CN113064167A

CN113064167A - SAR imaging scene division method based on interest region detection

Info

Publication number: CN113064167A
Application number: CN202110285633.1A
Authority: CN
Inventors: 皮亦鸣; 李晋; 闵锐; 王旭; 杨晓波; 曹宗杰; 崔宗勇
Original assignee: University of Electronic Science and Technology of China
Current assignee: University of Electronic Science and Technology of China
Priority date: 2021-03-17
Filing date: 2021-03-17
Publication date: 2021-07-02
Anticipated expiration: 2041-03-17
Also published as: CN113064167B

Abstract

The invention belongs to the technical field of radar imaging, and particularly relates to an SAR imaging scene division method based on interest area detection. According to the invention, by analyzing the phase error after the preprocessing of the synthetic aperture radar echo data and further optimizing the sub-scene division method on the existing phase gradient sub-scene division method, the focusing on the region of interest and the inhibition of edge defocusing can be well performed, and the imaging quality is improved. In addition, the invention does not relate to complex calculation operation and has simpler implementation mode.

Description

SAR imaging scene division method based on interest region detection

Technical Field

The invention belongs to the technical field of radar imaging, and particularly relates to an SAR imaging scene division method based on interest area detection.

Background

Synthetic Aperture Radar (SAR) accumulates doppler shift of echo signals during Radar motion by using correlation of the Radar echo signals, synthesizes equivalent Radar Aperture in the Radar motion direction, and realizes high-resolution imaging. The all-weather area monitoring imaging can be realized. Motion is the basis for SAR and is also the source of problems. Motion errors are the main source of phase errors, primary phase errors only affect the position of signals, secondary phase errors can defocus images, imaging quality is affected, and higher-order phase errors can generate false alarms.

The phase error due to motion error is generally two-dimensional space-variant and coupled, and the solution for the space-variant error is intuitive and simple, i.e., the imaging scene is divided into a plurality of sub-scenes and compensated for respectively. Limited by the complex phase error form, no quantitative partition criterion is given as to how to reasonably partition the sub-scenes. At present, the sub-scenes are divided into two main ways, one sub-scene dividing method is uniform division, because the phase error is symmetrical about the imaging scene center, the first quadrant of the imaging scene is uniformly divided into n multiplied by n blocks for block correction, and then the corrected sub-scenes are spliced; another method for dividing sub-scenes is based on phase error gradient division, and generally, when the phase error is less than pi/4, the influence of the phase error on the imaging quality is negligible, so that the sub-scenes can be divided by taking pi/4 as a gradient. However, when a large scene is imaged, the phase error between the scene center and the scene edge is large, if the scene is uniformly divided, the imaged scene edge still generates a relatively serious defocus phenomenon, and if the scene is divided according to the phase error gradient, a relatively accurate imaging result can be obtained, but the calculation amount is huge. Therefore, it is very important to find a sub-scene division method for dividing as few sub-scenes as possible while ensuring the focusing effect.

Disclosure of Invention

The invention aims to provide an SAR imaging scene division method based on interest area detection, which pre-divides sub-scenes through phase error gradient in the synthetic aperture radar imaging process, searches according to the energy density of each sub-scene and the average distance to the scene center, optimizes the division of the sub-scenes and respectively performs phase compensation, effectively improves the imaging quality of the scene edge, and improves the operation speed while ensuring the focusing effect. The method is an efficient scene division imaging method for the synthetic aperture radar.

The technical scheme of the invention is as follows: an SAR imaging scene division method based on interest region detection comprises the following steps:

step 1, preprocessing SAR original echo signal s (tau, t), namely, preprocessing SAR original echo signal s (tau, t) and reference signal s_ref(tau, t) is mixed to obtain a difference frequency signal s_if(τ, t), for s_ifFourier transformation is carried out on the fast time tau of (tau, t) to a frequency domain to obtain a difference frequency domain signal, and then the difference frequency domain signal and a corresponding compensation function S are carried out_c(f_i) Multiplication to obtain S_if(f_iT) to remove the residual video phase term and the envelope skew term, and then to perform inverse fourier transform to the time domain, where the target echoes at different distances are aligned in fast time.

Step 2, according to a phase gradient error division method, carrying out scene division on an imaging scene by taking a phase error of every pi/4 from the center phase of the scene as a gradient to obtain the number N of sub-scenes and energy ratio factors of the sub-scenes which are divided by the phase error

Average distance to scene center R_jJ is 1,2, …, N. Wherein, by the pair s_if(f_iT) performing a two-dimensional Fourier transform to obtain ρ_iFor detecting a Region of interest (ROI), which facilitates focusing on the ROI, the distance R from each sub-scene to the center of the scene is obtained by the ratio of the area of the sub-scene to the double perimeter_jThe method is convenient for focusing the scene edge and inhibiting the phenomenon of edge defocusing.

Step 3, presetting an optimized sub-scene number N which is sufficiently less than N according to

Search m_kK is 1,2, …, N, i.e. a mapping m for the number N of pre-divided sub-scenes to the number N of sub-scenes after optimization is obtained_k。

Step 4, passing through m_kObtaining the sub-scene division after combination, and then carrying out the echo data on each sub-scene frequency domain and the corresponding phase compensation function phi_kMultiplying the two-dimensional Fourier inverse transformation to obtain sub-scene scattering information delta_kAnd splicing to obtain imaging scene scattering information delta.

The invention determines the pre-division of the sub-scenes by a phase gradient error division method, and adaptively searches the optimal sub-scene combination and optimization division by calculating the energy ratio factor of each sub-scene and the average distance to the scene center, thereby ensuring the focusing effect and greatly improving the operation speed.

The method has the advantages of improving the operation speed of scene division, inhibiting the phenomenon of scene edge defocusing, optimizing the focusing effect on the ROI, improving the imaging quality, and not involving complex operation in the implementation process.

Drawings

FIG. 1 is a flow chart of the present invention;

FIG. 2 is a phase error gradient plot;

FIG. 3 is an optimized adaptive sub-scene partition diagram;

fig. 4 is the imaging results of 3 x 3 uniform division;

FIG. 5 is the scene edge portion of FIG. 4;

FIG. 6 is an imaging result of a gradient-binned scene with π/4 as phase error;

FIG. 7 is the scene edge portion of FIG. 6;

FIG. 8 is an imaging result after optimizing sub-scene partitioning;

fig. 9 is a scene edge portion of fig. 8.

Detailed Description

The invention is described in detail below with reference to the drawings and simulation examples to prove the applicability of the invention.

As shown in fig. 1, the method for dividing the SAR imaging scene based on the region of interest detection according to the present invention can effectively suppress the scene edge defocusing phenomenon after the input SAR original echo is subjected to imaging processing, and the specific implementation steps are as follows:

step 1: the distance direction preprocessing is carried out on the original echo: for received signal s (tau, t) and reference signal s_ref(tau, t) mixing to obtain difference frequency signal s_if(τ, t), expressed as:

wherein R is_a(t) is a reference pitch, Δ R (t) R_t(t)-R_ref(t)，T_pFor the signal pulse width, τ is the fast time of the radar signal, t is the slow time of the radar signal, c is the speed of light, α is the chirp rate of the radar signal, f_cTo the centre frequency, σ, of the signal_tIs the scattering intensity of the target;

then, for the difference frequency signal s_if(tau, t) deskewing, i.e. deskewing the difference frequency time domain signal s_if(tau, t) Fourier transforming the fast time tau to obtain a difference frequency domain signal, and corresponding phase compensation function S_c(f_i) Multiplying to obtain a deskewed difference frequency domain signal S_if(f_iT), the expression of which is:

step 2: performing sub-scene pre-division on the imaging scene by using a phase gradient error division method and taking a phase error of every pi/4 from the central phase of the scene as a gradient to obtain effective parameters of the pre-divided sub-scenes, wherein the effective parameters comprise the number N of the sub-scenes and energy ratio factors of the sub-scenes

Average distance to scene center R_jJ is 1,2, …, N. Wherein byThe ratio of the energy of the sub-scene to the scene energy yields ρ_jFacilitating focusing on the ROI; obtaining the distance R from each sub-scene to the center of the scene through the ratio of the area of the sub-scene to the double perimeter_jThe method is convenient for focusing the scene edge and inhibiting the phenomenon of edge defocusing.

And step 3: and dividing the SAR imaging scene based on the interest region detection. Presetting an optimized post-molecular scene number N which is sufficiently smaller than the preplanned molecular scene number N, and searching the preplanned molecular scene combination method m in a stepping mode_kNamely, determining the mapping relation from the pre-divided sub-scene number N to the optimized sub-scene number N.

Mapping m obtained by the above search scheme_kAn optimized sub-scene partition can be obtained.

And 4, step 4: for echo data on each sub scene frequency domain and corresponding phase compensation function phi_kMultiplying the two-dimensional Fourier inverse transformation to obtain sub-scene scattering information delta_kAnd splicing to obtain imaging scene scattering information delta.

Simulation example

According to the method, namely the flow operation shown in the attached figure 1, the specific setting of the simulation parameters is as follows: adopting an imaging mode of a circumferential SAR, wherein the distance from a radar to the center of a scene is 100 meters, the imaging radius is 80 meters, and the central carrier frequency of a transmitted signal is 220 GHZ;

fig. 2 shows a phase error gradient map obtained by simulation, where 10 is a step value, and a pre-division sub-scene diagram is drawn, where N is 400;

fig. 3 shows an optimized adaptive sub-scene division diagram, where n is preset to 20, and 2 is used as a step value to draw an optimized sub-scene division schematic;

fig. 4 shows the imaging result obtained by dividing a 3 × 3 uniform scene, and it can be clearly seen that a defocusing phenomenon exists at the edge of the scene, and the imaging time is only 19.08 seconds;

FIG. 5 shows the results of the imaging of the scene edges of FIG. 4;

FIG. 6 shows the pre-divided imaging results, with each point target having the best focus, but with an imaging time of 283.57 seconds;

FIG. 7 shows the scene edge imaging results of FIG. 6, and it can be seen that the edge defocus phenomenon is significantly suppressed;

FIG. 8 shows the imaging result of the method, which shows that the defocus phenomenon at the scene edge is suppressed, and the imaging time is 43.14 seconds;

fig. 9 shows the imaging result of the scene edge of fig. 8, compared with fig. 7, the side lobe suppression effect of the edge part is weakened, but still greatly improved compared with fig. 5, and the imaging efficiency of the method is significantly improved compared with the phase gradient error division.

In conclusion, the method provided by the invention has practical value in view of processing results.

Claims

1. An SAR imaging scene division method based on interest region detection is characterized by comprising the following steps:

s1, preprocessing SAR original echo signals S (tau, t) to align target echoes at different distances in a fast time;

s2, according to the phase gradient error division method, dividing the imaging scene by taking the phase error of every pi/4 from the central phase of the scene as the gradient, wherein the obtained number of sub-scenes is N, and the ratio of the energy of the sub-scenes to the energy of the scene is rho_jAverage distance to the scene center is R_jJ is 1,2, …, N, where ρ_jFor detecting the interested region, the distance R from each sub-scene to the scene center is obtained by the ratio of the area of the sub-scene to the double perimeter_j；

S3, presetting an optimized sub-scene number N which is sufficiently less than N according to

Search m_kObtaining the number N of sub-scenes to the number N of sub-scenes after optimizationMapping relation m_k；

S4, passing through m_kObtaining optimized sub-scene division, and performing phase compensation function phi on echo data on each sub-scene frequency domain_kMultiplying the two-dimensional Fourier inverse transformation to obtain sub-scene scattering information delta_kAnd splicing to obtain imaging scene scattering information delta.