CN106526592A - Frequency spectrum-based estimation method of SAR image low-scattering region scattering value - Google Patents

Abstract

The invention provides a frequency spectrum-based estimation method of an SAR image low-scattering region scattering value. The estimation method includes the following steps: generating and storing a single-look complex image according to original data received by an SAR; dividing the single-look complex image into n sub-image blocks, and performing Doppler center frequency estimation on each sub-image block and moving the Doppler center frequency to a zero frequency position; calculating a scattering rate initial value and a Doppler frequency spectrum with the Doppler center frequency moving to zero frequency of each sub-image block; utilizing maximum likelihood estimation and a Newton iteration method to estimate a relative scattering value of each sub-image block according to the scattering rate initial value and Doppler frequency spectrum of each sub-image block; and calculating an absolute scattering value of each sub-image block. The frequency spectrum-based estimation method of the SAR image low-scattering region scattering value derives a relation between target scattering values and frequency spectrums from a frequency domain, and derives a correlation equation, the scattering vale in the equation is estimated through the maximum likelihood estimation method and Newton iteration method, and the initial value of the Newton iteration method considers the factors of system noise and azimuth ambiguity, and thus an estimation result is more accurate.

Description

The method of estimation of the low scattering region scattering value of SAR image based on frequency spectrum

Technical field

The invention belongs to synthetic aperture radar (Synthetic Aperture Radar, SAR) image processing field, more It is related to a kind of method of estimation of SAR image based on frequency spectrum low scattering region scattering value body.

Background technology

In synthetic aperture radar (SAR) image, the normalization radar cross section scattering value of low scattering region is in the picture It is now dark areas.The frequency that dark areas occur in SAR image in ocean is higher, such as offshore spilled oil, organic film, low wind speed Area, sharp side, upper up-flow, the filaments of sun band of interior ripple and surge；Also there are many dark areas situations, the moon on such as mountain in the SAR image of land Face, smooth airfield runway etc. are all classical low scattering regions.

On the one hand, in SAR image, the signal intensity of low scattering region is close to the bottom of making an uproar of even below SAR system, so as to shadow The image intensity of the low scattering region of sound.By taking ocean surface as an example, under low wind speed and low incidence angle, for L, C and X-band sea The average scatter value scope on foreign surface is from -15dB to -25dB, and the scattering value of the low scattering region of ocean surface is than ocean surface Average scatter value it is lower, the scattering value of the low scattering region of ocean surface will be generally below -30dB, but most of satellite-borne SAR systems The scope of the equivalent noise figure (NESZ) of system is that -20dB arrives -30dB.In addition, in sea SAR image low scattering region it is backward Scattered signal intensity will be generally below the bottom of making an uproar of SAR system.

On the other hand, the azimuth ambiguity in strong scattering region can affect the image intensity of low scattering region, and this impact is main It is because that the Doppler frequency of the signal that antenna bearingt is reflected to the region that secondary lobe irradiates has exceeded radar pulse and repeated frequency Rate (PRF).The azimuth ambiguity signal location of target and actual position have certain deviation, and this skew is depending on PRF, platform speed With the doppler centroid of SAR system.The azimuth ambiguity value of classical satellite-borne SAR is about -15dB and arrives -20dB.If strong dissipate The signal intensity that region is penetrated in the low scattering region of blurred signal strength ratio of low scattering region is higher by 15dB to 20dB, azimuth ambiguity Signal will have a strong impact on the image intensity of low scattering region.Azimuth ambiguity signal occurs more in the intersection of land and the water surface Frequently, because the scattering value of land target is more much better than than the water surface.

Analysis display, will estimate the exact value of scattering value of low scattering region it is necessary to consider system noise and orientation above Fuzzy impact.And the Standard Ratio scaling algorithm of SAR image only considered system noise, the shadow of azimuth ambiguity is but ignored Ring.The existing method for improving interpretation picture quality is simply to remove system noise or optimize picture quality from image area. When calibration is performed, the scattering value of low scattering region inevitably can be affected by system noise and azimuth ambiguity, and this will The process to later image and interpretation is caused to bring puzzlement.

The content of the invention

(1) technical problem to be solved

Based on problem above, it is an object of the invention to propose a kind of low scattering region scattering of SAR image based on frequency spectrum The method of estimation of value, for solving at least one of above technical problem.

(2) technical scheme

To achieve these goals, the present invention proposes a kind of low scattering region scattering value of SAR image based on frequency spectrum Method of estimation, which includes that step is as follows：

Step 1, the Raw Data Generation received according to SAR simultaneously preserve haplopia complex pattern；

Step 2, haplopia complex pattern is divided into into n subimage block, doppler centroid is carried out to each subimage block and is estimated Meter, and doppler centroid is moved to into zero-frequency position；Wherein, n is the positive integer more than or equal to 1；

Step 3, calculate each subimage block scattered power initial value and doppler centroid move to zero-frequency Doppler frequency Spectrum；

Step 4, according to the scattered power initial value and Doppler frequency spectrum of each subimage block, with maximal possibility estimation and newton Iterative method estimates the relative scattering value of each subimage block；

Step 5, the absolute scatter value for calculating each subimage block.

Further, the scattered power initial value of above-mentioned each subimage block is expressed as：

Wherein,For the relative value of image intensity, x₀、y₀Orientation and distance respectively to coordinate, N₀It is SAR system Intrinsic noise,It is location of pixels [x₀+D_x- L/2, y₀+D_y] and [x₀+D_x+ L/2, y₀+D_y] between Average relative scattering value,It is location of pixels [x₀-D_x- L/2, y₀-D_y] and [x₀-D_x+ L/2, y₀-D_y] Between average relative scattering value, L be calculate Doppler frequency spectrum data length；D_xAnd D_yIt is azimuth ambiguity signal location respectively And locations of real targets between orientation and distance to side-play amount, σ_minFor the minima of absolute scatter value, A is orientation mould The paste factor.

Further, the iterative formula of above-mentioned Newton iteration method is：

Wherein, M is the points of the doppler spectral of each subimage block, F_rIt is the radar pulse repetition frequency of SAR system, P_c (f_i)、P_L(f_i)、P_R(f_i) it is respectively the energy spectrum of i-th of Doppler frequency spectrum main lobe, left secondary lobe and right secondary lobe points, p_n-X (f_i)、p_n(f_i)、p_n+X(f_i) be that the n-th-X is individual respectively, count for i-th of n-th and the n-th+X subimage block Doppler frequency spectrum Energy spectrum, α represents antenna elevation angle,Be the n-th -2X, the n-th-X respectively, N-thth, the relative scattering value of the n-th+X and the n-th+2X subimage block, i is the natural number more than or equal to 1.

Further, the formula of above-mentioned maximal possibility estimation is expressed as：

Wherein, the g (p in above formula_n(f₁) p_n(f₂) … p_n(f_M))、g(p_n-X(f₁) p_n-X(f₂) … p_n-X(f_M))、g (p_n+X(f₁) p_n+X(f₂) … p_n+X(f_M)) be n-th respectively, the probability density letter of the n-th-X and the n-th+X subimage block Number, p_n(f_i) be n-th subimage block i-th Frequency point energy spectrum.

Further, the absolute scatter value of each image block is calculated according to below equation：

Wherein, R, α and G represent the system gain of target oblique distance, antenna elevation angle and specific pixel respectively, and g (α) is pitching Two-way antenna gain when angle is α, K is scaling constant, R_ref、α_refAnd G_refOblique distance, the angle of pitch of reference position are represented respectively And system gain.

Further, the initial value of above-mentioned Newton iteration method is the scattered power initial value of each subimage block.

Further, when generating haplopia complex pattern in above-mentioned steps 1, the compression of orientation is adopted haves no right matched filter.

Further, the orientation energy spectrum of above-mentioned haplopia complex pattern has the identical aspect of model with SAR initial datas：

p_s(f, x₀, y₀)=p_r(f, x₀, y₀)|H(f)|²=p_r(f, x₀, y₀)；

Wherein,It is that orientation haves no right matched filter, p_s(f, x₀, y₀) and p_r(f, x₀, y₀) It is the energy spectrum of haplopia complex pattern and SAR initial datas respectively, f is doppler centroid, and λ is SAR system wavelength, and V is SAR The movement velocity of platform.

(3) beneficial effect

The method of estimation of the low scattering region scattering value of a kind of SAR image based on frequency spectrum proposed by the present invention, with following Beneficial effect：

1st, the present invention derives the relation of target scattering value and frequency spectrum from frequency domain, and pushes away to obtain correlate equation, through maximum Likelihood estimation and Newton iteration method estimate the scattering value in equation, and the initial value of Newton iteration method consider system noise and Azimuth ambiguity factor, therefore estimated result is more accurate；

2nd, the present invention from initial data carry out process generate haplopia complex pattern, and orientation compression using have no right matching Wave filter, can only change the phase place of Doppler frequency spectrum, the amplitude general without changing Doppler, therefore inherently extract data Useful information, it is to avoid the decay of the general intensity of Doppler；

3rd, for low scattering region, the present invention is more better than the method that tradition removes system noise, at the present invention The textural characteristics for becoming apparent from are had after reason.

Description of the drawings

Fig. 1 is the diagram of the SAR initial datas that one embodiment of the invention is proposed and other various energy ingredients；

Fig. 2 is the method for estimation of the low scattering region scattering value of the SAR image based on frequency spectrum that one embodiment of the invention is proposed Method flow schematic diagram；

Fig. 3 is the method for estimation of the low scattering region scattering value of the SAR image based on frequency spectrum that one embodiment of the invention is proposed Detailed process schematic diagram；

Fig. 4 is the ERS-2 satellite datas of the selection that one embodiment of the invention is proposed after without weights matched filtering imaging Haplopia complex pattern；

Fig. 5 is the comparison diagram before and after the doppler centroid movement that one embodiment of the invention is proposed；

Fig. 6 (a) is the result of the estimation scattering value method that one embodiment of the invention is proposed；

Fig. 6 (b) is the result of conventional estimated scattering value method；

Fig. 7 is estimated result Fig. 6 (a) and white line in Fig. 6 (b) of the estimation scattering value method that one embodiment of the invention is proposed The profile of position.

Specific embodiment

To make the object, technical solutions and advantages of the present invention become more apparent, below in conjunction with specific embodiment, and reference Accompanying drawing, the present invention is described in further detail.

A kind of Standard Ratio scaling algorithm of SAR image is as follows：

Wherein, I, R, α and G are the system increasing of image intensity, target oblique distance, antenna elevation angle and specific image pixel respectively Benefit；G (α) be the angle of pitch be α when two-way antenna gain, N₀It is system noise, K is scaling constant, R_ref、α_refAnd G_refPoint It is not oblique distance, the angle of pitch and the system gain of reference position.

However, accurate system noise N is hardly provided in business satellite-borne SAR data product₀.Although providing More accurate system noise N₀, the scattering value of low scattering region estimates may still to obtain and having no meaning less than or equal to zero The scattering value of justice, because the image intensity of low scattering region inherently is likely to less than system noise N₀Random change Amount.Moreover, in most of practical scattering values calibration applications, system noise be also it is ignored, therefore, equation (1) and quilt It is reduced to：

The composition diagram of the various energy that Fig. 1 is included in giving actual SAR data, it can be seen that actual SAR Data are included on the left of SAR initial datas, target and the strong target in right side is caused azimuth ambiguity, system noise.Therefore need to go Except system noise and azimuth ambiguity signal, SAR primary data information (pdi)s could be more accurately obtained.

Based on above reason, the invention discloses a kind of estimation of the low scattering region scattering value of SAR image based on frequency spectrum Method, which includes that step is as follows：

Step 1, the Raw Data Generation received according to SAR simultaneously preserve haplopia complex pattern；

Step 2, haplopia complex pattern is divided into into n subimage block, doppler centroid is carried out to each subimage block and is estimated Meter, and doppler centroid is moved to into zero-frequency position；

Step 3, calculate each subimage block scattered power initial value and doppler centroid move to zero-frequency Doppler frequency Spectrum；

Step 4, according to the scattered power initial value and Doppler frequency spectrum of each subimage block, with maximal possibility estimation and newton Iterative method estimates the relative scattering value of each subimage block；

Step 5, the absolute scatter value for calculating each subimage block.

When generating haplopia complex pattern wherein in step 1, being compressed into for orientation haves no right matched filter as employing, because In great majority application, the azimuth resolution of SAR primary signals is too low, in order to improve azimuth resolution, azimuth match wave filter Have to apply in SAR primary signals, and convert thereof into haplopia complex pattern.Selection haves no right matched filtering using orientation Device, then azimuth match wave filter can only change the phase place of Doppler frequency spectrum, without change Doppler frequency spectrum amplitude.Institute So that orientation energy spectrum and the SAR initial datas of haplopia complex pattern have the identical aspect of model.Haplopia complex pattern and SAR are original Relational expression between data can be expressed as：

p_s(f, x₀, y₀)=p_r(f, x₀, y₀)|H(f)|²=p_r(f, x₀, y₀)； (3)

Wherein,It is that orientation haves no right matched filter, p_s(f, x₀, y₀) and p_r(f, x₀, y₀) It is the energy spectrum of haplopia complex pattern and SAR primary signals respectively, f is Doppler frequency, and λ is SAR system wavelength, and V is SAR platform Movement velocity.

Estimation of Doppler central frequency is carried out to each subimage block, and doppler centroid is moved to into zero-frequency position, Its detailed process is Doppler's frequency that the time-domain image to each subimage block carries out that Fourier transformation obtains each subimage block Then the mid frequency of doppler spectral is moved to zero-frequency position, then the inverse Fourier transform for carrying out orientation, you can obtain many by spectrum Each subimage block of the haplopia complex pattern of Pu Le spectrum mid frequency movements.

The scattered power initial value and doppler centroid for calculating each subimage block moves to the Doppler frequency spectrum of zero-frequency：

First the Doppler frequency spectrum of SAR primary signals is analyzed herein, principle is enrolled according to radar return, its form can To be expressed as：

The Doppler frequency spectrum of orientation is the secondary lobe that fallen in the region by the main lobe and other positions of antenna radiation pattern and dissipate Penetrate what the weighting of rate was obtained；Wherein, (x₀, y₀) it is the center for carrying out Fourier transformation region, x₀And y₀It is orientation respectively With distance to coordinate.E [] is mathematic expectaion, p_rF () is the orientation energy spectrum of SAR initial datas.P_aF () is scattering value For the energy spectrum of the ideal point target of 0dB, its model is depending on bilateral antenna directional diagram.In addition, f₀It is Doppler center frequency Rate, F_rIt is the pulse recurrence frequency of SAR system,It is location of pixels [x₀-nD_x-L/2 y₀-nD_y] and [x₀-nD_x+L/2 y₀-nD_y] (L be calculate Doppler frequency spectrum data length) between average scatter value, D_xAnd D_yThe side of being respectively Between position blurred signal position and locations of real targets orientation and distance to side-play amount, they can be written as：

In formula (4), the signal that n=0 is reflected corresponding to antenna main lobe, n ≠ 0 is corresponding to azimuth ambiguity signal Affect.Under normal circumstances, azimuth ambiguity signal only considers the situation of n=-1 and 1, that is, the first secondary lobe of azimuth anteena.Cause This, equation (4) can be reduced to：

The true energy spectrum of one fritter of haplopia complex pattern is actually a stochastic process.Because the letter of haplopia complex pattern Number it is a multiple Gauss process, so the probability density function of each sample of energy spectrum obeys famous exponential

From equation (6), the backscatter signal of specific regionIt is made up of three frequency spectrums：p_s(f, x₀, y₀), p_s (f, x₀-D_x, y₀-D_y), p_s(f, x₀+D_x, y₀+D_y).Therefore the joint probability density function of all Frequency points is：

The scattered power initial value of each subimage block is the value derived according to the principle Analysis of SAR, can be expressed as：

Wherein,For the relative value of image intensity,It is location of pixels [x₀+D_x- L/2, y₀+D_y] [x₀+D_x+ L/2, y₀+D_y] between average relative scattering value,It is location of pixels [x₀-D_x- L/2, y₀-D_y] and [x₀-D_x+ L/2, y₀-D_y] between average relative scattering value, σ_minFor the minima of absolute scatter value, by figure The scattering value analysis of picture is obtained；A is the azimuth ambiguity factor, and which is represented by

Each subimage block doppler centroid moves to the Doppler frequency spectrum of zero-frequency i.e.：Doppler spectral mid frequency is moved The time-domain image of each subimage block of dynamic haplopia complex pattern, again passes by Fourier transformation and obtains Doppler frequency spectrum.

According to the scattered power initial value and Doppler frequency spectrum of each subimage block, the relative scattering of each subimage block is estimated Value：Estimate that scattering value is a classical Bayesian estimation problem according to Doppler frequency spectrum, the estimation side of whole DANFU visible image Journey can be expressed as

To clearly show that, herein g functions are divided into two lines expression, p (f are should be actually_{i-X_1}) ..., p (f_{i_M})....

For single subimage block, which estimates that equation is

Then maximal possibility estimation and Newton iteration method are used by solving equation below, it is possible to obtain equation (10) Estimated result, maximum Likelihood is represented by

Wherein, the g (p in above formula_n(f₁) p_n(f₂) … p_n(f_M))、g(p_n-X(f₁) p_n-X(f₂) … p_n-X(f_M))、g (p_n+X(f₁) p_n+X(f₂) … p_n+X(f_M)) be n-th respectively, the probability density letter of the n-th-X and the n-th+X subimage block Number, p_n(f_i) be n-th subimage block i-th Frequency point energy spectrum.

Newton iteration formula is

Wherein, M is the points of the doppler spectral of each subimage block, F_rIt is the radar pulse repetition frequency of SAR system, P_c (f_i)、P_L(f_i)、P_R(f_i) it is respectively the energy spectrum of i-th of Doppler frequency spectrum main lobe, left secondary lobe and right secondary lobe points, p_n-X (f_i)、p_n(f_i)、p_n+X(f_i) be that the n-th-X is individual respectively, count for i-th of n-th and the n-th+X subimage block Doppler frequency spectrum Energy spectrum, α represents antenna elevation angle,Be the n-th -2X, the n-th-X respectively, N-thth, the relative scattering value of the n-th+X and the n-th+2X subimage block, i is the natural number more than or equal to 1.

Formula (12) is that formula (8) is brought in formula (11), then with constraint solving method so that estimated result is not There is negative value, so there is last formula of formula (12).Iterative initial value be each subimage block as follows Scattered power initial value is shown in formula (9).Its iterative initial value is the scattered power initial value of each subimage block, until iteration convergence is obtained accurately The relative value of scattered power

Calculate the absolute scatter value of each image block：Obtained due to iterationIt is a relative back scattering Value, rather than absolute scatter value.So if the K constants of known radiation calibration, obtained with iterationReplace formula (1) In I-N₀, you can the relative back scattering value estimated is converted to into absolute scatter value, expression formula is as follows：

Below by way of specific embodiment estimating to the low scattering region scattering value of the SAR image based on frequency spectrum proposed by the present invention Meter method is described in detail.

Embodiment

As shown in Fig. 2 the present embodiment proposes a kind of estimation side of the low scattering region scattering value of SAR image based on frequency spectrum Method, which includes that step is as follows：

Step 1, the Raw Data Generation received according to SAR simultaneously preserve haplopia complex pattern；

Step 2, haplopia complex pattern is divided into into n subimage block, doppler centroid is carried out to each subimage block and is estimated Meter, and doppler centroid is moved to into zero-frequency position；

Step 3, calculate each subimage block scattered power initial value and doppler centroid move to zero-frequency Doppler frequency Spectrum；

Step 4, according to the scattered power initial value and Doppler frequency spectrum of each subimage block, with maximal possibility estimation and newton Iterative method estimates the relative scattering value of each subimage block；

Step 5, the absolute scatter value for calculating each subimage block.

The detail flowchart of the method is as shown in Figure 3：

First, the initial data for being received according to SAR, generates haplopia complex pattern simultaneously using matched filter is had no right in orientation This image is preserved, the present embodiment is chosen the data of ERS-2 satellites and carries out the solution of absolute scatter value, by without weights matched filtering Haplopia complex pattern after imaging is as shown in Figure 4；

Then haplopia complex pattern is divided into into n subimage block, white box part is one of subimage block in Fig. 4, The time-domain image of n subimage block is changed into by Doppler frequency spectrum using Fourier transformation, by the Doppler frequency spectrum for obtaining After frequency of heart moves to zero-frequency position, using orientation inverse Fourier transform obtain mid frequency movement each subimage block when Area image, mid frequency do not move and move after Doppler frequency spectrum comparison diagram it is as shown in Figure 5；

According to the operation principle of SAR, the scattered power initial value of each subimage block is calculated by formula (9), while adopting The time-domain image of each subimage block that mid frequency is moved by Fourier transformation is converted to the Doppler of each subimage block Frequency spectrum；

According to the scattered power initial value and Doppler frequency spectrum of each subimage block, using formula (10)-formula (12), by most Maximum-likelihood is estimated to obtain the relative scattering value of each subimage block with Newton iteration method, in iterative process, once obtains per iteration The value that should obtain with last iteration of value compared with, till loop convergence.

The relative scattered power value of each subimage block obtained according to iteration, obtains each subimage block using formula (13) Absolute scatter value.

Fig. 6 (a) and (b) are that white rectangle frame portion point is deducted by the present embodiment processing method and tradition and made an uproar in Fig. 4 respectively The two width images that the method for sound is obtained；

Two curves in Fig. 7 are the sectional view of white line part in two width figures of Fig. 6 respectively.Three lines that white line is passed through are seas Ripple in foreign.Wherein dotted line is results of the Fig. 6 (b) after the method that tradition deducts noise is processed, and solid line is Fig. 6 (a) by this The result that the method that embodiment is proposed is obtained after processing.Contrast from Fig. 7 can be seen that whether traditional treatment method or The processing method of the present embodiment, internal ripple crest (relatively strong scattering region) result is identical, and to low scattering region Interior ripple trough area, traditional method can only obtain the energy intensity of about 10db, and the present embodiment can be by trough after processing The energy intensity of position recovers to -25 to -30db or so, and the ability that this explanation the present embodiment recovers dark areas textural characteristics more connects Ripple actual value near.

Particular embodiments described above, has been carried out to the purpose of the present invention, technical scheme and beneficial effect further in detail Describe in detail bright, it should be understood that the foregoing is only the specific embodiment of the present invention, be not limited to the present invention, it is all Within the spirit and principles in the present invention, any modification, equivalent substitution and improvements done etc. should be included in the protection of the present invention Within the scope of.

Claims (8)

1. the method for estimation of the low scattering region scattering value of a kind of SAR image based on frequency spectrum, which includes that step is as follows：

Step 1, the Raw Data Generation received according to SAR simultaneously preserve haplopia complex pattern；

Step 2, the haplopia complex pattern is divided into into n subimage block, Doppler center is carried out to the subimage block described in each Frequency Estimation, and the doppler centroid is moved to into zero-frequency position；Wherein, n is the positive integer more than or equal to 1；

Step 3, calculate each subimage block scattered power initial value and doppler centroid move to zero-frequency Doppler frequency Spectrum；

Step 4, according to the scattered power initial value and Doppler frequency spectrum of each subimage block, with maximal possibility estimation and newton Iterative method estimates the relative scattering value of each subimage block；

Step 5, the absolute scatter value for calculating each subimage block.

2. the method for estimation of the low scattering region scattering value of SAR image based on frequency spectrum as claimed in claim 1, its feature exist In the scattered power initial value of each subimage block is expressed as：

Wherein,For the relative value of image intensity, x₀、y₀Orientation and distance respectively to coordinate, N₀It is the intrinsic of SAR system Noise,It is location of pixels [x₀+D_x- L/2, y₀+D_y] and [x₀+D_x+ L/2, y₀+D_y] between it is average With respect to scattering value,It is location of pixels [x₀-D_x- L/2, y₀-D_y] and [x₀-D_x+ L/2, y₀-D_y] between Average relative scattering value, L be calculate Doppler frequency spectrum data length；D_xAnd D_yIt is azimuth ambiguity signal location and true respectively Between real target location orientation and distance to side-play amount, σ_minFor the minima of absolute scatter value, A be azimuth ambiguity because Son.

3. the method for estimation of the low scattering region scattering value of SAR image based on frequency spectrum as claimed in claim 1, its feature exist In the iterative formula of the Newton iteration method is：

$\begin{array}{c}-\underset{i=1}{\overset{M}{\Σ}}\frac{P_C\left(f_i\right)}{\[{\stackrel{\‾}{\mathrm{\σ}}}_{n-X}{P}_L\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_n{P}_C\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_{n+X}{P}_R\left(f_i\right)+\frac{N_0}{F_r}\]}-\underset{i=1}{\overset{M}{\Σ}}\frac{P_R\left(f_i\right)}{\[{\stackrel{\‾}{\mathrm{\σ}}}_{n-2X}{P}_L\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_{n-X}{P}_C\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_n{P}_R\left(f_i\right)+\frac{N_0}{F_r}\]}\\ -\underset{i=1}{\overset{M}{\Σ}}\frac{P_L\left(f_i\right)}{\[{\stackrel{\‾}{\mathrm{\σ}}}_n{P}_L\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_{n+X}{P}_C\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_{n+2X}{P}_R\left(f_i\right)+\frac{N_0}{F_r}\]}+\underset{i=1}{\overset{M}{\Σ}}\frac{p_n\left(f_i\right)P_C\left(f_i\right)}{{\[{\stackrel{\‾}{\mathrm{\σ}}}_{n-X}{P}_L\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_n{P}_C\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_{n+X}{P}_R\left(f_i\right)+\frac{N_0}{F_r}\]}^2}\\ +\underset{i=1}{\overset{M}{\Σ}}\frac{p_{n-X}\left(f_i\right)P_R\left(f_i\right)}{{\[{\stackrel{\‾}{\mathrm{\σ}}}_{n-2X}{P}_L\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_{n-X}{P}_C\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_n{P}_R\left(f_i\right)+\frac{N_0}{F_r}\]}^2}+\underset{i=1}{\overset{M}{\Σ}}\frac{p_{n+X}\left(f_i\right)P_L\left(f_i\right)}{{\[{\stackrel{\‾}{\mathrm{\σ}}}_n{P}_L\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_{n+X}{P}_C\left(f_i\right)+{\stackrel{\‾}{\mathrm{\σ}}}_{n+2X}{P}_R\left(f_i\right)+\frac{N_0}{F_r}\]}^2}\\ +\frac{\mathrm{\α}}{\[1+\left(\mathrm{\α}{\stackrel{\‾}{\mathrm{\σ}}}_n\right)\]\[\frac{\mathrm{\π}}{2}+\arctan \left(\mathrm{\α}{\stackrel{\‾}{\mathrm{\σ}}}_n\right)\]}=0\end{array};$

Wherein, M is the points of the doppler spectral of each subimage block, F_rIt is the radar pulse repetition frequency of SAR system, P_c(f_i)、 P_L(f_i)、P_R(f_i) it is respectively the energy spectrum of i-th of Doppler frequency spectrum main lobe, left secondary lobe and right secondary lobe points, p_n-X(f_i)、p_n (f_i)、p_n+X(f_i) it is the n-th-X, the energy of i-th points of the Doppler frequency spectrum of n-th and the n-th+X subimage block respectively Spectrum, α represent antenna elevation angle,The n-th -2X, the n-th-X respectively, n-th, The relative scattering value of the n-th+X and the n-th+2X subimage block, i is the natural number more than or equal to 1.

4. the method for estimation of the low scattering region scattering value of SAR image based on frequency spectrum as claimed in claim 1, its feature exist In the formula of the maximal possibility estimation is expressed as：

$\begin{array}{c}\frac{\∂\ln g\left(\begin{array}{cccc}{p}_n\left(f_1\right)& p_n\left(f_2\right)& \mathrm{...}& p_n\left(f_M\right)|{\stackrel{\‾}{\mathrm{\σ}}}_n\end{array}\right)}{\∂{\stackrel{\‾}{\mathrm{\σ}}}_n}+\frac{\∂\ln g\left(\begin{array}{cccc}{p}_{n-X}\left(f_1\right)& p_{n-X}\left(f_2\right)& \mathrm{...}& p_{n-X}\left(f_M\right)\end{array}|{\stackrel{\‾}{\mathrm{\σ}}}_n\right)}{\∂{\stackrel{\‾}{\mathrm{\σ}}}_n}\\ +\frac{\∂\ln g\left(\begin{array}{cccc}{p}_{n+X}\left(f_1\right)& p_{n+X}\left(f_2\right)& \mathrm{...}& p_{n+X}\left(f_M\right)\end{array}|{\stackrel{\‾}{\mathrm{\σ}}}_n\right)}{\∂{\stackrel{\‾}{\mathrm{\σ}}}_n}+\frac{\∂\ln g_p\left({\stackrel{\‾}{\mathrm{\σ}}}_n\right|)}{\∂{\stackrel{\‾}{\mathrm{\σ}}}_n}=0\end{array};$

Wherein, the g (p in above formula_n(f₁)p_n(f₂)…p_n(f_M))、g(p_n-X(f₁)p_n-X(f₂)…p_n-X(f_M))、g(p_n+X(f₁)p_n+X (f₂)…p_n+X(f_M)) be n-th respectively, the probability density function of the n-th-X and the n-th+X subimage block, p_n(f_i) it is n-th The energy spectrum of i-th Frequency point of subimage block.

5. the method for estimation of the low scattering region scattering value of SAR image based on frequency spectrum as claimed in claim 1, its feature exist In the absolute scatter value of each image block according to below equation is calculated：

${\mathrm{\σ}}_{cal}(x_0,y_0)=10lg\[\stackrel{\‾}{\mathrm{\σ}}(x_0,y_0)\]-K+10lg\[{\left(\frac{R}{R_{ref}}\right)}^3\frac{sin\mathrm{\α}}{{\mathrm{sin\α}}_{ref}}\frac{G_{ref}}{g\left(\mathrm{\α}\right)G}\];$

Wherein, R, α and G represent the system gain of target oblique distance, antenna elevation angle and specific pixel respectively, and g (α) for the angle of pitch is Two-way antenna gain during α, K is scaling constant, R_ref、α_refAnd G_refThe oblique distance of reference position, the angle of pitch are represented respectively and are System gain.

6. the method for estimation of the low scattering region scattering value of SAR image based on frequency spectrum as claimed in claim 1, its feature exist In the initial value of the Newton iteration method is the scattered power initial value of each subimage block.

7. the method for estimation of the low scattering region scattering value of SAR image based on frequency spectrum as claimed in claim 1, its feature exist In, when generating haplopia complex pattern in the step 1, the compression of orientation is adopted haves no right matched filter.

8. the method for estimation of the low scattering region scattering value of SAR image based on frequency spectrum as claimed in claim 7, its feature exist In the orientation energy spectrum of the haplopia complex pattern has the identical aspect of model with the SAR initial datas：

p_s(f, x₀, y₀)=p_r(f, x₀, y₀)|H(f)|²=p_r(f, x₀, y₀)；

Wherein,It is that orientation haves no right matched filter, p_s(f, x₀, y₀) and p_r(f, x₀, y₀) be respectively The energy spectrum of haplopia complex pattern and SAR initial datas, f are doppler centroid, and λ is SAR system wavelength, and V is SAR platform Movement velocity.