A Fast Point Clouds Registration Algorithm for Laser Scanners

Round 1

Reviewer 1 Report

The paper presents an innovative technique to record point clouds obtained from a laser scanner. The authors provide an extensive introduction to the state of the art methodologies currently used in point cloud registration, and propose a new method based on five steps to reduce the computation time while maintaining high accuracy.

The article is well structured and explained. The algorithms used are well detailed and the result is accurate and meaningful. 

The only suggestion is to add the following acronyms: VICP, GICP, Go-ICP, PLICP.

As far as I am concerned, the article should be accepted as it is.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This manuscript provides the ICP algorithm combining RANSAC algorithm, ISS, and 3DSC. The main concept of the manuscript is interesting, but the contents of comparative analysis with existing techniques were not clearly written. Besides, there are not enough comparison experiment cases, so it is difficult to analyze the performance of the proposed algorithm.

There are several comments to improve the quality of the paper.

1. Abstract and introduction

In the last paragraph, the author explained the contribution of this paper, which is very insufficient. Differentiation from conventional researches and specific techniques you proposed should be written.

An analysis of the latest papers should be added to the revised manuscript. (There is very little mention in the latest paper.)

2. Proposed algorithm

It is necessary to separate the description of the proposed algorithm from that of the existing algorithm. Also, the main contribution of this manuscript is not clear compared with other researches.

The proposed technique is a simple combination of existing techniques, and RANSAC in particular is an old algorithm. In the case of a simple combination, it is necessary to add a newly developed technique to the proposed algorithm.

3.Result

Comparison experiments and simulations with recent algorithms should be added to the revised manuscript to analyze the performance of the proposed algorithm.

4.Minor comments

There are several grammatical errors throughout the manuscript. Please have an editor review and revise the manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper aims at proposing a point cloud registration method by combining various algorithms. The paper is well structured, appropriately describes the research context and includes relevant references to support it. On this topic, any contribution that improves point cloud registration accuracy is welcome, since it eases 3D scanning or photogrammetry data processing.

On the other hand, there are some changes to make:

- Please include line numbers in the manuscript. Otherwise, it will be more complicated to review and revise the changes needed.


TITLE

"Laser scanners". It should be plural, since the title is talking about the application to that technology, not just to a specific device.



KEYWORDS

"iterative closest point (ICP)" could be included as a keyword to increase the visibility of the paper.


INTRODUCTION

Please write in full any acronym in the paper the first time they are mentioned.

Please include a final paragraph describing how the rest of the paper is organised.



METHOD

Figure 7's caption and the text referring to it could state "the normals of the point cloud" (in plural) or "the point normals of the cloud" in order to confer that sense of considering all the normals in the point set.

The steps of each methodology subsection could be highlighted by using bullet points or lists for clarity.

The Method section has content that should be included into the Discussion section, since it analyses, for instance, the disadvantages of algorithms.

In Tables 5 and 6, the Authors could highlight the highest values using bold font. This would clearly show the reader where the best results are.



DISCUSSION

The limitations of this research should be addressed here.

Figure 12 (and the text referring to it) should be moved into Results, given the nature of this content. Instead, the Authors should address the implications of their contribution.

Discussion-wise parts of the Method section should be included here, as mentioned above (e.g. first paragraph in page 7).



CONCLUSIONS

Please support this section with the most representative quantitative data, e.g. time of the registration process against other algorithms, accuracies, etc.

Although the proposed algorithm improves the point cloud registration accuracy of other approaches, the 3D representation would benefit from non-subsampled point clouds in order not to lose any data, or to minimise data loss. In my opinion, future work should also address working with optimised real-world point clouds, understanding this optimisation as the process of removing noise and subsampling to its minimum expression.



LANGUAGE

Please thoroughly revise grammar and punctuation errors throughout the manuscript.

2nd line in the Abstract: "However," instead of "But" is more appropriate.

Introduction: "laser scanner has" should be "laser scanners have" for the same reason than in the title, unless the Authors use "laser scanning".

Introduction: "ICP algorithm [1], which requirements for good initial position and high". It should be "which requires a good... and a high".

"Recent years, many variants..." should be "In recent years, many variants..."

"which is mainly consist of" should be "which mainly consists of".

Subsection 2.5: "we uses" should be "we use".

Subsection 3.2: "how many points in the point cloud is involved in" should be "ARE involved in".

Among many others.

Discussion: "The registration results of each algorithm is shown" should be "ARE shown".

"showing good efficient" should be "showing a higher efficiency". Next, I would use "Therefore", "Hence", "Consequently", etc. instead of "So".

Please use commas to separate thousands in the numbers in Table 1. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors improved the introduction by providing additional information. However, the sentence "Optimization based methods search corresponding point of each point in another point cloud, then estimate the transformation matrix by using the corresponding relation." should be improved.

The updated labels better describe the meaning of the images.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

I read the response to the comment and revised manuscript, but I still think the contribution of the manuscript is not enough to be published in this journal. Although the authors have written a lot in the revised paper about the contribution of the paper, there is still a lack of explanation for the contribution of the paper. (Differentiation and performance improvement compared with conventional methods)

The response to the comment was not sufficiently written in the revised manuscript. In particular, the explanation of the improved RANSAC is not well written in the manuscript. Besides, A lot of RANSAC algorithms have been studied for decades, and it is also unclear which classical RANSAC algorithms the author compared with the proposed RANSAC algorithms.

The authors added the analysis of results, but there is no detailed performance analysis in the revised manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The Authors addressed all my comments. However, they could have highlighted the changes so that the review process was easier.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 3

Reviewer 2 Report

The response to the comment was not sufficiently written in the revised manuscript.