Three-Dimensional Unique-Identifier-Based Automated Georeferencing and Coregistration of Point Clouds in Underground Mines

Round 1

Reviewer 1 Report

The authors have attempted to develop a sound method for automated georeferencing and co-registration methodology for point cloud data in underground mines. The research topic is of importance to the mining industry and is of significance to researchers working in GNSS denied environment.

Although most of the content in the manuscript is of good quality, some improvements are  required to further enhance the quality.

Specific comments:

line 17: "georeference" instead of "georeferenced" and "co-register" instead of "co-registered"

line 20: avoid the use of "accurate, effective and efficient"; just providing the accuracy is sufficiently convincing for the readers.

line 38: "...few easily distinguishable control points are marked using ...": Are these control points inside mines or outside? Please clarify in text.

line 67: what does "multi-sensor integration" refer to? Please expand. "artificially induced"? Please expand on the 'how' of this.

line 68: How many 'discernable GCTs' are required per unit area? Please expand.

line 71: "power source"... please specify what kind of power source is required? Electrical?

line 73-74: "Coordinates of the installed location of tags...": which coordinates: global or local? Please specify for the purpose of clarity.

line 85-86: "... between the RFID tags were increased" due to which factors?

line 102-103: "... even in a low-resolution point cloud.": Please provide approximate density.

line 127: "The other GCTs which ...": what does 'other' refer to here?

line 140: "... and automatically georeference..."; "georeferences"

Figure 1: images not clear. Please provide better images to improve visibility. Overall, please check Figure 1 and the references made to it in the text. I see some inconsistencies. In the caption (line 196), the same symbol (∧) has been used to refer to two different assemblages. Please check and correct. Also, [(c) in line 197: a spacing of 2.5 cm is mentioned; whereas, in text (line 179) it's mentioned as 3.0 cm - please check and correct.

line 158-159: Replace "the top row in Fig.1a" with "Fig 1a, Row X". The shape ∧ with < when referring to Fig 1a, Row 3; otherwise, it's confusing to the reader because the vertical assembly is also ∧.

line 159: The depth (30 mm) is of significance to this experiment. Please provide depths for Fig 1a, Row 1 and Row 2 as well.

line 161: "... determining the separation threshold...": The separation threshold will have a relationship with scan density, frequency and scan distance. Please expand.

line 167: "... 3DUID was kept at 5 m...": will this distance of 5 m need to be calibrated for a different environment/scanner? What will be the influence of this 5 m distance on the design of 3DUID?

lines 172-174: Test 3 does not seem to support this statement. The higher the reflectivity, the better the backscatter, resulting in an increased number of scanned points, resulting in better characterisation/segregation. However, we experience noisier point cloud? Please check, verify and clarify in the text.

lines 183-186: Please provide a relevant reference to this statement. Also, provide specs/details of the laser instrument used to perform this experiment.

line 203: "algorithm" should be "Algorithm" (starts with an uppercase)

Figure 2: 'b' and 'c': are mirror images. Please consider flipping one of these to match the other.

Algorithm 1: line 7 - "... of 1 not connect to edge ..."; "connected"

line 256: consider replacing "leave-one-out error" with "leave-one-out validation" for the purpose of consistency.

line 282: Use a hyphen in between 'SLAM' and 'processed' to make it an adjective (SLAM-processed)

line 296-298: Requires a reference

line 298-299: Requires a reference

line 313: Requires a reference

line 319: what does the radius of 0.04 m depend on? Why was this number selected? Please explain.

line 345: "... small rod like structure...": please label this in Fig 3c

line 357: "... perimeter of a triangle was used [43].": Consider replacing "used" with "employed"

line 361: "... range is measured from the point to the nearest trajectory.": Does this refer to the Euclidean distance between point and the nearest trajectory?

line 372: "Geometrically, ...": will this not be "trigonometrically"?

line 380: "... from wall to 3DUID occur.": Should be "occurred"

line 390: why was 50 selected as a filtering threshold and not 51? Please expand on this. Also, please show the filtered point cloud in Fig 6e after spurious points were removed.

line 422: "combination" will be "combinations"

line 451: Please expand to clarify the meaning of "The two algorithms were chosen due to efficiency in processing time." If this was an intended part of the experiment, then it should be mentioned in the design/methods section.

lines 452-455: Does this mean that NDT takes longer to process the point cloud? If so, please mention.

line 516: "... within 1 m ...": according to Table 1, this is only true for 'best case'.

Discussion: please bring out any potential issues that 3DUID and the 3DReG workflow might have. For example, mention environments where this method might not be applicable, required density of point clouds, required dimensions and material for 3DUID and indicative cost. This is believed to not only strengthen the discussion and conclusion but will also identify future research.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors,

I have reviewed the paper entitled "Three dimensional unique identifier based automated georeferencing and coregistration of point clouds in underground mines". In my opinion, the paper is interesting but I am very confused because I dont understand the novelty of the Author's solution. Laser scanning in mining approach is commonly in use. Therefore, the developed approaches are used by companies which are managing mines. I think that by helping me understand the solution, the paper can be accepted for publication. For now, it needs improvements. More comments below.

Kind regards.

The Abstract is inconsistent. It is obvious that GNSS would not work in mines. Moreover, there is no word for the SLAM technology. In addition, it is unclear what the results are. What are the requirements for conducting measurements in mines. Is there a matrix installed in the mines to, for example, increase the accuracy of the results obtained? All of this requires a brief explanation.

Introduction is well written but still it needs improvements by focusing on the authors method.

At the very beginning of chapter 2, I would describe the equipment, tie points characteristics, the concept of using methods. Later, I would present a workflow chart that would introduce me to the whole technology. I would consider whether some of the definitions described above can be shortened, significantly increasing the readability of the article. In their current form, the materials and methods are heavy to assimilate.

The results are presented OK. The discussion, however, is too brief for me. Some of the information from chapter 4.2 should be included in the results. At the same time, coming back to the beginning of my attention. The discussion should include specific information about the advantage of the authors' method and then a discussion about similar other experiments to confirm Authors  thesis.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper presents a methodology for profile mapping and data extraction in underground mines.

The current knowledge in the field is presented in the Introductory section giving valuable input for the current research.

The methodology presented in the paper is using custom-defined indicators to correlate scans to ground measurements. The method for defining indicators is detailed, pointing to the reliable solution for constructing such devices.

Subsequently, the scanning procedure was presented. The method for data processing is presented in detail, and sufficient information is provided to explain the procedure for identifying indicators.

There is sufficient information related to the procedure of correlation between scanned data and ground data.

The work is complex, and the methods are explained in detail.

There are only minor that should be addressed by the authors:

• The use of “error” should be avoided. It is suggested to use “difference” as none of the methods provides absolute data;
• A summary pointing to the minimum actions required to acquire reliable data should be presented in the Concluding sections. This information is missing, as the authors are pointing only to the differences between the two data sate. It will be valuable to indicate the procedure to be followed in the first iteration and how a third-party user should relate to this method to obtain reliable data.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The Authors have sufficiently addressed the points raised in the previous review. Following are some minor edits and suggestions. Also, the some statements identified below will either need a relevant reference or a valid justification to support these statements.

lines 82-84: A higher number of GCTs comes with a cost. It is suggested to include cost implications associated with additional GCTs

lines 147-148: "..., which can be recognised in 3D point clouds are either difficult to construct or hard to decode.": Please either provide a reference or a justification to support this statement.

lines 171-173: The three main steps given in the manuscript are numbered 1, 2 and d. Please correct.

lines 201-203: Range (or any other) accuracy given in a scanner's manual is the accuracy of that scanner in a controlled environment and thus cannot be used unless tested in a desired environment. The statement should either be revised with caveats included or supported with a reference.

lines 357-358: Consider providing these numbers in millimeters

lines 434-435: "..., any value sufficiently large but less than average points present in 3DUID boundary grids works.": Please either provide a reference to support this statement or mention if it was empirically tested.

Author Response

Please see the attachment 

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors, 

Thank you for addressing all of my comments. In my opinion the paper can be accepted. 

Author Response

The authors of the manuscript thanks the reviewer for spending their valuable time in reviewing the manuscript and providing valuable comments.