Three-Dimensional Digital Documentation of Cultural Heritage Site Based on the Convergence of Terrestrial Laser Scanning and Unmanned Aerial Vehicle Photogrammetry

Round 1

Reviewer 1 Report

The manuscript titled “Three-dimensional Digital Documentation of Cultural Heritage Site Based on the Convergence of Terrestrial Laser Scanning and Unmanned Aerial Vehicle Photogrammetry” represents interesting original research.

This manuscript presents a known approach (Terrestrial Laser Scanning – TLS, and UAV-based photogrammetry) for documenting cultural heritage. Main goal of this research is a methodology for the integration of methods mentioned above for collecting a huge amount of 3D data. This problem is also already known, and there is many research papers on this topic. Despite this, the research is interesting as a case study and will attract an audience in the field of UAV and TLS data acquisition as well as, UAV-based photogrammetry and LiDAR data integration.

The manuscript title and abstract are well written and concise. In the entire manuscript, authors use a standard technical and scientific terminology. After Introduction, the authors explained in detailed study area and method, as well as achieved results. The conclusions are logical and based on the results of the research. The manuscript topics fit in ISPRS International Journal of Geo-Information aims and scope, especially in data collection and acquisition, algorithms as well as, applications of geoinformation technology.

While the topic of the manuscript is new and interesting, the manuscript should be improved. I recommend this manuscript should be reconsidered after major revision.

Comments for authors:

1.      Although one of the main goals of the research is the UAV-based and TLS data integration, it is needed to provide a more detailed description of the integration process. Suggest adding in introduction more state-of-the-art references about fusion laser scanning and photogrammetry data. This topic is not so new and adding these state-of-the-art references in the Introduction section will surely increase the quality of the paper. Proposed papers that can help authors to improve manuscript in the field of camera laser scanning and photogrammetry fusion: “Fusion of laser scanning data, maps, and aerial photographs for building reconstruction”; “Increase of readability and accuracy of 3D models using fusion of close range photogrammetry and laser scanning”; “Automatic fusion of photogrammetric imagery and laser scanner point clouds”; “Integration of Laser Scanning and Close-Range Photogrammetry – The Last Decade and Beyond”.

2.      Paper is not good structured, e.g. figure 10 must be part of the section Method. Sections Introduction and Methods must be increased. In my comments, I have mentioned several very important papers that I suggest to research it and include as references in this manuscript.

3.      Provide more data about the accuracy of the TLS and photogrammetry data (orientation accuracy, etc.).

4.      Did you provide camera calibration? How? Explain that? Have you done self-calibration in Agisoft? If you have done self-calibration each time, please provide us camera parameters variations (e.g. focal length, principal point coordinates etc.). That will be interesting to see. I am suggesting adding more details/numbers about UAS imagery orientation, camera calibration and achieved accuracy (e.g. phototriangulation RMSE). In the manuscript, new section or paragraph, as well as, state-of-the-art references about camera calibration must be added. This newly added paragraphs with the explanation of the processes mentioned above, as well as, camera calibration parameters could significantly improve the manuscript. Suggest researching state-of-the-art references about camera calibration. Proposed papers that can help authors to improve manuscript in the field of camera calibration: “Two-step camera calibration method developed for micro UAV's”; “Low cost surveying using an unmanned aerial vehicle”.

5.      Add more technical details about used UAV and picture. Did you flight autonomous or manual (UAV)? Was gimbal used or not? It will be interesting to see the camera's tilted angle stability accuracy (gimbal stability accuracy). That can be shown form exterior orientation parameters for each imagery. Suggest researching literature such as “Gimbal Influence on the Stability of Exterior Orientation Parameters of UAV Acquired Images”.

6.      Rename section name “Conclusion” to “Conclusions”.

7.      Use MDPI standard font (Palatino Linotype) on figures if you can. Remove frame on figure 5.

8.      Please provide producer details and versions for all used software.

9.      Double check reference style based on the Instructions for authors.

Author Response

Reply to Reviewer #1’s Comments (Red characters in revised manuscript)

 

Thank you for your constructive review and professional comments on our manuscript. We agree with some of the important points raised by you. Accordingly, explanations were added in the revised manuscript, or these points are addressed in the responses below.

 

1. More state-of-the-art references

- More state-of-the-art references regarding fusion of laser scanning and photogrammetry are added in the fifth paragraph of the section “1. Introduction.”

 

2. Figure 10 must be part of the section "Method"

- We agree with your comment. Further, we considered the location of figure 10 when I structured this paper. Consequently, I would like to summarize the workflow in the section “3.4. Integrated three-dimensional modeling” because one goal of this study was to establish the workflow of the fusion processes between terrestrial laser scanning and UAV photogrammetry. We hope that you will consider our response.

 

3. Sections “Introduction” and “Methods” must be increased

- Additional explanations and references for sections “1. Introduction” and “2.2. Methods” have been added.

 

4. Provide more data about the accuracy of the TLS and photogrammetry

- Overall, we reconstruct and rewrite the section “3.3. Accuracy assessment” to reflect the actual accuracy of the model.

- All the figures and tables related to the accuracy assessment are recreated.

 

5. Camera calibration

- As the reviewer pointed out, the photogrammetry requires camera calibration. However, a special camera calibration was not performed because this study used a new camera and lens for photogrammetry. These explanations and additional references are added in the second paragraph of the section “2.2. Method.” 

 

6. More technical details about used UAV and its gimbal

- Detailed explanations of the UAV and its gimbal are added in the second and third paragraphs of the section “2.2. Method.”

 

7. Rename section name “Conclusion” to “Conclusions”

- We have made the corresponding change.

 

8. Use MDPI standard font on figures and remove frame on figures

- We have made the corresponding change.

 

9. Provide producer details and versions for all used software

- We have added the producer details and versions for all used software.

 

10. Double check reference style

  - We have double checked the reference style, and some references are revised based on the instructions for authors.

11. English editing

 - The English, structure, and organization of manuscript were revised to be more systematic.

 

We hope that these responses are satisfactory and are grateful to the reviewer for the helpful comments.

 

Author Response File: Author Response.doc

Reviewer 2 Report

The paper presents the integration of terrestrial laser scanning (TLS) with digital photogrammetry for the documentation of large cultural heritage site.

Although the Authors put some efforts in the creation of the 3D model and in putting together the paper, the final result is not satisfactory in terms of statements, methodology, achievements and verification.

Indeed, the topic is not new at all and many more applications than the mentioned ones have been undertaken in different contexts and worldwile. A more careful literature review is suggested to the authors in this respect.

With respect to the case study, limited information are provided in the introduction about chronology, available bibliographic references or even real scope of the scanning: what is the scanning for? What extra information did the scanning provided and for what purpose? Was the initial requirement justified by the final level of details? Particularly "obscure" is the meaning of the phrase at line 70-71 and the multiple references to "Korean Treasure No." (perhaps a national record of monument? Please clarify).

In several place in the text, the Authors refer to "horizontal data" for UAV and "vertical data" for laser. Both references are not correct because both methods produce 3D (!) pointcloud and the 3 dimensions are X, Y, and Z, not only "vertical" or "horizontal".

At line 87 it is said that time of flight provides 3D geometrical information. This is not correct, since ToF provides only distances from the emittent to the target. It is the TLS that combines this information with the angle of the scanner "head" to produce 3d geometrical information.

Few lines after, when referring to the scanner, it is said that scanner has an accuracy of 6 to 4 mm. Normally such values are related with a distance from the scanner (i.e. "4 mm at 10 meters distance"). Please specify.

The paragraph from 99 to 105 needs rephrasing, since it is not clear (and actually quite confusing) the meaning of "direct levelling" and the procedure consisting in "fixing four integrated control points...".

At line 124-125 it is said that the scanning of roofs with TLS did not produce accurate results. Perhaps instead of inaccurate, authors should refers to the fact that point density on rooftops was not sufficient for geometrical reconstruction given the available positioning for the TLS.

At line 141 the authors say "geometric reconstruction using the seven GCP was performed to produce a more detailed 3D model". Ground control points do not provide "more details" at all. They are only used to scale and orient the model to the chosen reference system.

The statement at line 156-158 (about photogrammetry being mainly used as "visual data") is not correct. Once again, a more accurate literature review is suggested to the authors, since there are a number of paper using only (or mainly) photogrammetry as means of documentation when TLS is not available for example.

Figure 5 and 6 and their captions are not clear and need explanation as all the Accuracy assessment paragraph. It is not said, why distance between points has been chosen for accuracy assessment. What about the expected and real position of single reference points?

Furthermore, an error of 40mm in TLS makes me suspicious that a problem in TLS registration is present. Indeed, if the scanner has an accuracy of 4 to 6mm, an error of 40mm is quite high.

Authors do not report also on the error of the GCP on their own, which is an important information. Indeed, in principle TLS is more accurate than GNSS (which normally can get to an accuracy of a couple of centimeters).

Therefore, all the paragraph about accuracy assessment should be rewritten (and perhaps even the data should be reprocessed!) to reflect the real accuracy of the model.
Finally, the TLS and photogrammetry point cloud registration has been undertaken with "20 corresponding points". It is not clear what these are, but for sure there are plenty of more accurate algorithms for co-registration of pointclouds and the manual selection of arbitrary points cannot compete with the above methods.

Author Response

Reply to Reviewer #2’s Comments (Red characters in revised manuscript)

 

Thank you for your constructive review and professional comments on our manuscript. We agree with some of the important points raised by you. Accordingly, explanations were added in the revised manuscript, or these points are addressed in the responses below.

 

1. Add more references

- More references are added in the section “1. Introduction.”

 

2. Deeper research purpose

- More descriptions of the research purpose are added in the first, second, fifth, and last paragraphs of the section “1. Introduction.”

 

3. Obscure meaning

- We deleted the obscure phrase in the second paragraph of the section “2.1. Study Area.”

 

4. Modification of the terms “horizontal” and “vertical”

- We modified the terms “horizontal” and “vertical” to “planar” and “perpendicular,” respectively.

 

5. Specify the scanner’s accuracy according to a distance

- As the reviewer pointed out, the scanner’s accuracy according to a distance is added at line 98 of the section “2.2. Method.”

 

6. Rephrase GCP and levelling

- Confused contexts related to GCP and levelling was rephrased in the fourth and last paragraphs of the section "2.2. Method.”

- We mention the accuracy of the VRS GPS system used for this study in the fourth paragraph of the section “2.2. Method.”

 

7. Point density of roofs

- As the reviewer pointed out, the point density on rooftops was not sufficient. The explanation is revised as below:

High-level planar point clouds (e.g., the geometry of the roof) were not acquired sufficiently for geometrical reconstruction.

 

8. Incorrect explanation of GCPs input

- Incorrect explanation regarding the GCPs input for geometric reconstruction is revised in the second paragraph of the section “3.2. UAV photogrammetry.”

 

9. Statement about photogrammetry as visual data

- As the reviewer pointed out, statements regarding photogrammetry as visual data are deleted because it is confused and unclear.

 

10. Reconstruction of the section “3.3. Accuracy assessment”

- Overall, we reconstruct and rewrite the section “3.3. Accuracy assessment” to reflect the real accuracy of the model.

- All the figures and tables related to the accuracy assessment are recreated.

 

11. Registration for integrated three-dimensional modeling

- Detailed explanations of the registration processes are added in the first and second paragraphs of the section “3.4. Integrated three-dimensional modeling.”

 

12. English editing

  - The English, structure, and organization of manuscript were revised to be more systematic.

 

We hope that these responses are satisfactory, and we are grateful to the reviewer for the helpful comments.

 

Author Response File: Author Response.doc

Reviewer 3 Report

The presented paper has a hands-on content and meaning. The main idea of the paper is an application of the state-of-the-art technologies for culture heritage inventory and documentation. Authors gave a brief description of methods which they have used in their project. They were considering UAS and terrestrial laser scanning as complementary technologies with the aim of the following integration. Authors have chosen a good example of such integration. Despite on good experiment organization and accomplishment there are following remarks and recommendations:

16. it is questionable that "UAV photogrammetry yielded higher horizontal accuracy data accuracy than terrestrial laser scanning." Why is it so? I did not see any evidences through the paper.

46. the same as previously mentioned "UAV photogrammetry has a higher horizontal data acquisition rate than terrestrial laser scanning". I would add that it depends on object size.

85-89. the description of TLS is a commonplace, take it away.

100. please spell out why did you use GNSS and levelling for the same task (GCP)? Whether the vertical accuracy of GNSS is not enough or there were other reasons?

The main question is accuracy comparison. 

Firstly, from the statistical point of view, five measurements (distances) are not significant value for accuracy assessment. For reliable assessment we need 20-30 measurements at least, that would be good statistics.

Secondly, why did authors compare distances instead of coordinates, which are more informative (allow to identify not only discrepancies but also directions of discrepancies)?

Thirdly, the real measure of accuracy is root mean square error (RMS) or variance and not differences. Only from RMS we can infer about data quality.

Fourthly, there is another option for data comparison. We have two point clouds and a lot of artificial objects on those point clouds. Using of Leica Cyclone we can compare small objects as planes, first of all, and also sphere-like or cylinder-like objects. By the accuracy of recognition of those objects on two point clouds we can infer about overall accuracy of UAS and TLS. I realize that probably it is hardly to repeat the field experiment with more points. That is why I recommend authors to compare accuracy in a way I have recommended above.

Author Response

Reply to Reviewer #3’s Comments (Red characters in revised manuscript)

 

Thank you for your constructive review and professional comments on our manuscript. We agree with some of the important points raised by you. Accordingly, explanations were added in the revised manuscript, or these points are addressed in the responses below.

 

1. Incorrect explanation about accuracy of UAV photogrammetry

- Incorrect explanation regarding UAV photogrammetry is revised at line 16–17.

- Confused contexts regarding UAV photogrammetry was rephrased at line 55–56.

 

2. Delete commonplace description

- The commonplace description of terrestrial laser scanning is deleted in the first paragraph of the section “2.2. Method.”

 

3. Additional explanation of the levelling

- Explanation of the levelling is added at line 125.

 

4. Reconstruction of accuracy assessment

- Overall, we reconstructed and rewrote the section “3.3. Accuracy assessment” to reflect the real accuracy of the model.

- All figures and table related to the accuracy assessment are recreated.

 

5. English editing

 - The English, structure, and organization of manuscript were revised to be more systematic.

 

 We hope that these responses are satisfactory and are grateful to the reviewer for the helpful comments.

 

Author Response File: Author Response.doc

Round 2

Reviewer 1 Report

The authors have addressed all the reviewers' comments, and the manuscript in its current version is improved compared to the original.

Please, double check all reference style.

I have no further comments, and the revised manuscript can be accepted.

Author Response

Reply to Reviewer #1’s Comments (Blue characters in revised manuscript)

 

Thank you for your constructive review and professional comments on our manuscript. We double check all reference style raised by you.

We hope that this response is satisfactory and are grateful to the reviewer for the helpful comment.

Author Response File: Author Response.doc

Reviewer 3 Report

Authors made a significant improvement of their article.

After the second review what confused me it is information in tables 1 and 2.

Why the coordinate differences have the same sign in tables 1 and 2, is it really so? In such a case it forced me to think that in both cases there is a systematic shift between GPS+levelling from one hand and TLS and UAV data from the other hand (table 1). The same situation we have in table 2. If it is a real systematic error, then authors must explain a possible reason of that phenomenon or exclude that systematic shift. Otherwise, if the coordinate differences have different signs, the calculations of mean and RMS in tables are wrong and graphs in figures 5 and 7 have to be reconstructed. 

My guess, authors left numbers without signs and it was a fault.

After authors will correct the discrepancies above, I reckon that the paper can be published without additional review, insofar as those corrections will not change main conclusions of the paper.

Author Response

Reply to Reviewer #3’s Comments (Blue characters in revised manuscript)

 

Thank you for your constructive review and professional comments on our manuscript. We agree with some of the important points raised by you. Accordingly, explanations were added in the revised manuscript, or these points are addressed in the responses below.

 

1. Why the coordinate differences have the same sign in Tables 1 and 2

- As the reviewer pointed out, we left numbers without signs and it was our fault.

- We modified the same signs to the negative and positive signs.

 

2. Calculations of mean and RMS in tables are wrong

- The mean and RMS of the coordinate errors is revised in the tables 1 and 2 and the second and third paragraphs of the section "3.3. Accuracy assessment.”

 

3. Graphs in figures 5 and 7 have to be reconstructed

- Figures 5 and 7 related to the coordinate errors are recreated.

 

We hope that these responses are satisfactory and are grateful to the reviewer for the helpful comments.

Author Response File: Author Response.doc