A Practical Algorithm for the Viewpoint Planning of Terrestrial Laser Scanners

Round 1

Reviewer 1 Report

The authors present a viewpoint planning method for complex scenes, using terrestrial laser scanners. The topic is interesting for Geomatics topic.

 There are 2 major problems with this paper which do not let me infer the relevance of the mentioned results: the novelty of the paper and its "benchmark" paper.

The problem with the novelty of the paper are the references selected. There are 33 references and only 8 of them are from past 5 years (which 3 are from the same authors). More than 50% of the references should be from the past 5 years.

The second problem is with the benchmark technique. It has more than 10 years old and is not suitable for the comparison when the idea is to show advances in the field. I suggest to use another technique or to make a very detailed text justifying its selection (in the paper). 

There are 3 self-citations [22,23,25]. I recommend to revise if they are really needed in the text. If yes, please justify. 

When describing the paper contribution, please add the scientific contribution for the area.

The text and explanations along the text are ok; The figures are clear and complete.

In the results section, WGA average runtime is almost 5 times the OGA, and even more than Couto et. al. It really needs an better explanation of why this is acceptable. The cost-benefit of the increasing runtime vs the decrease of VP doesn't seems to be a good one.

I recommend using Confidence Interval of the results instead of using only the average. Or at least, the average with the standard deviation.

In the conclusion: "However, by having the processing time under eight minutes, it can already be used for on-site planning." -> This need to be further explained in the text, because there are so many variables thus it is hard to affirm something. I think it really depends.. 8 minutes was the result under the authors hardware, but it may take a lot more in a lesser configuration. Maybe the application uses even more than 500 vertices. Maybe the authors can specify in which conditions/application this can be useful.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Your research is excellent. I would try though to compare a real scanning project with the theoretical results. 

Author Response

Thank you for your kind review and suggestions. Yes, I agree it would be better to apply the method to a real scanning project. This would be investigated in our further research and publications.

Reviewer 3 Report

I have recently reviewed the original version of the manuscript entitled "A Practical Algorithm for the Viewpoint Planning of the Terrestrial Laser Scanners" on 2022-04-22 . The study in this paper was very interesting. The research have high novelty  and the descpription of the proposed method was clear, so I don't have any  suggestions. This paper can be accepted in this form.

Author Response

Thank you for your kind review. Your time and efforts are highly appreciated. 

Reviewer 4 Report

This paper proposed a practical strategy to resolve the problem of the optimal placement of the terrestrial laser scanner and tested on 540 polygons simulated with different sizes and complexities and further achieve good results compared with benchmark studies. I believe that it is an excellent study to fill such gaps in this area. However, there are a few concerns before I recommend it for publication.

1. Background research is a bit not well organized. I would suggest that the authors avoid using the term 'State-of-art of' and add 1-2 more paragraphs to better illustrate what are the potential problems facing and how do you improve them in this paper.

2. in a normal real TLS scanning project we would not have to deal with the random polygons over maybe 50 vertices, therefore some more detailed discussion with smaller vertices scenarios would be interesting for the real TLS planner, maybe the authors could add some discussion about it, but it is just a suggestion and you do not have to address this in the paper.

3. in 4.3. Computational efficiency, did you use the same computation configuration when compared with benchmark results?

4. line 267, should be  'Figure 7' instead of 'Figure 5'.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors did solve most of the problems that I mentioned, although it still can be further improved. While I respectful disagree with one or two answers, the paper has improved enough for publication.

There is one thing that I could not fully understand and maybe will be the same for the reader: Figure 13.  The blue lines are the same average results Couto et al. from Table 4. The orange ones are the reduced VPs (but it is not clear from where). When comparing the average results of WGA from the same table 4, the difference is not so great as indicated here. In #500 - orthogonal, for example, the difference between both averages is ~3 and not 8 as indicated.

Maybe this figure is from the same situation from Figure 12. Anyway, it just needs some clarification or correction.

The results / paper would be significantly better if authors had implemented the benchmark technique in the same machine for a direct comparison. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf