CN100357759C

CN100357759C - A self-focusing processing method for synthetic aperture sonar imaging

Info

Publication number: CN100357759C
Application number: CNB200510060491XA
Authority: CN
Inventors: 丛卫华; 张峰山; 谭咏青; 何卫彬; 董彦涛; 朱必波; 章灵祥; 何永军; 曹海林; 施国全; 付菊英; 王欣; 杨晓帆
Original assignee: Hangzhou Institute of Applied Acoustics
Current assignee: Hangzhou Ruili Marine Equipment Co ltd; Hangzhou Institute of Applied Acoustics
Priority date: 2005-08-25
Filing date: 2005-08-25
Publication date: 2007-12-26
Anticipated expiration: 2025-08-25
Also published as: CN1731212A

Abstract

The invention relates to a self-focusing processing method for synthetic aperture sonar imaging, which mainly includes the following steps: (1), each echo signal undergoes complex baseband demodulation and matched filtering processing to obtain a broadband analysis signal r _p (t); (2) ), correlation processing between adjacent echo analysis signals; (3), obtain the phase error estimation between adjacent echo analysis signals by calculating the complex phase angle; (4), obtain motion error caused by spatial position accumulation processing The cumulative phase error of ; (5), use the obtained phase error sequence {ε _p } to compensate for the echo data correction; (6), apply the corresponding image reconstruction algorithm to the compensated data, and then the focused synthetic aperture can be obtained sonar image. The beneficial effect of the invention is that it does not need iterative calculation, and has better real-time calculation performance and better image macroscopic correction ability. And the impact of strong target scattering points and noise on motion error estimation is further weakened by α weighting, thus broadening the application conditions.

Description

A Self-focusing Processing Method for Synthetic Aperture Sonar Imaging

技术领域technical field

本发明涉及合成孔径声纳成像的信号处理，主要是一种合成孔径声纳成像的自聚焦处理方法，用于获得更加精细的海底地貌测绘和沉物探测的声成像。The invention relates to signal processing of synthetic aperture sonar imaging, and mainly relates to a self-focusing processing method of synthetic aperture sonar imaging, which is used to obtain more precise acoustic imaging of seabed landform surveying and sinking object detection.

背景技术Background technique

合成孔径声纳(Synthetic Aperture Sonar，简称SAS)是一种高分辨力声学成像设备。它是利用小孔径基阵匀速直线运动，在运动过程中相继发射和接收信号，并将一段运动历程中接收信号进行相干迭加，从而得到等效于实际物理声阵几倍到几十倍的合成虚拟孔径阵，获得很高的分辨力和空间增益。然而在实际航行过程中这个条件并不满足，使得合成孔径图像产生强度失真和几何失真。实现运动误差补偿的自聚焦处理方法是合成孔径声纳的一项关键技术。Synthetic Aperture Sonar (SAS for short) is a high-resolution acoustic imaging device. It uses a small-aperture matrix to move in a straight line at a uniform speed, successively transmits and receives signals during the movement, and coherently superimposes the received signals during a period of movement, so as to obtain an acoustic array equivalent to several times to dozens of times the actual physical sound array. Synthetic virtual aperture array to obtain high resolution and spatial gain. However, this condition is not satisfied during the actual voyage, resulting in intensity distortion and geometric distortion in the synthetic aperture image. The autofocus processing method to realize motion error compensation is a key technology of synthetic aperture sonar.

合成孔径声纳的估计和校正方法通常分为两类：一类为采用传感器的硬件补偿，另一类为自聚焦算法补偿。由于传感器精度难于满足合成孔径声纳的要求，合成孔径声纳一般都需要采用自聚焦算法来实现运动误差补偿。自聚焦算法通常又可分为数据域自聚焦算法(如冗余相位中心自聚焦算法)和图像域自聚焦算法(如相位梯度自聚焦)算法两种。Estimation and correction methods of synthetic aperture sonar are generally divided into two categories: one is hardware compensation using sensors, and the other is self-focus algorithm compensation. Because the sensor accuracy is difficult to meet the requirements of synthetic aperture sonar, synthetic aperture sonar generally needs to use self-focusing algorithm to realize motion error compensation. Autofocus algorithms can usually be divided into data domain autofocus algorithms (such as redundant phase center autofocus algorithms) and image domain autofocus algorithms (such as phase gradient autofocus) algorithms.

国外合成孔径声纳所采用的自聚焦算法主要是冗余相位中心(Redundance Phase Center，简称RPC)自聚焦算法，它是利用接收回波信号的冗余性来处理的，需要采用冗余子阵设计或牺牲空间采样来实现。经典的相位梯度自聚焦(Phase Gradient Autofocus，简称PGA)算法，需要进行迭代处理才具有良好的估计性能，同时经典自聚焦处理方法需要选择强目标散射点。The self-focusing algorithm adopted by foreign synthetic aperture sonar is mainly the Redundance Phase Center (RPC) self-focusing algorithm, which uses the redundancy of the received echo signal to process, and needs to use redundant sub-arrays Design or sacrifice spatial sampling to implement. The classic Phase Gradient Autofocus (PGA for short) algorithm requires iterative processing to have good estimation performance, and the classic autofocus processing method needs to select strong target scattering points.

发明内容Contents of the invention

本发明的目的是为了克服上述方法的不足，而提供一种合成孔径声纳成像的自聚焦处理方法。The object of the present invention is to provide a self-focusing processing method for synthetic aperture sonar imaging in order to overcome the shortcomings of the above methods.

本发明解决其技术问题所采用的技术方案。这种合成孔径声纳成像的自聚焦处理方法，主要包括以下步骤：The technical solution adopted by the present invention to solve its technical problems. This self-focusing processing method for synthetic aperture sonar imaging mainly includes the following steps:

(1)、每次回波信号经过复基带解调以及匹配滤波处理得到宽带解析信号r_p(t)；(1), each echo signal undergoes complex baseband demodulation and matched filter processing to obtain a broadband analysis signal r _p (t);

(2)、相邻回波解析信号之间相关处理；相邻两个发射脉冲回波的宽带解析信号r_p(t)、r_p-1(t)的互相关运算为：(2), correlation processing between adjacent echo analysis signals; the cross-correlation operation of broadband analysis signals r _p (t) and r _p-1 (t) of adjacent two transmitted pulse echoes is:

${rr rr}_{p p} = = {&Integral; &Integral;}_{00}^{T T} {r r}_{p p}^{* *} ((t t)) {r r}_{p p - - 11} ((t t)) dt dt - - - - - - ((55))$

式中表示复共轭；In the formula, it means complex conjugate;

(3)、通过求复数相角，得到相邻回波解析信号之间的相位误差估计：(3), by calculating the complex phase angle, the phase error estimation between adjacent echo analysis signals is obtained:

$Δ Δ {\overset{^^}{φ φ}}_{p p} = = Arg Arg {{{rr rr}_{p p}}} - - - - - - ((66))$

(4)、通过空间位置累积处理获得运动误差引起的累积相位误差：(4), Accumulated phase error caused by motion error is obtained through spatial position accumulation processing:

${\overset{^^}{ϵ ϵ}}_{p p} = = {\overset{^^}{ϵ ϵ}}_{p p - - 11} + + Δ Δ {\overset{^^}{φ φ}}_{p p};;$ ${\overset{^^}{ϵ ϵ}}_{00} &equiv; &equiv; 00 - - - - - - ((77))$

(5)、利用获得的相位误差序列

进行补偿来实现回波数据校正：(5), using the obtained phase error sequence

Perform compensation to achieve echo data correction:

${R R}_{p p}^{ldeal ldeal} ((f f)) = = {R R}_{p p} ((f f)) exp exp ((j j {\overset{^^}{ϵ ϵ}}_{p p})) - - - - - - ((88))$

式中，R_p(f)为r_p(t)的傅立叶变换；where R _p (f) is the Fourier transform of r _p (t);

(6)、补偿后数据应用相应的图像重构算法，则可获得聚焦的合成孔径声纳图像。(6) Applying the corresponding image reconstruction algorithm to the compensated data, a focused synthetic aperture sonar image can be obtained.

本发明的有益的效果是：它是一种利用合成孔径声纳混响中冗余信息的数据域相关处理的运动误差估计方法。采用以均自聚焦算法为核心的、通过测量宽带解析信号复相关的相位角进行合成孔径声纳运动误差估计和补偿，从而获得合成孔径声纳成像的自聚焦处理。它不需要迭代运算，并具有较好的实时运算性能和较好的图像宏观校正能力。并且进一步通过α加权削弱强目标散射点和噪声对运动误差估计的影响，使得切变平均自聚焦算法既适应于空间缓变场景的宽波束SAS处理；也适用于存在强目标散射点的场景，从而拓宽了应用条件。The beneficial effect of the present invention is that it is a motion error estimation method utilizing data domain correlation processing of redundant information in synthetic aperture sonar reverberation. The self-focusing processing of synthetic aperture sonar imaging is obtained by using the homogeneous self-focusing algorithm as the core and estimating and compensating the motion error of synthetic aperture sonar by measuring the phase angle of complex correlation of broadband analytical signals. It does not need iterative operation, and has better real-time operation performance and better image macro correction ability. Moreover, the influence of strong target scattering points and noise on motion error estimation is further weakened by α weighting, so that the shear average self-focusing algorithm is not only suitable for wide-beam SAS processing of space slowly changing scenes, but also suitable for scenes with strong target scattering points. Thereby widening the application conditions.

附图说明Description of drawings

图1本发明实施例2的α加权切变平均自聚焦处理方法的流程图；Fig. 1 is a flow chart of the α-weighted shear average self-focus processing method of Embodiment 2 of the present invention;

具体实施方式Detailed ways

下面结合附图和实施例对本发明作进一步介绍：The present invention will be further described below in conjunction with accompanying drawing and embodiment:

实施例1：Example 1:

一、基于混响估计的运动误差模型1. Motion error model based on reverberation estimation

对于一个混响限制(混响占主导地位，而噪声相对较弱)慢变化场景海底模型，假定海底的各个散射体的尺寸远小于波长，可以把它们看作“点源”目标，在这种近似下可以不考虑频率和波束方向对散射的影响，这样混响模型可看为很多点源目标回波的线性之和。如果在测绘区域内共有Q个散射体，则第p个发射脉冲的回波可描述为：For a reverberation-limited (the reverberation dominates, while the noise is relatively weak) slow-changing scene seabed model, assuming that the size of each scatterer on the seabed is much smaller than the wavelength, they can be regarded as "point source" targets, in this In approximation, the influence of frequency and beam direction on scattering can be ignored, so the reverberation model can be regarded as a linear sum of many point source target echoes. If there are Q scatterers in the survey area, the echo of the p-th transmitted pulse can be described as:

${e e}_{p p} ((t t)) = = {Σ Σ}_{q q = = 11}^{Q Q} {s the s}_{p p} ((t t)) &CircleTimes; &CircleTimes; h h ((t t,, {τ τ}_{pq pq})) + + {n no}_{p p} ((t t)) - - - - - - ((11))$

式中表示卷积操作，s_p(t)是宽带发射信号，h(t，τ_pq)为点扩展函数(包括介质中传播函数，散射体散射响应以及基阵响应函数)，n_p(t)为噪声项。在没有运动误差时，回波信号经过复基带解调以及匹配滤波处理得到解析信号r_p(t)，表示为：where  represents the convolution operation, _sp (t) is the broadband transmission signal, h(t, τ _pq ) is the point spread function (including the propagation function in the medium, the scattering response of the scatterer and the matrix response function), n _p ( t) is the noise term. When there is no motion error, the echo signal undergoes complex baseband demodulation and matched filter processing to obtain the analytical signal r _p (t), expressed as:

${r r}_{p p}^{ldeal ldeal} ((t t)) = = | | {r r}_{p p}^{ldeal ldeal} ((t t)) | | exp exp ((j j {ψ ψ}_{p p}^{ldeal ldeal} ((t t)))) - - - - - - ((22))$

由于混响场中散射体服从均匀分布，接收回波为各个散射体散射信号之和，故而ψ_p ^ldeal(t)为[-π，π]内均匀分布的随机变量。Since the scatterers in the reverberation field obey the uniform distribution, the received echo is the sum of the scattered signals of each scatterer, so ψ _p ^ldeal (t) is a random variable uniformly distributed in [-π, π].

假设在声学换能器基阵发生摆动误差x_p，而散射体的主体近似不变，则对于场景中每一个散射体的回波延时误差都等于ε_p≈2x_p/c，第p个回波可表示为Assuming that the swing error x _p occurs in the acoustic transducer matrix, and the main body of the scatterer is approximately unchanged, the echo delay error for each scatterer in the scene is equal to ε _p ≈ 2x _p /c, the pth The echo can be expressed as

${r r}_{p p} ((t t)) = = {r r}_{p p}^{ldeal ldeal} ((t t - - {ϵ ϵ}_{p p})) exp exp ((- - j j 22 π π {f f}_{00} {ϵ ϵ}_{p p}))$

$= = | | {r r}_{p p}^{ldeal ldeal} ((t t - - {ϵ ϵ}_{p p})) | | exp exp ((j j {ψ ψ}_{p p}^{ldeal ldeal} ((t t - - {ϵ ϵ}_{p p})))) exp exp ((- - j j 22 π π {f f}_{00} {ϵ ϵ}_{p p})) - - - - - - ((33))$

故而获得测量相位为Therefore, the measured phase is obtained as

${ψ ψ}_{p p} ((t t)) = = Arg Arg {{{r r}_{p p} ((t t))}} = = [[{ψ ψ}_{p p}^{ldeal ldeal} ((t t - - {ϵ ϵ}_{p p})) - - 2π 2π {f f}_{00} {ϵ ϵ}_{p p}]] mod mod 22 π π - - - - - - ((44))$

式中Arg{·}为取复数相位操作。In the formula, Arg{·} is a complex phase operation.

如果我们能够对每一个ψ_p(t)进行去卷绕处理，则获得一个常量相移2πf₀ε_p和一个随机变量ψ_p ^ldeal(t-ε_p)，进一步通过在时间t上做平均处理就可获得常量相移，从而求出误差时延估计

If we can dewarp each ψ _p (t), we get a constant phase shift 2πf ₀ ε _p and a random variable ψ _p ^ldeal (t-ε _p ), which is further averaged over time t The constant phase shift can be obtained, so as to obtain the error delay estimate

二、合成孔径声纳成像的切变平均自聚焦处理方法2. Shear-averaged self-focusing processing method for synthetic aperture sonar imaging

(1)每次回波信号经过复基带解调以及匹配滤波处理得到宽带解析信号r_p(t)。(1) Each echo signal undergoes complex baseband demodulation and matched filtering processing to obtain a broadband analysis signal r _p (t).

(2)相邻回波解析信号之间相关处理。(2) Correlation processing between adjacent echo analysis signals.

利用(4)式中单个脉冲回波在时间t上做平均处理获得的常量误差相位2πf₀ε_p一般大于2π从而会导致相位模糊问题。一个可行的解决这种相位卷绕的方法就是估计相邻两个发射回波信号之间的相位差，其核心是一种相关处理。相邻两个发射脉冲回波的宽带解析信号r_p(t)、r_p-1(t)的互相关运算为：The constant error phase 2πf ₀ ε _p obtained by averaging a single pulse echo over time t in formula (4) is generally greater than 2π, which will lead to phase ambiguity. A feasible method to solve this phase wrapping is to estimate the phase difference between two adjacent transmitted echo signals, the core of which is a correlation process. The cross-correlation operation of the broadband analytical signals r _p (t) and r _p-1 (t) of two adjacent transmitted pulse echoes is:

式中^*表示复共轭。where ^* means complex conjugate.

(3)通过求复数相角，得到相邻回波解析信号之间的相位误差估计：(3) By calculating the complex phase angle, the phase error estimation between adjacent echo analysis signals is obtained:

(4)通过空间位置累积处理获得运动误差引起的累积相位误差(4) Obtain the cumulative phase error caused by the motion error through spatial position accumulation processing

(5)利用获得的相位误差序列

进行补偿就可实现回波数据校正：(5) Using the obtained phase error sequence

Echo data correction can be achieved by performing compensation:

式中，R_p(f)为r_p(t)的傅立叶变换。In the formula, R _p (f) is the Fourier transform of r _p (t).

(6)补偿后数据应用相应的图像重构算法，则可获得聚焦的合成孔径声纳图像。(6) Apply the corresponding image reconstruction algorithm to the compensated data, and then the focused synthetic aperture sonar image can be obtained.

实施例2：α加权切变平均自聚焦处理方法Embodiment 2: α-weighted shear average self-focus processing method

为了提高误差相位变化量估计的精度，K.A.Johnson等人提出最小均方意义加权平均核心，选择一种加权函数ω(t)＝r_p(t)|使其侧重于信噪比较大的相位差，从而有效抑制噪声相位的影响。使用加权方法对(5)式相关函数重新估计如下：In order to improve the accuracy of the estimation of error phase variation, KAJohnson et al. proposed the least mean square weighted average kernel, and selected a weighting function ω(t)=r _p (t)| to make it focus on the phase difference with a large signal-to-noise ratio , so as to effectively suppress the influence of noise phase. Using the weighting method to re-estimate the correlation function of (5) as follows:

$rr rr = = {{\frac{{&Integral; &Integral;}_{00}^{T T} {ω ω}^{22} ((t t)) {r r}_{p p}^{* *} ((t t)) {r r}_{p p - - 11} ((t t)) dt dt}{{&Integral; &Integral;}_{00}^{T T} {ω ω}^{22} ((t t)) dt dt}}} - - - - - - ((99))$

但是K.A.Johnson加权方法在抑制噪声影响的同时也加重了强目标散射体对相位函数的贡献。当有强目标存在时，(4)式中的ψ_p ^ldeal(t)不再能够视为一个[-π，π]内均匀分布随机变量，此时它主要表现为强目标的距离徙动相位，故而通过(6)式估计的时延变化量受到距离徙动变化量的影响。仿真结果和实际数据处理也证明当强散射体存在时使得运动误差估计有偏，必须采用合理方法抑制强目标散射体对相位估计影响。However, the KAJohnson weighting method also aggravates the contribution of strong target scatterers to the phase function while suppressing the influence of noise. When there is a strong target, ψ _p ^ldeal (t) in (4) can no longer be regarded as a uniformly distributed random variable in [-π, π], at this time it mainly appears as the distance migration phase of the strong target , so the delay variation estimated by (6) is affected by the distance migration variation. Simulation results and actual data processing also prove that the presence of strong scatterers makes the motion error estimation biased, and a reasonable method must be used to suppress the influence of strong target scatterers on phase estimation.

我们提出一种新的α加权加权函数，对应于(9)式中，We propose a new α-weighted weighting function corresponding to (9) where,

${ω ω}^{22} ((t t)) = = \frac{| | {r r}_{p p}^{* *} ((t t)) {r r}_{p p - - 11} ((t t)) | |}{{α α}^{44} + + {| | {r r}_{p p}^{* *} ((t t)) {r r}_{p p - - 11} ((t t)) | |}^{22}} - - - - - - ((1010))$

式中α参数为一可调整量，通过合理选择α值可折衷控制强目标散射体和噪声对误差相位估计的影响。从抑制强散射目标考虑，α取值越小越好，而从抑制噪声考虑，α取值越大越好；可结合实际的先验知识来优化α取值。The α parameter in the formula is an adjustable quantity, and the influence of strong target scatterers and noise on the error phase estimation can be compromised by choosing a reasonable value of α. From the consideration of suppressing strong scattering targets, the smaller the value of α, the better, and from the consideration of suppressing noise, the larger the value of α, the better; the value of α can be optimized by combining actual prior knowledge.

这种合成孔径声纳成像的α加权切变平均自聚焦处理方法如附图1所示。首先对回波数据进行复解调和匹配滤波，然后进行α加权切变平均处理得到运动误差时延变化量空间位置累积，并用运动误差补偿回波数据，最后应用图像重构算法进行合成孔径成像。The α-weighted shear-average self-focusing processing method of the synthetic aperture sonar imaging is shown in Fig. 1 . First, complex demodulation and matched filtering are performed on the echo data, and then the α-weighted shear average processing is performed to obtain the space position accumulation of the motion error time-delay variation, and the echo data is compensated by the motion error, and finally the image reconstruction algorithm is used for synthetic aperture imaging .

该自聚焦处理方法的试验结果：使用α加权切变平均自聚焦处理方法对实际合成孔径声纳湖试数据处理结果表示，自聚焦处理后T字型煤气瓶目标聚焦为三个亮点，而且自聚焦处理后图像的信混比明显增强。试验结果表明改进的α加权切变平均自聚焦算法能够较好估计运动误差，从而有效地抑制图像模糊。Experimental results of this self-focusing processing method: using the α-weighted shear-average self-focusing processing method to process the actual synthetic aperture sonar data, the results show that after the self-focusing processing, the target of the T-shaped gas cylinder is focused on three bright spots, and the self-focusing The signal-to-mixture ratio of the image after focus processing is significantly enhanced. Experimental results show that the improved α-weighted shear-averaged self-focusing algorithm can better estimate the motion error and effectively suppress image blur.

Claims

1, a kind of self-focusing processing method of synthetic aperture sonar imaging is characterized in that: mainly may further comprise the steps:

(1), each echoed signal obtains broadband analytic signal r through complex radical band demodulation and matched filter processing _p(t);

(2), relevant treatment between the adjacent echo analytic signal; The broadband analytic signal r of adjacent two transponder pulse echoes _p(t), r _P-1(t) computing cross-correlation is:

{rr}_{p} = {&Integral;}_{0}^{T} r_{p}^{*} (t) r_{p - 1} (t) dt - - - (5)

* represents complex conjugate in the formula;

(3), by asking plural phase angle, obtain the phase error estimation and phase error between the adjacent echo analytic signal:

Δ {\hat{φ}}_{p} = Arg {{rr}_{p}} - - - (6)

(4), obtain the accumulated phase error that kinematic error causes by the locus accumulated process:

{\hat{ϵ}}_{p} = {\hat{ϵ}}_{p - 1} + Δ {\hat{φ}}_{p};

{\hat{ϵ}}_{0} &equiv; 0 - - - (7)

(5), utilize the sequence of phase errors that obtains

Compensate and realize the echo data correction:

R_{p}^{Ideal} (f) = R_{p} (f) \exp (j {\hat{ϵ}}_{p}) - - - (8)

In the formula, R _p(f) be r _p(t) Fourier transform;

(6), compensation back data are used corresponding image reconstruction algorithm, the synthetic aperture sonar picture that then can obtain to focus on.

2, the self-focusing processing method of synthetic aperture sonar imaging according to claim 1 is characterized in that:

Adopt α weighting weighting function, in (9) formula,

rr = {\frac{{&Integral;}_{o}^{T} ω^{2} (t) r_{p}^{*} (t) r_{p - 1} (t) dt}{{&Integral;}_{o}^{T} ω^{2} (t) dt}} - - - (9)

ω^{2} (t) = \frac{| r_{p}^{*} (t) r_{p - 1} (t) |}{α^{4} + {| r_{p}^{*} (t) | r_{p - 1} (t)}^{2}} - - - (10)

But alpha parameter is an adjustment amount in the formula, controls strong target scattering body and noise to the error phase estimation effect by selecting the α value.

3, the self-focusing processing method of synthetic aperture sonar imaging according to claim 1 and 2 is characterized in that: suppose at acoustic transducer basic matrix generation stagger x _p, and the main body of scatterer is approximate constant, then the echo delay time error for each scatterer in the scene all equals ε _p≈ 2x _p/ c, p echo can be expressed as

r_{p} (t) = r_{p}^{Ideal} (t - ϵ_{p}) \exp (- j 2 π f_{0} ϵ_{p})

= | r_{p}^{Ideal} (t - ϵ_{p}) | \exp (j ψ_{p}^{Ideal} (t - ϵ_{p})) \exp (- j 2 π f_{0} ϵ_{p}) - - - (3)

Ψ _p ^IdealBe equally distributed stochastic variable in [π, π]; f ₀Be signal center frequency; When not having kinematic error, echoed signal obtains analytic signal r through complex radical band demodulation and matched filter processing _p(t), be expressed as r _p ^Ideal(t).