CN102129073A - Two-stage filter method for synthetic aperture radar azimuth pre-processing - Google Patents

Abstract

The invention discloses a two-stage filter method for synthetic aperture radar (SAR) azimuth pre-processing. The method comprises the following steps of: setting the order of a filter according to the processing time requirement of an SAR real-time imaging processor and the magnitude requirement of data volume, selecting a proper filter according to the set order, pre-filtering the sampled receiving signals, performing down sampling on the pre-filtered data, performing high-performance filtration on the filtered data, and performing subsequent processing. By reducing the order of the filter, the data volume and the processing time can be reduced, and the hardware resource requirements of the real-time imaging processor can be reduced.

Description

A kind of method of two-stage filtering of synthetic-aperture radar Azimuth Preprocessing

Technical field

The present invention relates to synthetic aperture radar (SAR) signal Processing field, especially a kind of implementation method of two-stage filtering of synthetic-aperture radar Azimuth Preprocessing.

Background technology

Synthetic aperture radar (SAR) is a kind of microwave remote sensing means of carrying out imaging observation on a surface target.Have can the round-the-clock all weather operations advantage, thereby be widely used in fields such as national defence, natural resources exploration, disaster monitoring.The real time imagery processor by to the SAR system acquisition to target echo signal handle real-time formation radar image, be SAR system essential important component for application with ageing requirement.In common carried SAR system, in order to reduce azimuth ambiguity, improve the signal to noise ratio (S/N ratio) of image, the frequency of the direct impulse of radar emission often is higher than the pulsed frequency of requirement, is equivalent to the orientation to the bandwidth of the system bandwidth far above signal.The data volume of considering SAR is bigger, and the bandwidth of useful signal is relatively low, therefore in imaging processor, adopt the orientation to reduce data volume usually, reduce pressure, guarantee the performance index of radar image simultaneously abilities such as processor calculating, storages to the down-sampled technology of pre-filtering.

The FIR wave filter is adopted in the orientation pre-filtering of SAR real time imagery processor usually, considers the restriction of computing and buffer memory, and the selection of filter order generally all can be more moderate.On the one hand, exponent number can not be excessive, in order to avoid increase too much calculated amount and metadata cache amount; On the other hand, also will be under limited exponent number condition, the wave filter that design is optimized as far as possible.

Summary of the invention

The object of the present invention is to provide a kind of method of two-stage filtering of synthetic-aperture radar Azimuth Preprocessing, to realize the requirement fast and efficiently of synthetic-aperture radar Azimuth Preprocessing.

For achieving the above object, the invention provides a kind of method of two-stage filtering of synthetic-aperture radar Azimuth Preprocessing, its key step comprises:

Require and the size requirements of data volume is provided with the exponent number of wave filter according to processing time of real-time processor;

Select corresponding wave filter according to the exponent number that sets;

Received signal behind over-sampling is carried out pre-filtering;

Carry out down-sampled to the data after the pre-filtering;

Carry out high-performance filtering for filtered data;

The filtered data of high-performance are carried out subsequent treatment;

In the such scheme, the described processing time according to real-time processor requires and the size requirements of data volume is provided with the exponent number of wave filter, specifically comprise: that supposes the real time processing system requirement pre-filtering T.T. can not surpass T, and be N the computing times of the required many consumption of the every increase single order of wave filter, and then the exponent number of prefilter can not surpass T/N.

In the such scheme, describedly select corresponding wave filter according to the exponent number that sets, specifically comprise:, select the frequency domain filter of response according to the requirement of filter type, to the requirement of filter cutoff frequency, to the requirement of filter order, to the passband fluctuating of wave filter and the requirement of stopband attenuation.

In the such scheme, described received signal behind over-sampling is carried out pre-filtering, specifically comprises: suppose the orientation to sampling interval be Δ x, be the n time echo usefulness r (n Δ of the target of R in distance _x) represent, with y (m Δ _x) expression wave filter coefficient, then pre-processing filter is output as:

$q\left(k{\mathrm{\Δ}}_x\right)=\underset{m=0}{\overset{N-1}{\mathrm{\Σ}}}r(k{\mathrm{\Δ}}_x-m{\mathrm{\Δ}}_x)y\left(m{\mathrm{\Δ}}_x\right)$

In the such scheme, describedly data after the pre-filtering are carried out down-sampled, specifically comprise: obtaining q (k Δ _x) pre-filtering output after, undertaken down-sampledly by the method that signal is extracted, obtain down-sampled signal q (n γ Δ _x), wherein γ is down-sampled coefficient.

In the such scheme, describedly carry out high-performance filtering, specifically comprise for filtered data: the discrete signal after data volume is reduced, select the wave filter of high-order to carry out filtering once more, extract useful signal, for subsequent treatment provides high-quality data.

In the such scheme, described the filtered data of high-performance are carried out subsequent treatment, specifically comprise: according to the obtain manner and the flight parameter of radar data, proceed the orientation to Filtering Processing, perhaps directly in conjunction with the orientation to imaging handle.

From technique scheme as can be seen, the present invention has following beneficial effect:

1) the present invention has realized processing fast and effectively to the mass data of synthetic aperture radar return by two-stage filtering.

2) the present invention has obtained diameter radar image clearly, and also can expand to the multiple-stage filtering processing according to specific requirement.

Description of drawings

Fig. 1 is the process flow diagram of implementation method of the two-stage filtering of synthetic-aperture radar Azimuth Preprocessing provided by the invention.

Fig. 2 is two not spectrum diagram of same order wave filter; Wherein, Fig. 2 a is depicted as 55 rank hamming windows, and Fig. 2 b is depicted as the spectrogram of 9 rank hamming windows.

Fig. 3 is down-sampled latter two spectrum diagram of same order wave filter not; Wherein, Fig. 3 a is depicted as the filter spectrum synoptic diagram of 55 rank hamming windows after down-sampled, and Fig. 3 b is depicted as the filter spectrum synoptic diagram of 9 rank hamming windows after down-sampled.

Fig. 4 is the spectrum diagram of the Brackman wave filter on 9 rank of use during certain is handled; Wherein, Fig. 4 a is the amplitude spectrum of Brackman wave filter, and Fig. 4 b is the frequency domain amplitude spectrum behind corresponding wave filter 1/2 double sampling.

Fig. 5 is that the diameter radar image that obtains during certain is handled compares; Wherein, the SAR image of Fig. 5 a for adopting common pre-filtering method to obtain, the SAR image of Fig. 5 b for adopting two-stage filtering to obtain.

Embodiment

For making the purpose, technical solutions and advantages of the present invention clearer, below in conjunction with specific embodiment, and with reference to accompanying drawing, the present invention is described in more detail.

As shown in Figure 1, Fig. 1 is the process flow diagram of the implementation method of this paper two-stage filtering that the synthetic-aperture radar Azimuth Preprocessing that provides is provided, and this method comprises:

Step S1: require and the size requirements of data volume is provided with the exponent number of wave filter according to processing time of real-time processor;

Step S2: select corresponding wave filter according to the exponent number that sets;

Step S3: the received signal behind over-sampling is carried out pre-filtering;

Step S4: carry out down-sampled to the data after the pre-filtering;

Step S5: carry out high-performance filtering for filtered data;

Step S6: the filtered data of high-performance are carried out subsequent treatment;

The described processing time according to real-time processor of above-mentioned steps 1 requires and the size requirements of data volume is provided with the exponent number of wave filter, specifically comprise: that supposes the real time processing system requirement pre-filtering T.T. can not surpass T, and be N the computing times of the required many consumption of the every increase single order of wave filter, and then the exponent number of prefilter can not surpass T/N.

Above-mentioned steps 2 is described selects corresponding wave filter according to the exponent number that sets, specifically comprise:, select the frequency domain filter of response according to the requirement of filter type, to the requirement of filter cutoff frequency, to the requirement of filter order, to the passband fluctuating of wave filter and the requirement of stopband attenuation.

Above-mentioned steps 3 is described carries out pre-filtering to the received signal behind over-sampling, specifically comprises: suppose the orientation to sampling interval be Δ x, be the n time echo usefulness r (n Δ of the target of R in distance _x) represent, with y (m Δ _x) expression wave filter coefficient, then pre-processing filter is output as:

$q\left(k{\mathrm{\Δ}}_x\right)=\underset{m=0}{\overset{N-1}{\mathrm{\Σ}}}r(k{\mathrm{\Δ}}_x-m{\mathrm{\Δ}}_x)y\left(m{\mathrm{\Δ}}_x\right)$

Above-mentioned steps 4 is described carries out down-sampledly to the data after the pre-filtering, specifically comprise: obtaining q (k Δ _x) pre-filtering output after, undertaken down-sampledly by the method that signal is extracted, obtain down-sampled signal q (n γ Δ _x), wherein γ is down-sampled coefficient.

Above-mentioned steps 5 is described carries out high-performance filtering for filtered data, specifically comprises: the discrete signal after data volume is reduced, and select the wave filter of high-order to carry out filtering once more, extract useful signal, for subsequent treatment provides high-quality data.

Above-mentioned steps 6 is described carries out subsequent treatment to the filtered data of high-performance, specifically comprises: according to the obtain manner and the flight parameter of radar data, proceed the orientation to Filtering Processing, perhaps directly in conjunction with the orientation to imaging handle.

Provide the process that actual SAR system median filter is selected below.

1, the relation of filter process effect and exponent number: as shown in Figure 2, the hamming window is a kind of wave filter commonly used in the signal Processing, and it is functional, realizes also uncomplicated.Fig. 2 a is depicted as 55 rank hamming windows, and Fig. 2 b is depicted as the spectrogram of 9 rank hamming windows.As seen from the figure, the attenuation band of 9 rank hamming windows is higher than 55 rank hamming windows far away, thereby makes its filter effect insert in 55 rank hamming windows, and after oversampling, the wave filter aliasing influence that exponent number is high is smaller, and the wave filter aliasing influence that exponent number is low is bigger.But, if use 55 rank hamming windows, could begin filtering after in real-time stream treatment, will storing 55 range lines, data volume by actual SAR system is calculated, the signal sampling of radar is counted and is the 32k byte, pulse repetition rate (PRF) is 2000hz, considers to be I, Q two paths of signals, then 55 range line 32k*55*2=3520k bytes after the radar mean frequency demodulation.This is for dealing with the sizable data volume of saying so in real time.So wish again to select the lower wave filter of exponent number, so just produced contradiction between data volume and the filter effect.

2, carry out down-sampled, with certain carried SAR system is example, and the signal sampling of radar is counted and is the 32k byte, and pulse repetition rate (PRF) is 2000hz, be I, Q two paths of signals after considering the radar mean frequency demodulation, the duration data rate of SAR radar is about the 128M byte per second.Such data transfer rate is born very heavy for processor, so will carry out down-sampled.With 1/2 down-sampled be example, the data transfer rate after the sampling is the 64M byte per second, and very big minimizing has been arranged.Filter spectrum after down-sampled as shown in Figure 3.

Yet 3, analysis chart 3a and Fig. 3 b can also find, if the frequency span of useful signal in 0.1 time, although the exponent number of wave filter is lower, the influence to useful signal that aliasing brings among Fig. 3 b is also smaller.If in follow-up Filtering Processing, for example carry out the filtering second time or compress processing etc. in conjunction with the orientation, the garbage signal outside the filtering useful signal then can obtain reasonable effect equally effectively.

Usually, the frequency domain characteristic of desirable Azimuth Preprocessing wave filter is

$H_d\left(e^{\mathrm{j\&omega;}}\right)=\left\{\begin{array}{cc}{e}^{-\mathrm{j\&omega;\&alpha;}},& -{\mathrm{\&omega;}}_c\&le;\mathrm{\&omega;}\&le;{\mathrm{\&omega;}}_{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="H2010100344059E00021.png,H2010100344059E00022.png"\; state="\{\{state\}\}">c</figure-callout>}}\\ 0,& {\mathrm{\&omega;}}_c\&le;\mathrm{\&omega;}\&le;1,-1\&le;\mathrm{\&omega;}\&le;-{\mathrm{\&omega;}}_c\end{array}\right.$

And in the secondary filter processing method that the present invention proposes, the ideal frequency domain characteristic of first order wave filter is required and can be expressed as

$H_d\left(e^{\mathrm{j\&omega;}}\right)=\left\{\begin{array}{cc}{e}^{-\mathrm{j\&omega;\&alpha;}},& -{\mathrm{\&omega;}}_c\&le;\mathrm{\&omega;}\&le;{\mathrm{\&omega;}}_{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="H2010100344059E00021.png,H2010100344059E00022.png"\; state="\{\{state\}\}">c</figure-callout>}}\\ 0,& {1-\mathrm{\&omega;}}_c\&le;\mathrm{\&omega;}\&le;1,-1\&le;\mathrm{\&omega;}\&le;-1+{\mathrm{\&omega;}}_c\end{array}\right.$

As can be seen at (ω _c, 1-ω _c) and (1+ ω _c,-ω _c) the interval in do not claim, this is equivalent to loosen greatly the Filter Design requirement.Also may provide theoretical foundation for what filter order reduced.

From top analysis as can be seen, carrying out two-stage filtering needs certain condition, requires sample frequency to want enough big with respect to signal bandwidth on the one hand, can satisfy low order Filtering Processing and down-sampled requirement, useful signal is not formed bigger influence simultaneously; On the other hand.The garbage signal outside the useful signal frequency spectrum can be effectively rejected in requirement in second level filtering and subsequent treatment.

4, according to above analysis, employing is carried out first order filtering as the Brackman wave filter on 9 rank of Fig. 4 in certain carried SAR imaging processing.Fig. 4 a is the amplitude spectrum of Brackman wave filter, and Fig. 4 b is the frequency domain amplitude spectrum behind corresponding wave filter 1/2 double sampling.For aviation SAR system, general pulse repetition rate is much larger than the Doppler frequency bandwidth of SAR radar, shown in Fig. 4 b.

5, through after the filtering double sampling, at effective signal bandwidth with interior aliasing influence even be less than the aliasing influence of the high-order wavenumber filter among Fig. 3 a.Follow-up processing can proceed the orientation to Filtering Processing, perhaps directly in conjunction with the orientation to imaging handle.As example, we utilize the SAR real data of the same area to carry out the comparative analysis of different disposal.The SAR image of Fig. 5 a for adopting common pre-filtering method to obtain, the SAR image of Fig. 5 b for adopting two-stage filtering to obtain.As can be seen, the image that utilizes two kinds of methods to handle to obtain is the same with preceding surface analysis, does not have tangible difference.To the result of this real data, the correctness and the validity of the two-stage filter processing method that this paper proposed has been described also.

Claims (7)

1. the method for the two-stage filtering of a synthetic-aperture radar Azimuth Preprocessing, its key step comprises:

Require and the size requirements of data volume is provided with the exponent number of wave filter according to processing time of real-time processor;

Select corresponding wave filter according to the exponent number that sets;

Received signal behind over-sampling is carried out pre-filtering;

Carry out down-sampled to the data after the pre-filtering;

Carry out high-performance filtering for filtered data;

The filtered data of high-performance are carried out subsequent treatment.

2. the method for the two-stage filtering of synthetic-aperture radar Azimuth Preprocessing according to claim 1, wherein, the described processing time according to real-time processor requires and the size requirements of data volume is provided with the exponent number of wave filter, comprising:

That sets the real time processing system requirement is no more than T pre-filtering T.T., and be N the computing times of the required many consumption of the every increase single order of wave filter, and then the exponent number of prefilter can not surpass T/N.

3. the implementation method of the two-stage filtering of synthetic-aperture radar Azimuth Preprocessing according to claim 1 wherein, is selected corresponding wave filter according to the exponent number that sets, and comprising:

According to the requirement of filter type, to the requirement of filter cutoff frequency, to the requirement of filter order, to the passband fluctuating of wave filter and the requirement of stopband attenuation, select the frequency domain filter of response.

4. the implementation method of the two-stage filtering of synthetic-aperture radar Azimuth Preprocessing according to claim 1 wherein, is carried out pre-filtering to the received signal behind over-sampling, comprising:

Suppose the orientation to sampling interval be Δ x, be the n time echo usefulness r (n Δ of the target of R in distance _x) represent, with y (m Δ _x) expression wave filter coefficient, then pre-processing filter is output as:

$q\left(k{\mathrm{\&Delta;}}_x\right)=\underset{m=0}{\overset{N-1}{\mathrm{\&Sigma;}}}r(k{\mathrm{\&Delta;}}_x-m{\mathrm{\&Delta;}}_x)y\left(m{\mathrm{\&Delta;}}_x\right)$

5. the implementation method of the two-stage filtering of synthetic-aperture radar Azimuth Preprocessing according to claim 1 wherein, is carried out down-sampledly to the data after the pre-filtering, comprising:

Obtaining q (k Δ _x) pre-filtering output after, undertaken down-sampledly by the method that signal is extracted, obtain down-sampled signal q (n γ Δ _x), wherein γ is down-sampled coefficient.

6. the implementation method of the two-stage filtering of synthetic-aperture radar Azimuth Preprocessing according to claim 1 wherein, is carried out high-performance filtering for filtered data, comprising:

Discrete signal after data volume reduced selects the wave filter of high-order to carry out filtering once more, extracts useful signal, for subsequent treatment provides high-quality data.

7. the implementation method of the two-stage filtering of synthetic-aperture radar Azimuth Preprocessing according to claim 1 wherein, is carried out subsequent treatment to the filtered data of high-performance, comprising:

According to the obtain manner and the flight parameter of radar data, proceed the orientation to Filtering Processing, perhaps directly in conjunction with the orientation to imaging handle.

Images