One-Step Enhancement Method for Data Registration Based on the Lidargrammetric Approach

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. It is recommended to be more specific about the data types (e.g., synthetic, semi-synthetic, real data) corresponding to “30-fold, 12-fold, 96-fold error reduction” in the abstract to enhance readability.
2. In Section 2.2, specific algorithms or examples of ULPI (Unique Point Identifier) generation are added to help readers understand the technical implementation.
3. In Section 3.4, it is suggested to add a comparison between the ICP algorithm and the lidargrammetry method in terms of computational efficiency (e.g., time-consuming, resource usage).
4. In Section 2.3.3, add background information such as the collection time of real data, weather conditions, etc. that may affect the results.
5. Equation (1) suggests using curly brackets on the rightmost side of the x and y formulas, and suggests avoiding page breaks

Author Response

Dear Reviewer!

We are extremely grateful for your comprehensive evaluation. It is our conviction that the aforementioned text has been improved by the implementation of the suggestions and comments that you so kindly provided.

We have responded to all the comments below.

All the comments were very helpful and each one has been taken into account, resulting in additional research and changes to the text.

Best regards,

Authors

 

Comments 1:

 

1. It is recommended to be more specific about the data types (e.g., synthetic, semi-synthetic, real data) corresponding to “30-fold, 12-fold, 96-fold error reduction” in the abstract to enhance readability.

Response 1:

We do not intend to longer the abstract. The specification of the data is inserted to the main text of the paper.

 

Comments 2:

1. In Section 2.2, specific algorithms or examples of ULPI (Unique Point Identifier) generation are added to help readers understand the technical implementation.

Response 2:

We do not intend to present all the details of the solution in this paper. It would enlarge it to much.

 

Comments 3:

1. In Section 3.4, it is suggested to add a comparison between the ICP algorithm and the lidargrammetry method in terms of computational efficiency (e.g., time-consuming, resource usage).

Response 3:

The method was tested and its implementation for assessment of the concept, not for productivity efficiency. The time and resources consuming is a subject of optimization, next step of method implementation.

 

Comments 4:

1. In Section 2.3.3, add background information such as the collection time of real data, weather conditions, etc. that may affect the results.

Response 4:

Extended analysis of the external conditions on the result of lidargrammetric approach application will be presented in a separated paper.

 

 

Comments 5:

1. Equation (1) suggests using curly brackets on the rightmost side of the x and y formulas, and suggests avoiding page breaks

Response 5:

Equation (1) is correct according to the literature.

Reviewer 2 Report

Comments and Suggestions for Authors

1. Summary

This manuscript introduces a LiDAR approach for registering and enhancing LiDAR data, particularly in scenarios where trajectory data is missing or unreliable.

Its core innovation is the projection of LiDAR point clouds into synthetic images (lidargrams), the assignment of unique identifiers (ULPIs) and the conduct of photogrammetric adjustments to refine 3D coordinates.

This method has been implemented in the custom-built PyLiGram tool and has been tested extensively using synthetic, semi-synthetic and real-world datasets. The results show significant improvements in geometric accuracy, including a reduction in error of up to 96x in real data.

 

2. General comments.

Improvement 1: Reproducibility.

Scientific validity and impact are increased when other researchers can replicate and build on the work. I suggest:

• Publish synthetic and semi-synthetic test datasets used in the study, ideally in a public repository (e.g., Zenodo, GitHub, or institutional archive). Include Point clouds in standard formats (e.g., LAS/LAZ, ASCII), Associated GCP and CHP files (XYZ, intensity/RGB) and JSON and XML configurations used in the processing workflow.
• Release the PyLiGram research tool or at least a basic version or script with example usage.
• Provide a minimal working example (MWE) showing how to process a simple dataset through the full pipeline (steps 1–4), ideally with output metrics (RMSE/Max).
• If open-source release is not possible, consider including a supplemental appendix with detailed pseudocode and data samples.

 

Improvement 2: Clarity

Dense text can obscure understanding, particularly for readers less familiar with lidargrammetric ideas.

I suggest to:

• Shorten or split overly long paragraphs, particularly in the Introduction and Discussion.
• Use bullet points or numbered lists where appropriate (e.g., outlining steps in data processing or matching).
• Consider including a conceptual overview or summary flowchart

 

Improvement 3: Figures

Visuals should enhance understanding even without direct reference to the main text. I suggest that each figure should have a descriptive caption that summarizes what the figure shows, such as type of data (synthetic, semi-synthetic, real) and purpose of the visual (e.g., data layout, GCP distribution, registration results).

In addition, you should add visual annotations (legends, axis labels, scale bars, or markers for GCP/CHP locations). For example, in Figures 3–5 (dataset descriptions), show strip numbers and orientation.

Please use color-coding consistently (e.g., GCPs in cyan, CHPs in orange), and always define them in the legend.

 

Improvement 4: Performance Summary Table

Consider adding a single table summarizing performance before the Discussion section (e.g., %RMSE reduction, across datasets and benchmarks). A summary table improves digestibility and comparison.

 

Improvement 5: Matching process

Matching is critical to accuracy and was acknowledged as a limitation in the paper. Expand discussion of how matching was performed in Metashape, including any challenges with intensity images.

Some suggestions:

• Expand the description of Agisoft Metashape's role: Were SIFT/SURF-like features used?, Was manual intervention required, especially in low-texture (e.g., intensity-only) data?
• Discuss preprocessing steps used for intensity images: Clarify how contrast/gamma correction was applied, Could histogram equalization or sharpening improve results further?
• Consider providing examples of lidargrams before/after enhancement, and describe what visual quality was necessary for successful matching.
• State whether deep learning-based methods were evaluated, given their mention in the Future Work.

 

3. Specific comments:

Improvement 6 – Abstract

The abstract is dense, consider simplifying the explanation of the method for accessibility.

 

Improvement 7 - Section 2

The comparison with commercial software is strong; consider tabulating key differences for reader clarity.  A table contrasting input requirements, trajectory use, flexibility, and target use cases (e.g., TerraMatch vs. PyLiGram) would improve reader understanding.

 

Improvement 8 - Section on Synthetic Data: 

Consider clearly stating why orthophotos were used and how RGB was derived. For example, explain if orthophotos were used only for visualization or to simulate real-world reflectance in lidargram generation.

 

Improvement 9 - Real Data Section

Good explanation of intensity-based matching, but contrast enhancement methods could be described more explicitly.  Indicate specific changes (e.g., contrast stretch parameters or gamma values) and whether they were automated or manual.

 

Improvement 10 - Table 8

Clarify the basis of the “trajectory approximation” comparison. Does the increased number of lidargrams equate to finer spatial resolution or something else?. Explicitly state how generating more lidargrams improves change (e.g., denser camera positions yielding more accurate EOP interpolation).

 

Improvement 11 - Discussion

The paragraph comparing PyLiGram to TerraMatch is fine. However, a more direct statement on where PyLiGram outperforms (non-rigid, trajectory-less scenarios) should be emphasized. Consider summarizing PyLiGram's niche, such as "ideal for legacy or GNSS-denied data where trajectory is missing or unreliable."

 

4. Ethics, references and data availability

Ethics and Conflicts: No ethical issues.

References: Enough, recent (many from past 5–10 years), and relevant. Some legacy references (e.g., early lidargrammetry origins) are necessary for historical context. Self-citations are not excessive.

Improvement 12 (related to 1) Data Availability

Marked “Not applicable.” This is a weakness. The authors should at least share synthetic and semi-synthetic data for replication.

Author Response

 

1. Summary

This manuscript introduces a LiDAR approach for registering and enhancing LiDAR data, particularly in scenarios where trajectory data is missing or unreliable.

Its core innovation is the projection of LiDAR point clouds into synthetic images (lidargrams), the assignment of unique identifiers (ULPIs) and the conduct of photogrammetric adjustments to refine 3D coordinates.

This method has been implemented in the custom-built PyLiGram tool and has been tested extensively using synthetic, semi-synthetic and real-world datasets. The results show significant improvements in geometric accuracy, including a reduction in error of up to 96x in real data.

 

Dear Reviewer,

We greatly appreciate your thorough review of our paper. Based on your suggestions, we have made corrections and revisions. We are excited to introduce a new idea in our paper, which is based on two concepts. Firstly, we propose utilizing photogrammetric methods for enhancing LiDAR data. Secondly, we introduce the concept of preserving the relationship between LiDAR points and their projection onto specific images. This novel approach allows for the development of a new processing workflow for LiDAR points.

Throughout our research, we have been developing a dedicated research tool specifically designed to test these new ideas and methods in both quantitative and qualitative ways. As we conclude this phase of our long-term research, we are eager to present our results.

While our solution is tailored to LiDAR data, it integrates closely with photogrammetric techniques, representing another case of LiDAR and photogrammetry integration. Externally, we employ image matching and bundle adjustment for determining new exterior orientation parameters (EOPs). The statistics used to evaluate the final discrepancies include RMSE and MAX values.

We have carefully reviewed and incorporated all the detailed suggestions into the text.

Best regards,

Authors

 

1. General comments.

Comments 1:

Improvement 1: Reproducibility.

Scientific validity and impact are increased when other researchers can replicate and build on the work. I suggest:

• Publish synthetic and semi-synthetic test datasets used in the study, ideally in a public repository (e.g., Zenodo, GitHub, or institutional archive). Include Point clouds in standard formats (e.g., LAS/LAZ, ASCII), Associated GCP and CHP files (XYZ, intensity/RGB) and JSON and XML configurations used in the processing workflow.
• Release the PyLiGram research tool or at least a basic version or script with example usage.
• Provide a minimal working example (MWE) showing how to process a simple dataset through the full pipeline (steps 1–4), ideally with output metrics (RMSE/Max).
• If open-source release is not possible, consider including a supplemental appendix with detailed pseudocode and data samples.

Response 1:

1. The synthetic and semi-synthetic datasets are available: https://fotogrametria.agh.edu.pl/pyligram
2. The PyLiGram release is also available there.
3. MWE dataset is also available. The synthetic dataset testing procedure is described in attached instruction_MWE_Synthetic_dataset.pdf

 

Comments 2:

Improvement 2: Clarity

Dense text can obscure understanding, particularly for readers less familiar with lidargrammetric ideas.

I suggest to:

• Shorten or split overly long paragraphs, particularly in the Introduction and Discussion.
• Use bullet points or numbered lists where appropriate (e.g., outlining steps in data processing or matching).
• Consider including a conceptual overview or summary flowchart

Response 2:

The entirety of the document was subjected to revision, with lengthy paragraphs being reduced in length and divided into shorter units. The application of numeration and bullet points occurred at all levels. The summary flowchart is divided into two phases. Following the incorporation of the more profound division of the text, the clarity of the content should be evident. It appears that the flowchart is not required in this instance.

Comments 3:

Improvement 3: Figures

Visuals should enhance understanding even without direct reference to the main text. I suggest that each figure should have a descriptive caption that summarizes what the figure shows, such as type of data (synthetic, semi-synthetic, real) and purpose of the visual (e.g., data layout, GCP distribution, registration results).

In addition, you should add visual annotations (legends, axis labels, scale bars, or markers for GCP/CHP locations). For example, in Figures 3–5 (dataset descriptions), show strip numbers and orientation.

Please use color-coding consistently (e.g., GCPs in cyan, CHPs in orange), and always define them in the legend.

Response 3:

As illustrated in Figures 1, 2 and 7, the summaries of the figures were expanded to provide an explanation of the processes. The generation of figures 3, 4, 5 and 8 was achieved through the utilisation of a scale bar (or XYZ axes) and a legend. The colours of the GCPs and CHPs were unified.

 

Comments 4:

Improvement 4: Performance Summary Table

Consider adding a single table summarizing performance before the Discussion section (e.g., %RMSE reduction, across datasets and benchmarks). A summary table improves digestibility and comparison.

Response 4: As illustrated in Table The description of Metashape usage is extended. The software was utilised in its default configuration. The text includes a detailed description of specific settings of accuracy, accompanied by the inclusion of additional figures. It is evident that no manual measurement was conducted, a fact that is duly noted in the text. The radiometric change of the images is described. It is not the intention of the author to extend the length of the paper any further. The radiometric enhancement through the integration of deep learning methodologies and/or adaptive interpolation techniques will be a subject to be explored in future research endeavours. The additional paragraph was incorporated into the Future Work chapter, accompanied by Fig. 6 of Metashape's Reference Settings window screenshot.

 

1. Specific comments:

 

Comments 6:

Improvement 6 – Abstract

The abstract is dense, consider simplifying the explanation of the method for accessibility.

Response 6:

The method is outlined in the abstract's opening paragraph, with subsequent sections detailing its testing and the outcomes. The conclusion provides information regarding the reference of our results to a globally recognised solution.

 

Comments 7:

Improvement 7 - Section 2

The comparison with commercial software is strong; consider tabulating key differences for reader clarity.  A table contrasting input requirements, trajectory use, flexibility, and target use cases (e.g., TerraMatch vs. PyLiGram) would improve reader understanding.

Response 7:  

Tab. 1 was incorporated in order to facilitate a comparison of the TerraMatch and PyLiGram methods.

 

Comments 8:

Improvement 8 - Section on Synthetic Data: 

Consider clearly stating why orthophotos were used and how RGB was derived. For example, explain if orthophotos were used only for visualization or to simulate real-world reflectance in lidargram generation.

Response 8:

The extension of the description is inserted in the paragraph.

 

Comments 9:

Improvement 9 - Real Data Section

Good explanation of intensity-based matching, but contrast enhancement methods could be described more explicitly.  Indicate specific changes (e.g., contrast stretch parameters or gamma values) and whether they were automated or manual.

Response 9:

The short explanation was added in the second paragraph of 3.3.

 

Comments 10:

Improvement 10 - Table 8

Clarify the basis of the “trajectory approximation” comparison. Does the increased number of lidargrams equate to finer spatial resolution or something else?. Explicitly state how generating more lidargrams improves change (e.g., denser camera positions yielding more accurate EOP interpolation).

Response 10:

The additional explanation was incorporated into 3.5.

 

Comments 11:

Improvement 11 - Discussion

The paragraph comparing PyLiGram to TerraMatch is fine. However, a more direct statement on where PyLiGram outperforms (non-rigid, trajectory-less scenarios) should be emphasized. Consider summarizing PyLiGram's niche, such as "ideal for legacy or GNSS-denied data where trajectory is missing or unreliable."

Response 11:

The final paragraph of Section 4.5 has undergone expansion.

 

1. Ethics, references and data availability

Ethics and Conflicts: No ethical issues.

References: Enough, recent (many from past 5–10 years), and relevant. Some legacy references (e.g., early lidargrammetry origins) are necessary for historical context. Self-citations are not excessive.

 

Comments 12

Improvement 12 (related to 1) Data Availability

Marked “Not applicable.” This is a weakness. The authors should at least share synthetic and semi-synthetic data for replication.

Response 12:

Link with the release, MWE and other data is added: https://fotogrametria.agh.edu.pl/pyligram

 

Reviewer 3 Report

Comments and Suggestions for Authors

This paper presents an approach for processing LiDAR data using image and photogrammetric techniques and conducts extensive experimental verification.

However, I have the following questions:

1. The presentation of experimental results in the paper is too limited, relying solely on tables. It is recommended that the author incorporate charts or graphs to enhance the presentation.
2. Why are commas used as decimal separators in the table values?
3. Figures 1, 2 and 6 in the paper are unprofessional and visually unappealing; they must be redrawn.
4. The references in the paper are not up-to-date and should be replaced with more recent ones.
5. Part 5 is overly redundant and should be condensed.
6. The paper has significant language issues and requires thorough proofreading and editing. The formatting of the paper also needs to be reconsidered.

Author Response

Dear Reviewer!

We are extremely grateful for your comprehensive evaluation. It is our conviction that the aforementioned text has been improved by the implementation of the suggestions and comments that you so kindly provided.

We have responded to all the comments below.

All the comments were very helpful and each one has been taken into account, resulting in additional research and changes to the text.

Best regards,

Authors

 

This paper presents an approach for processing LiDAR data using image and photogrammetric techniques and conducts extensive experimental verification.

However, I have the following questions:

Comments 1:

1. The presentation of experimental results in the paper is too limited, relying solely on tables. It is recommended that the author incorporate charts or graphs to enhance the presentation.

Response 1:

Adding charts and graphs would extend already long paper. The tables were corrected.

 

Comments 2:

1. Why are commas used as decimal separators in the table values?

Response 2:

Corrected.

 

Comments 3:

1. Figures 1, 2 and 6 in the paper are unprofessional and visually unappealing; they must be redrawn.

Response 3:

Checked.

 

Comments 4:

1. The references in the paper are not up-to-date and should be replaced with more recent ones.

Response 4:

Checked. Corrected.

 

Comments 5:

1. Part 5 is overly redundant and should be condensed.

Response 5:

The part 5 was divided into sub-chapters and its redundancy was limited.

 

Comments 6:

1. The paper has significant language issues and requires thorough proofreading and editing. The formatting of the paper also needs to be reconsidered.

Response 6:

Language and formatting were revised.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,

Most issues have been satisfactorily addressed. You have implemented nearly all key reviewer suggestions, particularly regarding reproducibility, clarity, data access, and methodological transparency.

There are two remaining areas for improvement (not required but encouraged):

• Performance Summary Table: Still recommended to synthesize performance metrics and help with cross-dataset comparisons.
• Abstract Simplification: One or two sentences simplifying the core method for broader accessibility would enhance the paper’s reach.

Below is a point-by-point evaluation of the authors’ revisions.

 

General Comments

Comment 1:

Sufficient. No further action required.

Authors have released the synthetic and semi-synthetic datasets, the PyLiGram tool, and a minimal working example (MWE) with documentation.

 

Comment 2:

Sufficient

Long paragraphs were revised and broken down, bullet points and structured formatting were added.

 

Comment 3:

Sufficient. No further action required.

Authors revised captions, unified color codes for GCPs/CHPs, and added legends, scale bars, and axis information.

 

Comment 4:

It has been partially addressed.

Authors acknowledge the suggestion but did not add a dedicated summary table (as recommended before the Discussion section). Instead, detailed tables remain distributed across sections.

Recommendation: A single summary table (before the Discussion or as an appendix) that synthesizes %RMSE reduction across synthetic, semi-synthetic, and real datasets and compares PyLiGram with CloudCompare/TerraMatch would still greatly improve readability and impact.

 

Comment 5:

Sufficient. No further action required.

The text now includes more information on Metashape’s configuration, settings, and limitations. Enhancements and plans for deep learning-based improvement are noted in the Future Work section.

 

Specific Comments

Comment 6:

Partially addressed

Authors state the method is outlined early and results are referenced later. However, the abstract remains dense, with technical terms (e.g., ULPIs, EOPs) appearing early without simplification.

Recommendation: A slightly more accessible abstract would still help general readers. One or two sentences simplifying the core idea in plain terms would improve outreach.

 

Comment 7:

Sufficient. No further action required.

Table 1 was added to compare TerraMatch and PyLiGram methods.

 

Comment 8:

Sufficient. No further action required.

Clarification was added to explain the role of orthophotos in deriving RGB and simulating real reflectance.

 

Comment 9:

Sufficient. No further action required.

A short explanation of radiometric adjustments (contrast, gamma) was added to the real data section.

 

Comment 10:

Sufficient. No further action required.

Authors clarified how denser lidargram sampling (via more images) improves EOP estimation and results.

 

Comment 11:

Sufficient. No further action required.

The discussion now explicitly emphasizes PyLiGram’s value in non-rigid, trajectory-less or GNSS-denied environments.

 

Ethics, References, and Data Availability

Comment 12: Data Availability

Sufficient. No further action required.

Authors added a data/code link: https://fotogrametria.agh.edu.pl/pyligram

 

Author Response

Comment 4:

Authors acknowledge the suggestion but did not add a dedicated summary table (as recommended before the Discussion section). Instead, detailed tables remain distributed across sections.

Recommendation: A single summary table (before the Discussion or as an appendix) that synthesizes %RMSE reduction across synthetic, semi-synthetic, and real datasets and compares PyLiGram with CloudCompare/TerraMatch would still greatly improve readability and impact.

Response 4:

We agreed with the comment. The table was added. 

Specific Comments

Comment 6:

Authors state the method is outlined early and results are referenced later. However, the abstract remains dense, with technical terms (e.g., ULPIs, EOPs) appearing early without simplification.

Recommendation: A slightly more accessible abstract would still help general readers. One or two sentences simplifying the core idea in plain terms would improve outreach.

Response 6:

We agreed with the comment. Two first sentences were added to the abstract.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The author has made detailed revisions to the issues raised previously, and the article has improved to some extent. However, it is recommended that the author shorten the introduction, as the current version contains too much content.

Comments on the Quality of English Language

 The English could be improved to more clearly.

Author Response

Comment:

The author has made detailed revisions to the issues raised previously, and the article has improved to some extent. However, it is recommended that the author shorten the introduction, as the current version contains too much content.

Response:

The introduction was divided into subparagraphs. And we believe its capacity is adequate to the subject importance and size.

Author Response File: Author Response.docx