entities or operation or sequentially.And, term " includes ", " comprising " or its any other variant are intended to comprising of nonexcludability, so that include the process of a series of key element, method, article or equipment not only include those key elements, but also include other key elements being not expressly set out, or also include the key element intrinsic for this process, method, article or equipment.When there is no more restriction, statement " including ... " key element limited, it is not excluded that there is also other identical element in including the process of described key element, method, article or equipment.

In this specification, each embodiment adopts the mode gone forward one by one to describe, and what each embodiment stressed is the difference with other embodiments, between each embodiment identical similar portion mutually referring to.

Described above to the disclosed embodiments, makes professional and technical personnel in the field be capable of or uses the present invention.The multiple amendment of these embodiments be will be apparent from for those skilled in the art, and generic principles defined herein can without departing from the spirit or scope of the present invention, realize in other embodiments.Therefore, the present invention is not intended to be limited to the embodiments shown herein, and is to fit to the widest scope consistent with principles disclosed herein and features of novelty.

Claims (6)

1. the remote sensing quantitative inversion method of a crop physical and chemical parameter, it is characterised in that including:

Will simulation Vegetation canopy reflectance look-up table LUT_sIt is divided into N according to spectral band_bIndividual Vegetation canopy reflectance look-up table LUT_si, wherein, N_bRepresent described LUT_sThe spectral band number of middle Vegetation canopy reflectivity data, described LUT_sFor according to PROSAIL model forward mode simulate obtain with crop physical and chemical parameter look-up table LUT_pCorresponding simulation Vegetation canopy reflectance look-up table；

According to each described LUT_siIn the size of spectral band reflectivity values corresponding to the Vegetation canopy reflectance that comprises, require each described LUT according to predetermined order_siIt is ranked up, the Vegetation canopy reflectance look-up table LUT after being sorted_{si_sort}；

For LUT each described_{si_sort}, utilize binary search method, find out NL_iIndividual and true Vegetation canopy reflectivity data R_mI the similarity of () meets the simulation Vegetation canopy reflectivity data of preset requirement, wherein, and described NL_iAccording to the simulation Vegetation canopy reflectivity data sum TNL and weight W utilizing binary search method to find out_iCalculating obtains, 0≤W_i≤ 1,1≤i≤N_b；

By each described simulation Vegetation canopy reflectivity data and described simulation Vegetation canopy reflectivity data at described LUT_sThe content of middle corresponding entry builds form with the form of corresponding relation, obtains dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}；

Calculate described LUT_{s_new}In each simulation Vegetation canopy reflectivity data and the reflectance comprehensive differences index TD of the true Vegetation canopy reflectivity data corresponding with this simulation Vegetation canopy reflectivity data；

Minima is found out from calculated all described TD, and using the crop physical and chemical parameter data corresponding for the described minima optimal solution as lookup table algorithm.

2. remote sensing quantitative inversion method according to claim 1, it is characterised in that described NL_iCalculating process include formula (1) and formula (2):

NL_i=round (W_i×TNL)(1)；

In formula, described W_iRepresent weight, 0≤W_i≤ 1,1≤i≤N_b, described TNL represents the simulation Vegetation canopy reflectivity data sum utilizing binary search method to find out and described TNL < N, and described N represents described LUT_sIn entry number, round represents bracket function；

$W_i=\frac{{\left(\frac{R_m\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{{\mathrm{\&Sigma;}}_{k=1}^{N_b}{\left(\frac{R_m\left(k\right)}{{\mathrm{\&sigma;}}_s\left(k\right)}\right)}^2}\&times;100\%---\left(2\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_mK () represents R_mAt the reflectance of kth spectral band, σ_sI () represents R_sThe standard deviation of (i), σ_sK () represents R_sThe standard deviation of (k), 1≤i≤N_b, 1≤k≤N_b。

3. remote sensing quantitative inversion method according to claim 1, it is characterised in that the calculating process of described TD includes formula (3), formula (4) and formula (5):

TD=α × NRMSE+ (1-α) × DCC (3)；

In formula, α represents weight, 0≤α≤1, and NRMSE represents normalization root-mean-square error；DCC represents direction variation coefficient；

$NRMSE=\sqrt{\frac{{\mathrm{\&Sigma;}}_{i=1}^{N_b}{\left(\frac{R_m\left(i\right)-R_s\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{N_b}}---\left(4\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_sI () represents R_sAt the reflectance of i-th spectral band, σ_sI () represents R_sThe standard deviation of (i), 1≤i≤N_b；

$DCC=\frac{sum\left(abs\right(\mathrm{sgn}\left({\mathrm{RC}}_m\right)-\mathrm{sgn}\left({\mathrm{RC}}_s\right)\left)\right)}{\frac{N_b}{2}}---\left(5\right);$

In formula, RC_m=(R_m(2)-R_m(1), R_m(3)-R_m(2) ..., R_m(N_b)-R_m(N_b-1)), RC_s=(R_s(2)-R_s(1), R_s(3)-R_s(2) ..., R_s(N_b)-R_s(N_b-1)), 1≤i≤N_b, sum represents that summing function, abs represent the function that takes absolute value, and sgn represents sign function.

4. the remote sensing quantitative inversion system of a crop physical and chemical parameter, it is characterised in that including:

Cutting unit, for simulating Vegetation canopy reflectance look-up table LUT_sIt is divided into N according to spectral band_bIndividual Vegetation canopy reflectance look-up table LUT_si, wherein, N_bRepresent described LUT_sThe spectral band number of middle Vegetation canopy reflectivity data, described LUT_sFor according to PROSAIL model forward mode simulate obtain with crop physical and chemical parameter look-up table LUT_pCorresponding simulation Vegetation canopy reflectance look-up table；

Sequencing unit, for according to each described LUT_siIn the size of spectral band reflectivity values corresponding to the Vegetation canopy reflectance that comprises, require each described LUT according to predetermined order_siIt is ranked up, the Vegetation canopy reflectance look-up table LUT after being sorted_{si_sort}；

First searches unit, for for LUT each described_{si_sort}, utilize binary search method, find out NL_iIndividual and true Vegetation canopy reflectivity data R_mI the similarity of () meets the simulation Vegetation canopy reflectivity data of preset requirement, wherein, and described NL_iAccording to the simulation Vegetation canopy reflectivity data sum TNL and weight W utilizing binary search method to find out_iCalculating obtains, 0≤W_i≤ 1,1≤i≤N_b；

Form construction unit, for by each described simulation Vegetation canopy reflectivity data and described simulation Vegetation canopy reflectivity data at described LUT_sThe content of middle corresponding entry builds form with the form of corresponding relation, obtains dimensionality reduction simulation Vegetation canopy reflectance look-up table LUT_{s_new}；

Computing unit, is used for calculating described LUT_{s_new}In each simulation Vegetation canopy reflectivity data and the reflectance comprehensive differences index TD of the true Vegetation canopy reflectivity data corresponding with this simulation Vegetation canopy reflectivity data；

Second searches unit, for finding out minima from calculated all described TD, and using the crop physical and chemical parameter data corresponding for the described minima optimal solution as lookup table algorithm.

5. remote sensing quantitative inversion system according to claim 4, it is characterised in that described NL_iCalculating process include formula (1) and formula (2):

NL_i=round (W_i×TNL)(1)；

In formula, described W_iRepresent weight, 0≤W_i≤ 1,1≤i≤N_b, described TNL represents the simulation Vegetation canopy reflectivity data sum utilizing binary search method to find out and described TNL < N, and described N represents described LUT_sIn entry number, round represents bracket function；

$W_i=\frac{{\left(\frac{R_m\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{{\mathrm{\&Sigma;}}_{k=1}^{N_b}{\left(\frac{R_m\left(k\right)}{{\mathrm{\&sigma;}}_s\left(k\right)}\right)}^2}\&times;100\%---\left(2\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_mK () represents R_mAt the reflectance of kth spectral band, σ_sI () represents R_sThe standard deviation of (i), σ_sK () represents R_sThe standard deviation of (k), 1≤i≤N_b, 1≤k≤N_b。

6. remote sensing quantitative inversion system according to claim 4, it is characterised in that the calculating process of described TD includes formula (3), formula (4) and formula (5):

TD=α × NRMSE+ (1-α) × DCC (3)；

In formula, α represents weight, 0≤α≤1, and NRMSE represents normalization root-mean-square error；DCC represents direction variation coefficient；

$NRMSE=\sqrt{\frac{{\mathrm{\&Sigma;}}_{i=1}^{N_b}{\left(\frac{R_m\left(i\right)-R_s\left(i\right)}{{\mathrm{\&sigma;}}_s\left(i\right)}\right)}^2}{N_b}}---\left(4\right);$

In formula, R_mRepresent true Vegetation canopy reflectivity data, σ_sRepresent R_sStandard deviation, R_sRepresent simulation Vegetation canopy reflectivity data, R_mI () represents R_mAt the reflectance of i-th spectral band, R_sI () represents R_sAt the reflectance of i-th spectral band, σ_sI () represents R_sThe standard deviation of (i), 1≤i≤N_b；

$DCC=\frac{sum\left(abs\right(\mathrm{sgn}\left({\mathrm{RC}}_m\right)-\mathrm{sgn}\left({\mathrm{RC}}_s\right)\left)\right)}{\frac{N_b}{2}}---\left(5\right);$

In formula, RC_m=(R_m(2)-R_m(1), R_m(3)-R_m(2) ..., R_m(N_b)-R_m(N_b-1)), RC_s=(R_s(2)-R_s(1), R_s(3)-R_s(2) ..., R_s(N_b)-R_s(N_b-1)), 1≤i≤N_b, sum represents that summing function, abs represent the function that takes absolute value, and sgn represents sign function.