Road Information Extraction from High-Resolution Remote Sensing Images Based on Road Reconstruction

Round 1

Reviewer 1 Report

The article can not be published as it is. There are a lot of grammatical errors and typos. Even the theoretical concepts are confusing. There is much work to do until this article can be published. Below, I provide a set of comments and suggestions to improve the article in order it can be published in Remote Sensing. Other comments and questions are in the article in pdf.

1. Please, revise the English grammar since there are many typos and poorly constructed sentences.

2. This article is about a problem that has been studied for too long, so there is a vast literature regarding it. That’s why I consider the Introduction is too short. Accordingly, I recommend to extend this section providing a more wide explanation of the state of the art in this topic.

3. Figure 1 resumes the method proposed by the authors, so it should be written in a section called Methodology.

4. The FMM method is poorly explained, with important errors that are not admissible in a scientific article (see comments in the pdf).

5. I think it would be better to use a percentile, such as the 90%, to estimate the road width, especially if there are many shadows or objects that reduce it in the image.

6. In general, I find very difficult to understand Section 2.2 in terms of image processing.

The same problem is present in Sections 2.3 and 2.4. I recommend rewriting it to explain the concepts taking into account the potential readers of the article and that you are working with images. Moreover, I do not understand why it is necessary all this. If you have the boundary of the road, calculate a distance map M_D, then the contour line with maximum value or M_D will follow the centerline, or not?

Comments for author File: Comments.pdf

Author Response

Dear reviewer,

Thank you very much for your constructive comments to improve our manuscript. We really appreciate your precious time and great efforts on reviewing the manuscript. We have carefully addressed all the review comments and improved the quality of this manuscript accordingly. Please find below our detailed response to the comments point by point.

Point 1: Please, revise the English grammar since there are many typos and poorly constructed sentences. 


 

Response 1: We are so sorry for our unprofessional English writing. Thank you very much for your suggestions, the questions you have raised are very pertinent, and our level of professional English does need to be further improved. After carefully check, we found many grammar and sentence errors, and have modified the manuscript accordingly. Furthermore, we have invited several English teachers help correct grammar and sentences, and we hope the revised paper will be clearer and more accurate on expressions. If there are any other mistakes, please inform us timely for us to revise.

 

Point 2: This article is about a problem that has been studied for too long, so there is a vast literature regarding it. That’s why I consider the Introduction is too short. Accordingly, I recommend to extend this section providing a more wide explanation of the state of the art in this topic. 


 

Response 2: Thank you for your suggestion, and we are aware of the problem. According to the article, we have added some useful references to support our article.

 

Point 3: Figure 1 resumes the method proposed by the authors, so it should be written in a section called Methodology.

 

Response 3: Thank you for your reminding, we have modified the title of the second 2 to Methodology and adjusted the flowchart to section 2.

 

Point 4: The FMM method is poorly explained, with important errors that are not admissible in a scientific article (see comments in the pdf).

 

Response 4: For the FMM method, we mainly cite the methodology in the reference (Subvoxel precise skeletons of volumetric data based on fast marching methods). We have revised your recommendations in PDF accordingly. But there are many theories of the reference, and we may not be able to express all the details clearly, could you please refer to the reference for more information? If there are some other inappropriate expression, could you please to point out so that we can revise timely.

 

Point 5: I think it would be better to use a percentile, such as the 90%, to estimate the road width, especially if there are many shadows or objects that reduce it in the image.

 

Response 5: Thanks for your suggestions, we have revised the road wide assessment. In addition, each road branch has only one road width, so I think it is reasonable to compare and analyse the randomly selected road width from reference road with the extracted road width. If we understand it inappropriately, may you give us some suggestions and we will revise in time.

 

 

Point 6: In general, I find very difficult to understand Section 2.2 in terms of image processing.

 

Response 6: We have revised section 2.2 in the hope that this part could be explained clearly. For this section we cite method from the reference (Subvoxel precise skeletons of volumetric data based on fast marching methods) in chapter III, and we hope the reference will help you to understand.

 

Point 7: The same problem is present in Sections 2.3 and 2.4. I recommend rewriting it to explain the concepts taking into account the potential readers of the article and that you are working with images. Moreover, I do not understand why it is necessary all this. If you have the boundary of the road, calculate a distance map M_D, then the contour line with maximum value or M_D will follow the centerline, or not?

 

Response 7: We have rewritten sections 2.3 and 2.4 seriously and drawn some diagrams to help to explain our concepts. In addition, the maximum values in the M_d follow the centreline. However, the points on the centreline are local maximum values and their values may be not equal, so the branch backing-tracking method is used to find the pixels with the local maximum values.

Author Response File: Author Response.docx

Reviewer 2 Report

General comment:
The issue of the paper is very interesting for me, and the approach to retrieve road information looks like very improved one. But there is a need to provide further information for the readers. Please give some explanation regarding the following comments.
Comments:
1. In real, roads have various width, therefore there is a range of road width to extract road from image. Please give the applicable road width range.
2. If a road has roadside trees with long distance, there is a chance to look like disconnected or there is no road (or reducing road width? Maybe). I wonder the suggested approach is available to recognize this.

Author Response

Dear reviewer,

Thank you very much for your constructive comments to improve our manuscript. We really appreciate your precious time and great efforts on reviewing the manuscript. We have carefully addressed all the review comments and improved the quality of this manuscript accordingly. Please find below our detailed response to the comments point by point.

Point 1: In real, roads have various width, therefore there is a range of road width to extract road from image. Please give the applicable road width range. 


 

Response 1: We are so sorry for our unprofessional English writing. Thank you very much for your suggestion, we may not explain the concept of toad with. Using back-tracking method, different branches of the road can be obtained and the width of different branches is different. We use the average value of a bit of width of each branch to serve as the road width of this branch. The road width of each branch was calculated by the average value of twice times the distance from all the pixels in the centreline to the boundary.

 

Point 2: If a road has roadside trees with long distance, there is a chance to look like disconnected or there is no road (or reducing road width? Maybe). I wonder the suggested approach is available to recognize this.

 

Response 2: Thank you for your reminding, and the question you pointed out is one of the inadequacies of this article. If the road width is reduced but not fracture, then the centreline can be detected. By means of road width extraction, the width of the road width reduced position can be instead of the road wide from other positions of the branch, and the reduced road can be compensated by road reconstruction. If the fracture is too large, then the tensor voting algorithm cannot connect the broken centreline, and the road width extraction and road reconstruction cannot be carried out. Thank you very much for reminding us of this issue, we are already studying this issue and we believe it will be resolved soon.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Although you have improved the manuscript, it is still very difficult to understand in some parts, mainly due to the English.

It is mandatory to submit the manuscript to a Native English speaker or to other person who can write it appropriately.

Some parts of the article must be corrected or improved. Comments and suggestions are in the attached pdf

Comments for author File: Comments.pdf

Author Response

Dear Reviewer:

Thank you for your comments concerning our manuscript. Although we have modified our manuscript, we are so sorry for the remained mistakes. The comments are all valuable and very helpful for revising and improving our paper. We have studied comments carefully and have made correction which we hope meet with approval. We responded to your suggestions accordingly in the pdf. The responds to the reviewer’s comments are as flowing:

 

Point 1: Although you have improved the manuscript, it is still very difficult to understand in some parts, mainly due to the English. 


 

Response 1: We apologize for the mistakes we still have after the revision. We are so sorry for our unprofessional English writing and is very kind of you to point out the mistakes in the pdf. We have tried our best to revise the articles, but we are not sure if there are still errors. If there are any other mistakes, please inform us timely for us to revise.

 

Point 2: It is mandatory to submit the manuscript to a Native English speaker or to other person who can write it appropriately. 


 

Response 2: Thank you for your suggestion, we have invited an English to help correct grammar and sentences, and we hope the revised paper will be clearer and more accurate on expressions. If there are any other mistakes, please inform us timely for us to revise.

 

Point 3: Some parts of the article must be corrected or improved. Comments and suggestions are in the attached pdf.

 

Response 3: Thank you for your careful revision, we carefully modified the manuscript according to the recommendations on the pdf.

Author Response File: Author Response.pdf