UAV-Mounted GPR for Object Detection Based on Cross-Correlation Background Subtraction Method

Round 1

Reviewer 1 Report (Previous Reviewer 2)

I would like to thank the editor to give me the opportunity to review this interesting work. The impression of the paper is interesting. I have a minor comment about the future studies, which will guide the researchers to continue this good work.

The article is well written and easy to understand. However, few of my feedback can be considered to improve the quality of the paper.

1.     Introduction may be improved, adding the highlights and the problem statements.

2.     You could improve writing, link better the ideas flow in the Introduction.

3.     Review references because some of them are unstandardized.

4.    The conclusion needs improvements towards major claimed contribution.

5.    Write some future directions in the conclusion section.

6.     The difference between your proposal and related works is not clear, you could to details better. I suggest add a comparative table in ''Related Literature'' to contrast your solution in front of related works.

7.     You could discuss the relationship between your solution and past literature.

Author Response

Q1)

Introduction may be improved, adding the highlights and the problem statements.

A1)

Thank you for the valuable suggestion. The highlights and the problem statements have been added in Introduction section.

Q2)

 You could improve writing, link better the ideas flow in the Introduction.

A2)

Thank you for the valuable suggestion. We have improved the writing of Introduction to link the ideas flows in the revised manuscript.

Q3)

Review references because some of them are unstandardized.

A3)

Thank you for the valuable suggestion. The references have been standardized in the revised manuscript.                                                             

Q4)

The conclusion needs improvements towards major claimed contribution.

A4)

Thank you for the valuable suggestion. The conclusion has been modified in the revised manuscript to highlight the major contribution.

Q5)

Write some future directions in the conclusion section.

A5)

Thank you for the valuable suggestion. Some future directions are added into the conclusion section, as shown follows.

“However, the research of UAV-GPR still faces many challenges, such as environmental interference, signal attenuation, miniaturization and security, which will be our next research interest.”

Q6)

The difference between your proposal and related works is not clear, you could to details better. I suggest add a comparative table in ''Related Literature'' to contrast your solution in front of related works.

A6)

Thank you for the suggestion. A performance comparative table between our proposed method and related works is added in Simulations and Experiments section, as shown below.

Q7)

You could discuss the relationship between your solution and past literature.

A7)

Thank you for the suggestion. The relationship of our solution and past literature is added in the revised manuscript.

“The structural similarity (SSIM) is adopted to measure the relationship between the processed images and the original image. It uses three indicators of brightness, contrast and structure to evaluate the similarity between the two images. According to Fig. 10, it can be seen that different algorithms lead to the transformation of underground structures, which is difficult for the eyes to distinguish. However, the similarity between them can be calculated through SSIM (range 0-1). The higher the similarity, the closer it is. We calculated the similarity between Fig. 16 (b), (c), (d) with (a). To avoid the influence of surface direct waves, only the underground part is calculated (the image below the surface direct wave is taken for calculation). The calculation results are shown in Table I.

It can be seen from the calculation results that the CCBS method is the closest to the original image underground structure, which indicates that the algorithm can well preserve the shape and intensity of the underground structure. Although the mean subtraction method can retain good intensity, the average processing between adjacent echoes will change the structure. Besides, the moving average method will change the intensity of the echo.”

Author Response File: Author Response.docx

Reviewer 2 Report (Previous Reviewer 1)

The authors have answered all remarks and recommendation of the reviewer. The article can be accepted for publication.

Author Response

Thank you very much.

Reviewer 3 Report (Previous Reviewer 3)

This manuscript has a 99% similarity to the previously submitted version.

A text portion of the introduction has been added, as well as a discussion.

The manuscript still has conceptual problems that the previous version had, so I will not refer to them again.

The discussion is poor, as are the conclusions.

Processing has design issues

The manuscript is not publishable, so it needs to be restructured properly.

Author Response

Q1)

This manuscript has a 99% similarity to the previously submitted version. A text portion of the introduction has been added, as well as a discussion. The manuscript still has conceptual problems that the previous version had, so I will not refer to them again. The discussion is poor, as are the conclusions. Processing has design issues. The manuscript is not publishable, so it needs to be restructured properly.

A1)

Thank you for the comment, and your suggestions can greatly help improve our manuscript. For the antenna spectrum, since it was measured in a complex environment, it may not be ideal. We will further optimize the test environment in our next work. And the discussion and conclusion part have been modified in the revised manuscript, as shown below.

Discussion part：

“Currently, the UAV-GPR is facing many challenges, so it is not used commercially. For the areas that cannot be reached by humans, the UAV-GPR can replace humans to work and detect underground structures, such as mines, marshes, underground cavities and so on. This paper explores an integrated UAV-GPR and conducts the experimental verifications, but there are still many problems needed to be improved in the future.

1) Since the UAV-GPR system is working at low frequencies, the size of the antenna is very large. To improve the beam energy of antenna, we use an ultra-wideband antenna with multiple transmitting antennas and multiple receiving antennas. The size and weight of the radar can be further reduced by using an integrated antenna module, thus improving the working efficiency of the UAV-GPR system.

2) In this paper, we propose a processing algorithm for UAV-GPR, which aims to better remove interference signals and improve image resolution. This algorithm provides an effective solution to the interference caused by multi rotor UAV. In addition, there are some distance between the radar and the ground, and the signal will attenuate before reaching the ground surface. By compensating this attenuation, the radar image quality can be further improved. For more complex terrain, such as rugged road surface, the accuracy of the algorithm needs to be verified further.

3) Three experiments are designed in this paper to study different characteristics of UAV-GPR. Experiments are carried out on static UAV rotor and motor interference, floor penetrability, and bridge UAV-GPR imaging respectively. However, due to the complexity of the environment and underground structure, some areas in the images are difficult to explain. It is necessary to further analyze the characteristics of different detection targets through some comparative experiments, such as steel bars, concrete and water surface.”

Conclusion part：

“In this paper, we have developed a UAV-GPR system with miniaturization and penetrability, this system has the advantages of independent detection and data processing integration. Based on the system, a new cross correlation-based background subtraction (CCBS) method is proposed, and interference suppression technique is adopted in the signal processing step. The proposed CCBS method takes advantage of the similarity between each A-Scan echo and a reference wave by establishing a background-removal model. Moreover, we have explored the effects about the electromagnetic interference in environment and the interference from UAV rotor itself by practical testing, and a Lateral-Doppler-Filtering technique are adopted to effectively suppress the active interferences. Besides, a calibration method for measured data is introduced to calculate the dielectric constant of the measured medium. As a result, the simulations and experimental results indicate that the proposed method in terms of clutter subtraction has better performance than the traditional methods, and the penetrability of the UAV-GPR system is also validated. In this paper, experiments carried out in different scenarios provide meaningful reference for future research. However, the research of UAV-GPR still faces many challenges, such as environmental interference, signal attenuation, miniaturization and security, which will be our next research interest.”

Author Response File: Author Response.docx

Reviewer 4 Report (New Reviewer)

In this paper, the authors discuss a UAV-mounted GPR system with the frequency band at 150 MHz - 309 MHz based on a cross correlation-based background subtraction (CCBS) method and and an interference suppression technique. In general, the method is easily understood. However, more discussions should be added.

1. In section 1, the literature review should be enhanced. The authors should comprehensively review the advantages and disadvantages of related methods. This can help authors understand the novelty of authors’ method.

2. In Eq. (3), the noise is is considered. Is this noise subjected to Gaussian noise? Up till now, non-Gaussian noise like Middleton noise model [C1] and alpha distribution [C2] is characterized by heavy tail. The object detection methods with heavy tail noise are widely studied by researchers. The reviewer wanders to know whether the authors’ method can work well with these non-Gaussian noise models. The authors should discuss this issue in their paper.

[C1]Zhang，et al．Parameter estimation of underwater impulsive noise with the Class B model．IET Radar, Sonar & Navigation，2020，Doi: 10.1049/iet-rsn.2019.0477．

[C2]Mahmood, et al, “Modeling Colored Impulsive Noise by Markov Chains and Alpha-Stable Processes,” in OCEANS 2015 MTS/IEEE, (Genoa, Italy), Doi: 10.1109/OCEANS-Genova.2015.7271550.

3. In Eq. (6), the reviewer wanders to know why the parameter σ is set to 0.5. The authors should discuss the influence of this parameter on the performance of their method in detail.

4. Fig. 12 shows the processing results of authors’ method. To further evaluate the performance, the authors are suggested to use the quality parameters. With this operation, the conclusions based on this result would be much more convincing.

5. The authors should present the results of each step corresponding to Fig.7, as the authors’ method includes many key steps like CCBS and interference suppression technique. Based on this operation, the readers can easily understand that the authors’ method is effective.

Author Response

The response is listed in the attached file.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report (Previous Reviewer 3)

Thank you for the improvement, namely for include the discussion section. About the discussion, this section still have some issues. It is very summarized. A discussion section serves for discuss the problems identified and relate with previous publications regarding the same or similar subject. Therefore, a final effort must be made to improve the discussion and turn the manuscript 100% with an article structure.

The figure captions must have a discription of the image, the most important aspects.

Author Response

Thank you for the valuable comments. The details of revision are listed in the attached file.

Author Response File: Author Response.docx

Reviewer 4 Report (New Reviewer)

The authors well addressed my comments.

Author Response

Thank you very much.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

General description

In the present work, an UAV-GPR system with the frequency bandwidth from 150 MHz to 309 MHz, and an algorithm are proposed to process the raw radar echo data. The algorithm consists of four processing parts: a cross correlation-based background subtraction (CCBS) method to remove the background signal; a vertical filter to get rid of the active electromagnetic interference in the environment; a Lataral-Dopple- Filtering (LDF) to weaken the impact of passive interference of the UAV rotor electrical current; estimating the dielectric constant of the detected soil through the calibration of measured radar data. The authors provide numerical simulations and experimental results obtained by an UAV-GPR system that demonstrate better detection capability of the proposed algorithm and great potential in detecting underground buried targets.

Remarks:

Page 2, row 47: It is written: the dielectric constant of detected soil are estimated through the calibration. Correct: the dielectric constant of detected soil is estimated through the calibration.

Page 3, bottom: It is written Then, the second-order differential of is calculated to get the second-order Gaussian pulse as follows. Correct: Then, the second-order differential of the Gaussian pulse is calculated to get the second-order Gaussian pulse as follows.

Page 8, formula 5: X[t, u] is a cross correlation one-dimensional function. It is not clear the second argument u tahat can be seen in formulas (6) and (7).

There are several places where is written Fig. ?? (a), Fig. ?? (b) .

It is recommended to correct grammar mistakes in the paper.

Reviewer 2 Report

I would like to thank the editor to give me the opportunity to review this interesting work. The impression of the paper is interesting. I have a minor comment about the future studies, which will guide the researchers to continue this good work. 

The article is well written and easy to understand. However, few of my feedback can be considered to improve the quality of the paper.

[A] You could improve writing, link better the ideas flow in the Introduction.

[B] Introduction may be improved, adding the highlights and the problem statements. Review references because some of them are unstandardized.

[C] The conclusion needs improvements towards major claimed contribution.

[D] Write some future directions in the conclusion section.

[E] The difference between your proposal and related works is not clear, you could to details better. I suggest add a comparative table in ''Related Literature'' to contrast your solution in front of related works.

[F] You could discuss the relationship between your solution and past literature.

Reviewer 3 Report

The article proposal have several conceptual problems, since the equipment assembly (antenna spectral content problems).

The results also have problems, in the processing. The acquisition is made in the air, so the contact between air and ground have a contrast in the dieletric constant that prevent the transmission of the wave to the ground. The processing applied remove the "direct wave" of the interface air-ground, but the remain signal stil have problems in the amplitude. You need to apply more operations to improve the data to allow to see reflections of the ground.

THe article proposal don't have a discussion and the conclusions are very poor.

You can improve the manuscript to increase the quality to turn the document good to have the name "article".

Good work.