Automatic Extraction of Indoor Structural Information from Point Clouds

Round 1

Reviewer 1 Report

Overall, I find the results of this paper to be scientifically sound and interesting. The paper is well written.

There are some minor things such as:

Line 29: Did you mean "provided by lidar" instead of "provided by radar"

Line 57 and Line 81: Please accept the reviews. Commas are red and also check for other locations.

Line 300: Check the indentation (line1, line2 ..) and overlapping

Author Response

Thanks for the reviewer's suggestion, we have corrected these errors and marked them in red in the revision.

Reviewer 2 Report

This manuscript proposes a workflow to extract structural information and reconstruct 3D models from indoor point clouds. The method consists of four algorithms. It is validated on two indoor data sets. Some major comments and questions are listed below:

1. The title of your manuscript was “Automatic Acquiring Indoor Structural Information from Point Cloud Data”. “Automatic Acquisition of Indoor Structural Information from Point Clouds” is suggested from my side. Furthermore, it is suggested to put some key words in the title to reflect the proposed algorithm.

2. In both the abstract and conclusion, you claim that your method is efficient. However, you did not do any quantitative evaluation in the efficiency. I suggest to remove all the statement about the speed/efficiency of your method.

3. Before presenting the framework in Figure 1, it is suggested to give a detailed introduction how you come up with a framework in the current way. In other words, what is the idea behind the current workflow in Figure 1? Please elaborate on that.

4. In Table 6, you list the characteristics of different reconstruction methods. However, a quantitative comparison is missing, which is necessary for a scientific paper. The section of “experiment results and comparison” is quite short compared to other sections.

5. The contribution of the paper is not that strong. For 3D reconstruction of indoor environment, it is not a common method that the 3D point clouds are projected to 2D ground and then recovered to 3D. Why not extract 3D plane from 3D point clouds directly using 3D Hough transform? Then you may already obtain nice structural shapes from your two data sets.

6. Regarding with your data from two study areas, the two scenes are not complicated. So it is convincing that your method achieved satisfactory results on these data sets.

7. In terms of scientific writing, you should put forward clearly that which is proposed by yourself, and which is an algorithm from others. Now it seems that PCA is from others but the workflow/framework is yours.

8. The reference formats are not uniform, such as the authors in Ref. 1 and Ref. 2.

9. The English writing still needs much improvement. Please check my comments below:

L1: lidar -> LiDAR. You should also give the full name of LiDAR at the first time of mentioning it.
L27: point cloud data (PCD), I have never seen this abbreviation (PCD) before, better use “point cloud data” itself.
L28: “radar”: No! radar is totally different from LiDAR.
L38: No, laser scanner is fine. Remove “or lidar” since LiDAR is a technique but not hardware.
L40: high precise -> precise
L41: in subsequent processing -> in the subsequent processing
L54: divide -> segment
L48-L85: Here is verbose. It is suggested to summarize all the work from others, find the research gap, and then motivate your own work. Do not simply list the work from others one by one.
L86: What are their “general” shortcomings?
L87: stipulate -> make a hypothesis?
L115: frame->framework
L36-L113: The statement is not clear. It seems that the “related work” is mixed with “research gap” and your own contribution.
L285: Better use (a) Grid points   (b) The positions of points in the image   (c) Image
L286: converting to the image -> converting grid points to the image
L18: Line finding method -> Line searching method?
L445: Experimental result -> Experimental results
L472: radar -> LiDAR
L477: * -> Ⅹ

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

The paper is very relevant to the scientific community. Automated reconstruction of objects is highly relevant. The indoor area is particularly important here. Although there is already some work in this area, the paper still represents an important and new contribution. 

Author Response

Thanks for the reviewer's recognition of our work, we will continue our efforts in relevant areas.