CN1089442C

CN1089442C - Non-linear compressor for original data of synthetic aperture radar

Info

Publication number: CN1089442C
Application number: CN96114076A
Authority: CN
Inventors: 王岩飞
Original assignee: Institute of Electronics of CAS
Current assignee: Institute of Electronics of CAS
Priority date: 1996-12-27
Filing date: 1996-12-27
Publication date: 2002-08-21
Anticipated expiration: 2016-12-27
Also published as: CN1171555A

Abstract

The present invention relates to a non-linear compressor for original data of a synthetic aperture radar. Signals of the original data of the synthetic aperture radar directly enter a non-linear encoder after quantized by A/D. The present invention has the advantages of simple circuits and wide dynamic ranges and is particularly suitable for real-time compression of original data of a space-borne synthetic aperture radar.

Description

Non-linear compressor for original data of synthetic aperture radar

The invention belongs to synthetic-aperture radar, particularly the non-linear compressor of satellite-borne synthetic aperture radar.

Synthetic aperture radar (SAR) passes through the research and development of decades as a kind of microwave remote sensing means of widespread use, develops into spaceborne system from mobile system.Because the work characteristics of SAR system, the satellite that big data quantity causes become a key issue of the necessary solution of development satellite-borne SAR institute to the high data rate problem on ground.Though, can take in from the system index design for the solution of High Data Rate problem, for example, reduce resolving power, mapping bandwidth etc.; Also can take in, for example, reduce visual dynamic range, radiation resolving power etc. from the design of image parameter index.But these considerations that solve the data volume problems all are that quality with final reduction radar image is a cost.Therefore, the method that reduces data transfer rate by data compression technique under the preceding topic that does not reduce or reduce less image quality has as far as possible obtained increasing attention.

In polarization sensitive synthetic aperture radar system, raw data is different from pictorial data, is the stack of the many little scatterer response that receives of radar antenna, can equivalence be the mutually independent random variables sum.Because the area of radar antenna irradiation is very big, according to central limit theorem, the stack of each little scatterer response can be approximated to be Gaussian distribution.Compress for this radar raw data, cause losing of information always usually, thereby have influence on the quality of image with high entropy.In fact data compression problem is exactly to trade off between the complexity of image quality and data compression algorithm complexity and realization according to different purposes requirements.

The quantification of branch block floating point and the block adaptive quantization method that have been used on Mai Zhelun synthetic-aperture radar and the space shuttle synthetic-aperture radar all are the data compression methods that the higher bit data is converted to low bits of encoded according to the partial statistics characteristic of radar signal, see R.Kwok andW.T.K.Johnson, " Block Adaptive Quantization of Magellan SAR Data. " IEEETrans.Geos.and Remote Sensing vol.GES-27.1989.T.H.Joo, D.N.Held, " AnAdaptive Quzntization Method For Burst Mode SAR " IEEE InternationalRadar Conference, 1985.Bryan L.Huneycutt " Spaceborne Imaging Radar-Cinstrument ", IEEE Trans.on Geoscience and Remote Sensing, Vol.27, No.2,1989.These two kinds of data compression methods all belong to the adaptive quantizing method, and the key distinction is that coding method is slightly different.The adaptive quantizing method is to carry out the method for adaptive quantizing according to the statistical parameter that calculates the radar data that extracts, owing to will carry out statistical computation constantly to the radar return data, so hardware circuit implements more complicated, its principle of work as shown in Figure 1.

The purpose of this invention is to provide a kind of hardware and realize simply, keep radar signal to have non-linear compressor than wide dynamic range with low Bit data.

Principal feature of the present invention is that the original data of synthetic aperture radar signal directly enters non-linear encoder after A/D quantizes.

Circuit of the present invention is simple, and wide dynamic range is specially adapted to the raw data Real Time Compression of satellite-borne synthetic aperture radar.

Description of drawings

Fig. 1 is that block adaptive quantizes the compressor reducer principle

Fig. 2 is the signal to noise ratio (S/N ratio) graph of relation of nonlinear quantization

Fig. 3 is the circuit theory diagrams of non-linear compressor.

Below in conjunction with accompanying drawing in detail the present invention is described in detail.In uniform quantization, quantization step is apart from irrelevant with the signal level size.The maximum instantaneous value of quantization error equals half of quantization step distance.So signal level is low more, signal to noise ratio (S/N ratio) is more little; Signal level is high more, and signal to noise ratio (S/N ratio) is big more.Certainly, when signal level surpasses certain limit, can produce the supersaturation noise signal to noise ratio (S/N ratio) is descended.

The idea of nonlinear quantizer produces at the deficiency that overcomes uniform quantizer.We wish that quantization step changes apart from the size that can follow incoming signal level.When signal level is low, apart from quantification, then use big quantization step apart from quantification to high level signal with little quantization step.The ratio that so just makes input signal and quantizing noise is consistent basically in the gamut of large-signal at small-signal, and much less when large-signal quantized required quantification progression than uniform quantization.

At spaceborne synthetic aperture radar (SAR) system, the present invention has provided and has a kind ofly carried out the non-linear, digital Methods for Coding later in the A/D uniform quantization.By signal is carried out the uniform quantization of higher bit position, for example 8 bits carry out non-uniform encoding to low number of bits, for example 3 bits or 4 bits according to the size of signal again.Can reduce data volume so on the one hand; Can make the signal after the quantification have the dynamic range of broad on the other hand.This paper has provided from 8 bits to 4 bits and two kinds of non-uniform encoding schemes from 8 bits to 3 bits, and the signal to noise ratio (S/N ratio) curve that has calculated nonlinear quantization according to the statistical property of signal as shown in Figure 2.Wherein curve (1) is 4 bit uniform quantization signal to noise ratio (S/N ratio) relation curves, curve (2) is 3 bit uniform quantization signal to noise ratio (S/N ratio) relation curves, curve (3) be 8 bits to 4 bit non-uniform quantizing signal to noise ratio (S/N ratio) relation curves, curve (4) is that 8 bits are to 3 bit non-uniform quantizing signal to noise ratio (S/N ratio) relation curves.Nonlinear quantization although the maximum signal to noise ratio value can descend to some extent, but has the effect of broadening dynamic range of signals with respect to uniform quantization as can be seen.In polarization sensitive synthetic aperture radar system, be 4 if require the quantization bit figure place of signal, the signal to noise ratio (S/N ratio) of signal is not less than 12 decibels, and as can be seen from Figure, the dynamic range of signals of nonlinear quantization exceeds about 25 decibels than the dynamic range of signals of uniform quantization.

Choosing of compression scheme is that statistical property with signal is relevant.Signal with different statistical properties all has a corresponding optimal compression scheme, makes quantizing noise reach minimum.But also should be taken into account problems such as compression scheme is easy to realize in actual applications in circuit.Among the present invention, two kinds of non-uniform encoding schemes that we adopted from 8 bits to 4 bits with from 8 bits to 3 bits be expressed as respectively map section 0,1}, (1,2}{2, and 4}, 4,8}, { 8,16}, 16, and 32}, 32,64}, { 64,127} and { 0,1}, 1, and 7}, 7,32}, { 32,127}, and be odd symmetric.

As can be seen, the cataloged procedure of nonlinear quantizer and the statistical property of radar signal do not have direct relation from top discussion.Coding for radar data can directly be realized by coding circuit very simply.As shown in Figure 3, become the digital signal of 8 bits after radar I, Q quadrature demodulated signal quantize through A/D.Wherein the sign bit of 1 bit is directly exported, and the digital signal process non-linear encoder of other 7 bits is encoded to the digital signal of 2 bits, is encoded to the output of 3 bit signals at last.

In the circuit, crucial part is the realization of non-linear encoder.This partial circuit can be realized with the mode of tabling look-up, and promptly with the address signal of 7 Bit datas of importing as a ROM (read-only memory), the coded data that keeps is in advance therefrom read, and so just can finish the function of non-uniform encoding.The scheme that can adopt I, Q signal to quantize respectively to encode simultaneously in side circuit is simplified coding circuit.Also can adopt the programmable circuit device directly to encode in addition, can choose different encoding schemes easily.

Clearly as can be seen, the scheme of nonlinear quantization is very simple in circuit is realized, almost is equal to the scheme of uniform quantization, and does not also have additional data to enter number biography passage.Decode procedure is easy to realize too.But also should be noted that the non-uniform encoding scheme of this implemented in digital, is the data that by the A/D quantizer of higher bit number radar signal quantized to be encoded to later again low bit number.So require the A/D quantizer that certain figure place will be arranged.The figure place of A/D quantizer is the 8-16 bit.

Claims

1. non-linear compressor for original data of synthetic aperture radar is characterized in that comprising:

A/D converter is used to quantize the original data of synthetic aperture radar signal, and the output two paths of signals;

Non-linear encoder, the road signal of exporting for the A/D transducer carries out non-uniform encoding;

2. by the described non-linear compressor for original data of synthetic aperture radar of claim 1, it is characterized in that described A/D transducer quantization digit is the 8-16 bit.

3. by the described non-linear compressor for original data of synthetic aperture radar of claim 1, it is characterized in that described non-linear encoder encodes to one tunnel 7 Bit data of A/D variator output, described non-linear encoder is output as 2 bit digital signal.