A Modular Method for GPR Hyperbolic Feature Detection and Quantitative Parameter Inversion of Underground Pipelines

Round 1

Reviewer 1 Report

GENERAL OVERVIEW

The paper describes the study on the characterization of underground pipelines using a GPR method. The authors develop a two-stage algorithm for modular detection and quantitative inversion for hyperbolic features that are commonly present in GPR images. The results of FDTD simulations are presented and processed. New YOLOv7 network is introduced for processing of GPR data. By the comparison to previously developed YOLOv5 network, some important advantages of newly-developed algorithm are described.

The discussed problem is of significant interest of NDT researchers and it meets the scope of the journal. The results has the potential to be practically used in non-destructive testing of real underground structures. The manuscript is well written in general, the description of the research performed is appropriate. No substantive flaws are present in the paper. In my opinion, the paper can be accepted for publication in Remote Sensing in a current version. A few comments are presented below.

 

CRITICAL REMARKS

1) Structure. The structure is logical, the following sections are organised reasonably.

2) Formatting. The Journal’s template is followed appropriately. The manuscript is well written. Some minor flaws are present:

– the variables in the text are sometimes written without using italic, e.g., line 145, where variables x and z should be written in italic,

– in the Introduction Section – there is a lack of spaces between sentences and references,

3) Figures. The figures have a good quality, they are easy to read.

4) Language. The quality of the language used is sufficient. No important language errors were identified, besides some minor flaws (typos) that can be easily detected and corrected in proofreading process.

Author Response

Thank you very much for your kindly comments. We have made corrections carefully according to your comments. Bellows are the point-to-point response.

Comment 1: The variables in the text are sometimes written without using italic, e.g., line 145, where variables x and z should be written in italic.

Response: Thank you so much for your careful check, variables in this paper have been modified to italic.

 

Comment 2: In the Introduction Section – there is a lack of spaces between sentences and references.

Response: We are very sorry for our negligence, spaces between sentences and references have been added.

Reviewer 2 Report

1.     The performance of YOLOv7 algorithm for automatically detecting hyperbolas in GPR images should be evaluated. Authors are required to provide false positives, precision, recall, etc., which are common in target recognition by machine learning. Without this evaluation, the detection method proposed in this manuscript cannot be considered successful.

2.     Authors mention that this manuscript proposed "a modular detection method". The "modular" here is not explicitly explained in the manuscript.

3.     The authors mention “the radius determines only the expansion angle of the hyperbola”(line 156 in Page 6). This description is incorrect. It is the velocity of electromagnetic waves in the medium that determines the expansion angle of the hyperbola, rather than the radius of the pipe. Therefore, please explain in detail the inversion method for pipe radius to demonstrate that the theory is correct.

4.     In section 4.2, what sort of points can be recognized as key points?

5.     A Colorbar needs to be added to Figure 9 to display the dielectric constant distribution of the medium.

6.     There are writing errors like ”Tuntil thehe relative” in line 185 page 6.

7.     It should be “Figure 11” instead of “Figure 12” in line 254 page 9.

8.     Figure 11 has three sub-figures. Please indicate a, b, and c and add the figure caption.

Author Response

Thank you very much for your kindly comments. We have made corrections carefully according to your comments. The point-to-point response is uploaded as an attached file. 

Sincerely

Chengke Zhu and Hongxia Ye

Fudan University, Shanghai China

Author Response File: Author Response.docx

Reviewer 3 Report

Comments for author File: Comments.pdf

Author Response

Dear Reviewer, 

Thank you very much for your kindly comments. We have made corrections carefully according to your comments. The point-to-point response to your comments is uploaded as an attached file. Please kindly considered it for review again. Thanks a lot!

Sincerely

Chengke Zhu and Hongxia Ye

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

I don't have a new comments.