A SAR remote sensing detection performance evaluation method

Abstract

The present invention discloses a SAR remote sensing detection performance evaluation method, belonging to the field of remote sensing detection technology. The evaluation method divides factors affecting SAR detection performance into two categories: spatial performance influencing factors and radiation performance influencing factors. Spatial performance influencing factors are characterized using the Johnson criterion, resolution, and target characteristic size, while radiation performance influencing factors are characterized using backscatter coefficient difference, target scattering characteristics, and background scattering characteristics. Resolution-limited and backscatter-limited probability models are established, respectively, and the probabilities of discovery, identification, and confirmation that characterize SAR detection performance are further obtained, providing a technical basis for quantitative evaluation of SAR detection performance.

Description

Technical Field

[0001] The invention relates to a SAR remote sensing detection performance evaluation method, belonging to the technical field of remote sensing detection. Background Art

[0002] Synthetic Aperture Radar (SAR), a research hotspot in the field of international radar remote sensing, has always been difficult to evaluate. SAR detection performance is influenced by two factors: one is the SAR system itself, including the radar system, correction system, and imaging algorithm; the other is related to the ground object and its surrounding scene, including the target's geometry, motion speed, and dielectric constant. Unlike optical images, SAR images reflect the backscatter coefficient of electromagnetic waves from the object, and therefore have strong speckle noise, as well as unique quality issues such as ghosting, non-uniform gain, and motion blur.

[0003] Currently, SAR detection performance evaluation is primarily based on the quality of SAR images, with SAR detection performance reflected through images of varying quality. Currently, objective evaluation methods are used to assess SAR image quality. These methods construct mathematical models associated with image capabilities, analyze the images being evaluated, and generate evaluation results. These methods are convenient, fast, and accurate. SAR image quality involves many independent quality factors, such as resolution and noise level. Each quality factor can be quantitatively described by one or several numerical values. However, specific quality factors generally only reflect a specific aspect of a SAR image, making it difficult to comprehensively and directly reflect the SAR imaging capability for target detection, identification, and confirmation. Furthermore, this method is a post-hoc evaluation, requiring only SAR images as a basis, making it ineffective for effectively evaluating detection performance during the demonstration and design phases. Summary of the Invention

[0004] The technical problem solved by the present invention is: to overcome the shortcomings of the existing technology and propose a SAR remote sensing detection performance evaluation method. According to the backscatter coefficient difference and the corrected equivalent strip line pair number, a backscattering limited probability model and a resolution limited model are established respectively, and the backscattering limited probability and the resolution limited probability are further calculated to quantitatively evaluate the SAR's target discovery, identification, and confirmation detection capabilities.

[0005] The technical solution of the present invention is:

[0006] A SAR remote sensing detection performance evaluation method, comprising:

[0007] The detection index value is calculated based on the backscatter characteristics of the target and background and the signal-to-noise ratio within the radar synthetic aperture time, which is used to characterize the detection capability of SAR limited by backscatter.

[0008] Based on the detection index, the backscattering characteristic difference between the target and the background, a backscattering limited probability model is constructed, and the values ​​of the above two are substituted into the backscattering limited probability.

[0009] A resolution-limited probability model is established to calculate the equivalent size of the target. Based on the ground resolution of SAR remote sensing, the number of equivalent strip line pairs of the target is obtained. The number of equivalent strip line pairs required for the detection level of discovery, identification, and confirmation is then substituted into the resolution-limited probability model to obtain the resolution-limited probability at the detection level of discovery, identification, and confirmation.

[0010] The probabilities of target discovery, identification, and confirmation are calculated using the backscatter-limited probability and the resolution-limited probability at the discovery, identification, and confirmation detection levels, respectively.

[0011] Furthermore, based on the detection index and the backscattering characteristic difference between the target and the background, a backscattering limited probability model is constructed. The backscattering limited probability model is:

[0012]

[0013] Where, P σ is the backscattering limitation probability, MRΔσ is the detection index, Δσ is the backscattering characteristic difference between the target and the background, E σ =2.7+0.7(Δσ / MRΔσ).

[0014] Furthermore, the calculation method of the detection index MRΔσ is:

[0015]

[0016] Where, P t is the radar peak power, G t is the antenna gain, R is the distance from the point target to the radar, λ is the wavelength, l s is the system loss factor, k is the Boltzmann constant, T s is the system equivalent noise temperature, B s is the system receiving bandwidth, SNR DT The threshold signal-to-noise ratio for human eyes to distinguish targets, n r is the range pulse compression gain, n a is the azimuth pulse compression gain, and M is the number of views.

[0017] Furthermore, the method for obtaining the number of equivalent strip line pairs required for discovery, identification, and confirmation detection levels is as follows: based on the Johnson criterion, the number of equivalent strip line pairs required for discovery, identification, and confirmation detection levels is obtained, and then the above-mentioned equivalent strip line pairs are corrected to obtain the final number of equivalent strip line pairs required for discovery, identification, and confirmation detection levels for SAR remote sensing detection performance evaluation; wherein the correction factor is 2-4.

[0018] Furthermore, a resolution-limited probability model is established, which is:

[0019]

[0020] Where N is the number of target equivalent strip line pairs; n ε is the number of equivalent strip line pairs required at different detection levels, and the detection levels of discovery, identification, and confirmation are selected respectively; E ε =2.7+0.7(N / n ε );P J is the resolution-limited probability at the corresponding detection level.

[0021] Furthermore, the probability of target discovery, identification and confirmation is calculated respectively. The calculation method is: Backscattering limited probability P σ The resolution-limited probability P under the corresponding detection level J The product obtained by multiplying them is the probability of the target at the corresponding detection level.

[0022] Furthermore, the probability of target discovery, identification and confirmation is calculated respectively. The calculation method is:

[0023] P=(1-P σ )×(1-P J )

[0024] Where, P σ is the backscattering limited probability, P J is the resolution-limited probability at the corresponding detection level, and P is the probability of the target at the corresponding detection level.

[0025] The advantages of the present invention compared with the prior art are:

[0026] (1) The SAR detection performance evaluation method of the present invention takes into account both spatial performance factors and radiation performance factors, and uses resolution-limited probability and backscatter-limited probability to represent them respectively. Finally, the detection performance of SAR is evaluated by the joint probability of discovery, identification, and confirmation. Compared with the traditional evaluation method of SAR image quality indicators, the evaluation results are quantitative and intuitive, which can guide the design of SAR detection systems.

[0027] (2) The backscattering limited probability model proposed in the present invention can adaptively modify parameters such as the number of views and the threshold signal-to-noise ratio at which the human eye can distinguish targets according to the actual application scenario to adapt to different evaluation scenarios. It is more flexible in application and has strong operability.

[0028] (3) The traditional Johnson criterion is mainly used for interpretability evaluation of optical detection, rather than for application in the field of SAR detection. Based on the empirical relationship between optical detection capability and SAR detection capability, the present invention innovatively adopts a correction factor to correct the traditional Johnson criterion to meet the needs of SAR detection performance evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0030] Figure 1 This is a flow chart of the SAR remote sensing detection performance evaluation method of the present invention. DETAILED DESCRIPTION

[0031] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0032] The present invention proposes a SAR remote sensing detection performance evaluation method, which divides the factors affecting SAR detection performance into two categories, namely, spatial performance influencing factors and radiation performance influencing factors. The spatial performance influencing factors are characterized by the Johnson criterion, resolution and target characteristic size, and the radiation performance influencing factors are characterized by the minimum resolvable backscatter coefficient difference (MRΔσ), target scattering characteristics and background scattering characteristics. Resolution-limited and backscatter-limited probability models are established respectively, and the joint probability of discovery, recognition and confirmation that characterize SAR detection performance is further obtained, providing a technical basis for quantitative evaluation of SAR detection performance.

[0033] The present invention realizes the process as follows Figure 1 The specific steps are as follows:

[0034] Step 1: Assume that the backscattering coefficients of the target and background are σ t and σ b (m 2), are used to characterize the scattering characteristics of the target and background respectively, and the backscattering difference Δσ=|σ t -σ b |.

[0035] Step 2: The signal-to-noise ratio (SNR) within the synthetic aperture time is derived based on the radar equation. MRΔσ is then derived by combining it with the SNR threshold at which the human eye can distinguish targets. MRΔσ is used as a comprehensive indicator to characterize the detection capability of SAR limited by backscatter.

[0036]

[0037] P t is the radar peak power, G t is the antenna gain, R is the distance from the point target to the radar, λ is the wavelength, l s is the system loss factor (l s >1), k is the Boltzmann constant, T s is the system equivalent noise temperature, B s is the system receiving bandwidth, SNR DT The threshold signal-to-noise ratio for human eyes to distinguish targets, n r is the range pulse compression gain, n a is the azimuth pulse compression gain, and M is the number of views.

[0038] Step 3: To construct a backscatter-limited probability model and quantitatively evaluate the detection capability of SAR limited by backscatter, substitute MRΔσ and Δσ into the probability transfer function to obtain the backscatter-limited probability P σ :

[0039]

[0040] Among them, E σ =2.7+0.7(Δσ / MRΔσ).

[0041] Step 4: The detection capability limited by the resolution is related to the size of the target. The equivalent size is used to characterize the spatial characteristics of the target. Assuming that the target size is L (length) and W (width), the present invention adopts a two-dimensional representation method and uses the geometric mean of the target length and width as the equivalent size of the target (a one-dimensional representation method can also be used, that is, using length or width as the target equivalent size), that is, the equivalent size is (L×W) 1 / 2 , then the target equivalent strip line pair number N is:

[0042]

[0043] Among them, GSD is the ground resolution of SAR remote sensing.

[0044] Step 5: Since the traditional Johnson criterion is based on the number of equivalent strip line pairs n required for different detection levels proposed by optical images, the traditional Johnson criterion is modified for SAR detection applications. The modified equivalent strip line pair number n is ε =ε·n, where ε is a correction factor. Based on the experience of professional interpreters in interpreting optical and SAR images, in general, to achieve the same image interpretation effect, the resolution of optical images should be three times that of SAR images. This may vary in different situations. The present invention proposes that the optimal value range of ε is 2-4. The value can be selected according to the situation. When the SAR resolution is high and the target scattering is strong, a value less than or equal to 3 can be selected. Otherwise, a value greater than or equal to 3 can be selected.

[0045] Step 6: Construct a resolution-limited probability model to quantitatively evaluate the detection capability of SAR limited by resolution, and compare N with n ε Substituting into the probability transfer function, we get the resolution-limited probability P J :

[0046]

[0047] Among them, E ε =2.7+0.7(N / n ε ).

[0048] Step 7: Based on the backscattering limited probability P σ and resolution-limited probability P J , constructing a joint probability model, the present invention adopts the method of multiplying the two probabilities as the joint probability, that is, it is considered that the target can be detected only when both limiting factors meet the conditions. The probability of target discovery, recognition and confirmation is: P = P σ ×P J .

[0049] The following is further described by specific examples:

[0050] Taking the indicators of a hypothetical SAR satellite as an example, the detection performance of a certain type of tank is evaluated.

[0051] Step 1: Assume that the tank target is on wet soil, and the backscatter difference Δσ between the target and the background is shown in Table 1:

[0052] Table 1 Statistics of the difference Δσ between the backscatter of the target and the background

[0053] Azimuth (°) 0-20 20-40 40-60 60-80 80-100 100-120 120-140 140-160 160-180 <![CDATA[Median value (m 2 )]]> 2.88 0.58 0.47 0.87 24.90 0.63 0.14 0.19 3.59

[0054] Step 2: Assume that the SAR satellite indicator is: radar peak power P t 1200W, antenna gain G tis 30dB, the distance R from the target to the radar is 550km, the wavelength λ is 0.0306m, and the system loss factor is l s is 6dB, and the Boltzmann constant k is 1.38054×10 -23 J / K, system equivalent noise temperature T s The system receiving bandwidth is 290K. s 1×10 -9 Hz, the threshold signal-to-noise ratio (SNR) at which the human eye can distinguish targets DT is 2.25, and the range pulse compression gain n r 5×10 4 , azimuthal pulse compression gain n a 1.68×10 5 , the viewing number M is 1.

[0055] The calculated SAR MRΔσ is 0.6897m 2 .

[0056] Step 3: Substitute MRΔσ and Δσ into the probability transfer function to obtain the backscattering limited probability P σ , the results are shown in Table 2:

[0057] Table 2 Backscattering limited probability P σ Statistics

[0058] Azimuth (°) 0-20 20-40 40-60 60-80 80-100 100-120 120-140 140-160 160-180 <![CDATA[P σ ]]> 1.00 0.36 0.23 0.70 1.00 0.42 0.01 0.02 1.00

[0059] Step 4: Assuming the target size L is 9.83m and W is 3.66m, the equivalent size of the target is (L×W) 1 / 2 The target equivalent strip line pair number N is 6.

[0060] Step 5: The traditional Johnson criterion finds, identifies, and confirms the number of equivalent strip line pairs n required for the detection level to be 0.75, 3, and 6 respectively. Assuming the correction factor ε is 3, the corrected number of equivalent strip line pairs n ε They are 2.25, 9, and 18 respectively.

[0061] Step 6: N and n ε Substituting into the probability transfer function, we get the resolution-limited probability P J , see Table 3.

[0062] Table 3 Resolution limited probability P J Statistics

[0063]

[0064] Step 7: Based on the backscattering limited probability P σ and resolution-limited probability P J, the target discovery, recognition and confirmation probabilities are shown in Table 4.

[0065] Table 4 Target discovery, identification and confirmation probability statistics

[0066]

[0067] Alternatives:

[0068] (1) The backscatter difference Δσ between the target and the background in step 1 can be real collected data or obtained by simulation.

[0069] (2) The equivalent size in step 4 can also be represented in one dimension, that is, length or width is used as the target equivalent size.

[0070] (3) The correction factor ε in step 5 is generally 2-4 and can be selected according to the actual detection environment.

[0071] (4) Step 7 can also be done by using P=(1-P σ )×(1-P J ) is taken as the joint probability. At this time, it is considered that the target can be detected if one of the two limiting factors meets the condition.

[0072] The above-described embodiments are only preferred specific implementations of the present invention. Common changes and substitutions made by those skilled in the art within the scope of the technical solution of the present invention should be included in the protection scope of the present invention.

Claims

1. A SAR remote sensing detection performance evaluation method, characterized in that: include: The detection index value is calculated based on the backscatter characteristics of the target and background and the signal-to-noise ratio within the radar synthetic aperture time, which is used to characterize the detection capability of SAR limited by backscatter. Based on the detection index, the backscattering characteristic difference between the target and the background, a backscattering limited probability model is constructed, and the values ​​of the above two are substituted into the backscattering limited probability. A resolution-limited probability model is established to calculate the equivalent size of the target. Based on the ground resolution of SAR remote sensing, the number of equivalent strip line pairs of the target is obtained. The number of equivalent strip line pairs required for the detection level of discovery, identification, and confirmation is then substituted into the resolution-limited probability model to obtain the resolution-limited probability at the detection level of discovery, identification, and confirmation. The probabilities of target discovery, identification, and confirmation are calculated using the backscatter-limited probability and the resolution-limited probability at the discovery, identification, and confirmation detection levels, respectively.

2. The SAR remote sensing detection performance evaluation method according to claim 1, characterized in that: Based on the detection index, the backscattering characteristic difference between the target and the background, a backscattering limited probability model is constructed. The backscattering limited probability model is: Where, P σ is the backscattering limitation probability, MRΔσ is the detection index, Δσ is the backscattering characteristic difference between the target and the background, E σ =2.7+0.7(Δσ / MRΔσ).

3. A SAR remote sensing detection performance evaluation method according to claim 1 or 2, characterized in that: The calculation method of the detection index MRΔσ is: Where, P t is the radar peak power, G t is the antenna gain, R is the distance from the point target to the radar, λ is the wavelength, l s is the system loss factor, k is the Boltzmann constant, T s is the system equivalent noise temperature, B s is the system receiving bandwidth, SNR DT The threshold signal-to-noise ratio for human eyes to distinguish targets, n r is the range pulse compression gain, n a is the azimuth pulse compression gain, and M is the number of views.

4. The SAR remote sensing detection performance evaluation method according to claim 1, characterized in that: According to the equivalent size of the target and the ground resolution of SAR remote sensing, the number of equivalent strip line pairs of the target is obtained. The calculation method is: Where N is the number of target equivalent strip line pairs, GSD is the ground resolution of SAR remote sensing, and the target equivalent size is S.

5. The SAR remote sensing detection performance evaluation method according to claim 4, characterized in that: The target equivalent size S is (L×W) 1 / 2 , the target length is L and the width is W.

6. A SAR remote sensing detection performance evaluation method according to claim 4, characterized in that: The target equivalent size S is the target length or width.

7. The SAR remote sensing detection performance evaluation method according to claim 1, characterized in that: The method for obtaining the number of equivalent strip line pairs required for the detection level of discovery, identification, and confirmation is as follows: based on the Johnson criterion, the number of equivalent strip line pairs required for the detection level of discovery, identification, and confirmation is obtained, and then the above equivalent strip line pairs are corrected to obtain the final equivalent strip line pairs required for the detection level of discovery, identification, and confirmation for SAR remote sensing detection performance evaluation; wherein the correction factor is 2-4.

8. The SAR remote sensing detection performance evaluation method according to claim 1, characterized in that: A resolution-limited probability model is established. The resolution-limited probability model is: Where N is the number of target equivalent strip line pairs; n ε The number of equivalent strip line pairs required at different detection levels is selected for the discovery, identification, and confirmation detection levels respectively; E ε =2.7+0.7(N / n ε );P J is the resolution-limited probability at the corresponding detection level.

9. The SAR remote sensing detection performance evaluation method according to claim 1, characterized in that: Calculate the probability of target discovery, identification and confirmation respectively. The calculation method is: backscatter limited probability P σ The resolution-limited probability P under the corresponding detection level J The product obtained by multiplying them is the probability of the target at the corresponding detection level.

10. The SAR remote sensing detection performance evaluation method according to claim 1, characterized in that: Calculate the probability of target discovery, identification and confirmation respectively. The calculation method is: P=(1-P σ )×(1-P J ) Where, P σ is the backscattering limited probability, P J is the resolution-limited probability at the corresponding detection level, and P is the probability of the target at the corresponding detection level.

Images