3D Structure from 2D Dimensional Images Using Structure from Motion Algorithms

Round 1

Reviewer 1 Report

Dear Author,

Thanks for your effort in using efficient algorithms to build 3D structures from 2D Images. This is a good contribution that reduces costs on relevant implementations. However, a lot of references, non-clear images, and a lack of mathematical derivations are barriers, which leads to your paper needing a major correction at this moment. The correction is suggested as follows:

 1. References/Citations: they are not well shown in the article; lots of them are missed, please correct them (Line No.: 87, 93, 102, 106, 117, 128, 154, 161, 227, 232, 238, 247, 253, 254, 269, 270, 286, 291, 293, 297 and 299)

 2. Figure: resize Figure 3 and full information of the figure should be given. 

 3. Table: In Table 3: no need to show the unit again in the table as you give in the title (of the table).

 4. Mathematical derivations: it could give related derivations in mathematics to show and explain Motion Algorithms, which could help readers to understand in depth.  

5. Please remove ultra space between two words in the paper, such as in Figure 5, in front of Regard3D software; Check carefully your paper and then correct others.

There may have other grammar/spell errors, which are not listed above. Please have a double check before your next submission. 

Best wishes.

Author Response

Reviewer:1
Thanks for your effort in using efficient algorithms to build 3D structures from 2D Images. This is a good contribution that reduces costs on relevant implementations. However, a lot of references, non-clear images, and a lack of mathematical derivations are barriers, which leads to your paper needing a major correction at this moment. The correction is suggested as follows:

1. References/Citations: they are not well shown in the article; lots of them are missed, please correct them (Line No.: 87, 93, 102, 106, 117, 128, 154, 161, 227, 232, 238, 247, 253, 254, 269, 270, 286, 291, 293, 297 and 299)

Responses:

The titles of the figures and tables have been reinserted, and they have been renumbered and positioned suitably inside the text body. Please note all figures and tables.

3. Table: In Table 3: no need to show the unit again in the table as you give in the title (of the table).

Responses:

The duplicated units have been eliminated; thank you for your comment; and please see table 3 on page 332.

4. Mathematical derivations: it could give related derivations in mathematics to show and explain Motion Algorithms, which could help readers to understand in depth.

Responses:

Thank you for your valuable comment, and actually there are three paradigms of SFM method, incremental, global and hierarchical. Each of the three sorts of solutions has its own set of long equations and derivatives, and I don't think it's necessary to include all of them in the current study, which is a repetition. I included a few references for the in-depth reader to utilize as needed.

5. Please remove ultra space between two words in the paper, such as in Figure 5, in front of Regard3D software; Check carefully your paper and then correct others.

Responses:

Thank you for your note, and please note the final manuscript word file, which was verified to detect and delete extra whitespace between words.

There may have other grammar/spell errors, which are not listed above. Please have a double check before your next submission.

Responses:

Thank you for your feedback; the MDPI staff will conduct an English language assessment, and their certificate will be attached.

Regards,

Author

 

Reviewer 2 Report

Overall the paper is well written, however, the authors did not discuss state of the arts using deep learning. The following works are relevant, which need to to be discussed in the related works : Feature boosting network for 3D pose estimation

 

Author Response

Reviewer:2
Overall the paper is well written, however, the authors did not discuss state of the arts using deep learning. The following works are relevant, which need to to be discussed in the related works : Feature boosting network for 3D pose estimation

Responses:

The text has been updated to include more details and a literature review on applying deep learning for SFM solutions. There are some recent references as well. Thank you for your valuable comment. Please note pages.

Regards,

Author

 

 

Reviewer 3 Report

This paper presents a simple method to obtain 3D surface information from 2D photo and software. It is a topic of interest to the researchers in the related areas. However, frankly speaking, there are many errors in figures and main text. The paper needs very significant improvement before acceptance for publication. My detailed comments are as follows:

 

1. Please check the reference. There are many “Error!” which is shown in Line 117, 128, 154, 161, 231, 238, 247, 253, 254, 269, 270, 287, 291, 293, 297, 299. The Reference source could not be found.
2. I suggest to delete the sentences from Line 84-87, which describe “Historical Emara Palace in Najran”. There is no necessary to demonstrate the background of the building.
3. There are a lot of problem in Figures. Please rearrange them and improve the quality. For example, there is huge empty space between two photos (Figure 1). Also, there is no scalebar in any photos. Some figure is too large (Figure 2).
4. You cannot directly use the reference umber as the subject to start your sentence (Line 58, 60, 227, 232 and so on). You should use the researchers’ name.
5. I definitely agree the 3D modeling of heritage buildings is significant and necessary for their identification, monitoring, and restoration. However, the 3D modeling of a building is not only the wall outside. What about the other rooms behind the walls? Please discuss more about it.
6. In section 4.3.2, why there is no root mean square error without ground control points in the method VisualSFM and Regard3D. Please add the explanation.
7. I suggest author to move Table 4, 5 to supplementary information. There is no necessary to show in such table in main text and it is not easy to read. Can you change the data into some other figures?
8. In Figure 8(a), the variations are less than 5 cm, but the C2C absolute distance shows the difference is nearly 2. There are only one small spot is red. How this red spot come from?
9. In Line 292, the mean distance was 0.067m. You have too many typos. Please double check before submission.
10. In the figure caption, remove the first label ”(a)”.
11. There are some useless space in the main text. Such as Line 334, 335 and so on. Please double check before submission.

Author Response

Reviewer:3
This paper presents a simple method to obtain 3D surface information from 2D photo and software. It is a topic of interest to the researchers in the related areas. However, frankly speaking, there are many errors in figures and main text. The paper needs very significant improvement before acceptance for publication. My detailed comments are as follows:

 

1. Please check the reference. There are many “Error!” which is shown in Line 117, 128, 154, 161, 231, 238, 247, 253, 254, 269, 270, 287, 291, 293, 297, 299. The Reference source could not be found.

Responses:

The titles of the figures and tables have been reinserted, and they have been renumbered and positioned suitably inside the text body. Please note all figures and tables.

1. I suggest to delete the sentences from Line 84-87, which describe “Historical Emara Palace in Najran”. There is no necessary to demonstrate the background of the building.

Responses:

Thank you for your comment; however, in Saudi culture, there is an increasing awareness of the importance of documenting heritage sites and spreading this knowledge to all sectors of society. As a result, I used the royal palace, which is one of the most well-known sites in the Najran region, as a case study in my research.

 

1. There are a lot of problem in Figures. Please rearrange them and improve the quality. For example, there is huge empty space between two photos (Figure 1). Also, there is no scalebar in any photos. Some figure is too large (Figure 2).

Responses:

All figures have been downsized and increased in quality. In addition, figure 2 now has a scale bar.

1. You cannot directly use the reference umber as the subject to start your sentence (Line 58, 60, 227, 232 and so on). You should use the researchers’ name.

 

Responses:

The IEEE recommended style for The Sustainability journal generates the authors' names with numbers. If the style is altered, the authors' names will display as subjects in the text. Thank you for your comment.

1. I definitely agree the 3D modeling of heritage buildings is significant and necessary for their identification, monitoring, and restoration. However, the 3D modeling of a building is not only the wall outside. What about the other rooms behind the walls? Please discuss more about it.

Responses:

I completely agree with you that numerous images should be acquired for inner portions and linking outside and inside with enough overlapping images in order to generate a complete 3D model for the heritage structures. Thank you for your comment, please note additional information in last paragraph discussions part page 338-343

 

 

1. In section 4.3.2, why there is no root mean square error without ground control points in the method VisualSFM and Regard3D. Please add the explanation.

 

Responses:

I was unable to discover an answer to this topic by looking for how to compute the RMSE for the BA step without employing any GCPs for the VisualSFM and Regard3D programs. I expect that they are relatively new and open-source applications, and that they will be implemented in the future.

 

1. I suggest author to move Table 4, 5 to supplementary information. There is no necessary to show in such table in main text and it is not easy to read. Can you change the data into some other figures?

Responses:

I appreciate your note, but In fact, tables 4 and 5 indicate the bundle block adjustment accuracy of the GCPs and CPs in X, Y, and Z values. Each column indicates the difference between total station coordinates (reference) and the SFM algorithm solution approach at each ground control point. The RMSE values were calculated using Equation 1. Presenting the RMSE values is useful for evaluating the procedures that were employed. While the discrepancies were small, as shown in the tables, the values might be displayed as a table or in figures and chosen to be displayed in the tables.

1. In Figure 8(a), the variations are less than 5 cm, but the C2C absolute distance shows the difference is nearly 2. There are only one small spot is red. How this red spot come from?

Responses:

 

The red area is located behind the wall next to the entrance to the palace and is not in its level, which led to a large error, as you can see in the figure below inside black circle.  Thank you for your comment.

 

Figure 1. Red area

1. In Line 292, the mean distance was 0.067m. You have too many typos. Please double check before submission.

 

Responses:

Thank you for your advice; I double-checked the data, and as shown in Figure 2 below, the same results were achieved. Please note the numbers within the red circles.

 

Figure 2.

 

 

 

 

1. In the figure caption, remove the first label ”(a)”.

 

Responses:

The first label ”(a)”removed.

 

1. There are some useless space in the main text. Such as Line 334, 335 and so on. Please double check before submission.

 

Responses:

Thank you for your advice, I scanned the entire text for any unnecessary space and removed it.

 

Regards,

Author

 

Reviewer 4 Report

In the manuscript entitled " 3D Structure from 2D Dimensional Images Using Structure from Motion Algorithms” author reported the comprehensive study 3D modeling of historical Emara Palace in Najran, Saudi Arabia. Unfortunately, heritage buildings around the world are at risk due to natural catastrophes and/or human activity and by using a 3D modelling their identification, preservation and restoration can be achieved. The topic is interesting and relevant to the field due to the fact that it researches a new approach of preservation the historical buildings by using a 2D Dimensional Images and Structure from Motion (SfM) Algorithms. In this manuscript a low-cost digital camera and three SfM processing software were used to obtain a 3D surface information for historical Emara Palace. Set of captured image datasets were utilized to evaluate the quality of SfM software products: Agisoft Metashape, VisualSfM, and Regard3D.

The manuscript is well organized, it clearly presents the objectives of research, the author used the scientific methods, and they are adequately described. Experimental part is appropriate and the results and discussion are correctly described. Conclusion part is consistent with the presented objectives. Author emphasizes in the conclusion section the advantages of the usage of a low-cost photogrammetry software and indicates their possible limitations.

The figures, tables and images are appropriate and can be interpreted easily.

The references cited in the manuscript are recent, mostly within the last 10 years.

Having in mind that the presented theme is interesting and the topic is attractive, some issues should be corrected before the suggestion to publish the manuscript.

One can see that the part of the reference numbers in the manuscript are not displayed correctly and instead of number and Error text is displayed. That should be corrected.

References in the brackets are not correctly written. For two or more references one bracket should be used, do not use separate bracket for every reference number.

Lines 30 and 73 have too much space between words.

Links to Internet addresses should not be mentioned in the text, they should be listed in references.

The caption below Figure 1 should contain a) and b) markings of the images.

The caption below Figure 2 is incorrectly stated and part of the image in Figure 2 appears to be missing.

The caption below Figure 3 is incorrectly stated and part of the image in Figure 3 appears to be missing.

The reference list is not correctly written.

Corrections are also marked in the attached file.

Acceptance of the manuscript is suggested after all the corrections are made.

Comments for author File: Comments.pdf

Author Response

Reviewer:4
In the manuscript entitled " 3D Structure from 2D Dimensional Images Using Structure from Motion Algorithms” author reported the comprehensive study 3D modeling of historical Emara Palace in Najran, Saudi Arabia. Unfortunately, heritage buildings around the world are at risk due to natural catastrophes and/or human activity and by using a 3D modelling their identification, preservation and restoration can be achieved. The topic is interesting and relevant to the field due to the fact that it researches a new approach of preservation the historical buildings by using a 2D Dimensional Images and Structure from Motion (SfM) Algorithms. In this manuscript a low-cost digital camera and three SfM processing software were used to obtain a 3D surface information for historical Emara Palace. Set of captured image datasets were utilized to evaluate the quality of SfM software products: Agisoft Metashape, VisualSfM, and Regard3D.

The manuscript is well organized, it clearly presents the objectives of research, the author used the scientific methods, and they are adequately described. Experimental part is appropriate and the results and discussion are correctly described. Conclusion part is consistent with the presented objectives. Author emphasizes in the conclusion section the advantages of the usage of a low-cost photogrammetry software and indicates their possible limitations.

The figures, tables and images are appropriate and can be interpreted easily.

The references cited in the manuscript are recent, mostly within the last 10 years.

Having in mind that the presented theme is interesting and the topic is attractive, some issues should be corrected before the suggestion to publish the manuscript.

One can see that the part of the reference numbers in the manuscript are not displayed correctly and instead of number and Error text is displayed. That should be corrected.

References in the brackets are not correctly written. For two or more references one bracket should be used, do not use separate bracket for every reference number.

Lines 30 and 73 have too much space between words.

Responses:

Thank you for your advice, I scanned the entire text for any unnecessary space and removed it.

 

Links to Internet addresses should not be mentioned in the text, they should be listed in references.

Responses:

When citing a webpage in Modern Language Association (MLA) or University of Chicago style, the footnote information is placed at the bottom of the page. However, some journals advise against using footnotes.

The caption below Figure 1 should contain a) and b) markings of the images.

Responses:

The figure has been updated, and any missing letters have been replaced. Sea figure 1, thank you for your valuable comment.

The caption below Figure 2 is incorrectly stated and part of the image in Figure 2 appears to be missing.

Responses:

The figure has been updated, and any missing letters have been replaced. Sea figure 2, thank you for your valuable comment.

 

The caption below Figure 3 is incorrectly stated and part of the image in Figure 3 appears to be missing.

Responses:

The figure has been updated, and any missing letters have been replaced. Sea figure 2, thank you for your valuable comment.

 

The reference list is not correctly written.

Responses:

The reference list has been updated, thank you for your valuable comment.

 

Corrections are also marked in the attached file.

Acceptance of the manuscript is suggested after all the corrections are made

 Regards,

Author

 

Round 2

Reviewer 1 Report

Dear Author,

Your effort can be seen in your amended paper which is significantly improved in quality and could be submitted for publication after you finalise the paper and have a final check, specifically such error on Line 261.

Best wishes

Author Response

Thank you for your valuable review and I’m so grateful for all the constructive comments. Some corrections were made such as lines 91, 110, 261, 256, 257, and 273. Also, you could use the trach changes to note all added corrections. I hope my revision has improved the paper to a level of your satisfaction.

Reviewer 3 Report

Thank you for your reply.

Author Response

Thank you for your valuable review and I’m so grateful for all the constructive comments. Some corrections were made such as lines 91, 110, 261, 256, 257, and 273. Also, you could use the trach changes to note all added corrections. I hope my revision has improved the paper to a level of your satisfaction.