Automatic Forest DBH Measurement Based on Structure from Motion Photogrammetry

Round 1

Reviewer 1 Report

Dear Authors,

I have finished my review on your paper. I think that it could add to our knowledge (+) but it still needs improvement, mainly in terms of terminology and phrasing (-). Please refer to my comments and suggestions given in the following.

Best regards,

Rev.

PAPER IN GENERAL:

• The topic is highly important and the approach of the authors could be sound; however, I found this paper very difficult to read and understand. This is mainly due to the language quality which needs to be significantly improved in all the sections of the manuscript. I would say that the first thing to do would be to send the paper to a professional translation service with the goal of improving the phrasing.
• Other things to consider would be: improving the introduction mainly in terms of phrasing; building a better materials and methods section, in a structured way to help readership in understanding the approach. The paper is about diameter estimation and not about the software used. Although the software part is important, it should not be the main topic herein. Perhaps a better way to understand how the software works would be that of providing a link to a downloadable version.

 

SPECIFIC COMMENTS:

• I would not give here the exact lines where the paper should be improved in terms of phrasings due to the fact that it requires improvements in most of the parts. The Authors are encouraged to search for professional assistance for translating their manuscript;
• Do not mix the text and information in the sections and try to put each part of the text where it should belong. For instance, lines 386-391 should belong to materials and methods;
• Some figures are not referenced in the text (for example Figure 8 and Figure 9). They should be referenced and some explanations should be given about the data shown in them. If the method of Bland & Altman is not used, Figure 9 would benefit from adding some more lines which should be parallel to the diagonal one at 1, 2 and 3 cm on both sides and in different colors. This would help understanding the deviation of data from an absolute match.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Does the camera use to have GPS or other positioning systems? If not, it may have affected the quality of the generated point cloud.
Explain why they did not use ground control points (GCPs). Could this have influenced the results? Explain.

Will the technique work in forests with low (< 4 cm) DBH values? Explain.
And sloping terrain, the proposed technique will work, explain.

Resume
Line 25 - Inform the mean relative error (3.16 cm).

Introduction
Line 68 - Can you mention an open-sourcing SfM is used?
Table 1 - Remove No. Insert N.Inform the max values. And min. DBH?
Line 150 - Are all trunks symmetrical? Different from what was described above.
Line 162 - “The same part of the target trees’ trunk was contained in the five images”. How was this possible? Explain. Has the camera's vertical/horizontal axis been rotated? If yes, what angle?
Line 163 - “..baselines between the five images must be large enough “. What is the distance used in the study?
Line 169 - Was the measurement error of DBHs estimated? If not, explain why.
Line 175 - Show a picture with the marks.

Figure 3 - I suggest inserting a part “C”, containing the five photos taken in the field of an individual.
Line 243 - Not set to 1.3 m height? Exact value? Explain. How to ensure that this height was the exact measurement in the field?
Line 271 - Why, explain.
Line 273 - “we roughly pre-estimate DBH value”, how?

Results

Line 337 and 338 - How was this estimated? It was compared with other methods, explain.

Line 370 - PhtoScan - Make the citation of the program.
That sentence doesn't make much sense. Neither commented on the methodology. I suggest removing it.

Line 377 - “Detection rates”. How was this estimated? Inform.

Lines 384-391 - Remove for methodology.

Lines 394-395 - -0.09 cm and 0.83 cm. Where did these values ​​come from?

discussion
 Lines 411 - 412 - “the efficiency, and accuracy of DBH measurement by” . Can you really say that? RMSE values ​​were lower when compared to other studies.
Line 446 - It makes no sense to compare absolute value. In relative terms, the RMSE% was lower than that observed in

Conclusion
Every conclusion must be redone. Conclude on the results found.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The manuscript “Automatic forest DBH measurement based on Structure from Motion Photogrammetry” authored by Qiang Gao and Jiangming Kan presents a novel approach to realize accurate and efficient automatic measurement based on SfM photogrammetry and increase image acquisition efficiency.

The manuscript is relatively novel and has contribution to community since it do remain largely limitations that commercial software and redundant images consume large amounts of money and time. It is acceptable for publication but I suggest the authors to solve the issues below.

First, all the images were taken by camera in automatic mode held by hand in this research. To me, the authors should present more detail on data acquisition. Whether the data derived from these images would suffer uncertainties from researchers? What is the height of the camera from the tree and the ground when collecting images? How should the route be designed to get a sequence of overlapping images?

Second, will your software Auto-DBH be freely available? This is encouraged and might be needed since you argue in this paper that commercial software cost much. 

Minor comments:
Line 121: there is a lack of [cm] after [40.1]
Line 141-143: What is the specific process to take images with a camera held by hand? Whether errors from researchers would arise?
Line 304: there is a lack of blank between [plot] and [4]
Line 370: [PhotoScan] rather than [PhtoScan]
Line 385-391: This introduction should be presented in [Materials and Methods]
Line 493-514: The conclusions could be divided into three paragraphs.

Figure 1: It would be better that the color of the scale bar is adjusted.
Figure 2: Inconsistent font size of title 
Figure 4: There is a lack of blank after [&]
Figure 9: The meaning of blue lines and points should be presented in the title. 
Table 1: Check that if Latin names of main species are correctly presented. Italic? Whether the first letter should be capitalized under column [Difiiculty Level]?

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

Review of Gao & Kan: Forest DBH

 

This paper provides a most useful rapid method of determining forest inventory, though a number of clarifications are needed as outlined below.

 

Line 10-11: “…is an essential but laborious task in traditional forest inventory..”

Line 27: “..is relatively fast and on average takes only 2 seconds to estimate the DBH of a tree, which is much more rapid than direct physical measurements of tree trunk diameter.”

 

Lines 68-69:  Change to “..in the data collection stage which can be saved if open-sourced SfM is used.  Therefore, SfM photogrammetry is a relatively economical….”

Line 100: Change to “The following work has been…”

 

Table 1: No. Trees not “No. Tree”  Also in Table 2

 

Line 128: “The level of difficulty of SfM reconstruction…”

Line 139: “Figure 2. Large variations in tree trunks in plot 3.  …..”

Line 146: “The maximum resolution is 3648x2736…”

Line 150: “..that reduces the number of images that need to be collected and so shortens the time needed for assessments.”

 

Line 165-66: this is confusing. Do you mean “The entire data collection program would be like scanning the whole forest tree by tree using a camera.”

Line 169: “Ground truth values of DBHs…”

Figure 3: You need to clarify how 5 images taken from the 3 positions.  Something like: “..directions and positions of images taken along the photographing path are indicated by the black dashed lines.  Each tree had three positions for photography: right, middle (where bottom and top pictures were also taken) and left, giving five images per tree.  (b) The red parallelograms and arrows illustrate the positions and directions of the five images per tree.”

Line 224: “The DBH extraction pipeline…”

Line 230: “The dense points cloud was used…”

Line 287: “….plot 4, we reduced the distance…”

 

Line 329-334: “Auto DBH helps improve….”  Is this better in discussion?

Line 375 after DBH estimation of sampled trees rest of trees estimated in 1-4 minutes!

 

Line 385-388: Define bias, absolute bias, RMSE in words as well as formulae:

The accuracy of SfM DBH in the different plots are shown in Table 3 where values are provided for Bias: the difference between estimated and measured DBH; for Absolute bias: the Absolute value of the difference between estimated and measured DBH; Mean relative error:  absolute bias as a percent of the circumference of the tree trunk; Root mean square error (RMSE) as a cm value as well as a percent of the circumference of the tree trunk (Relative RMSE).

Line 396: “..of the two plots are 1.41 and 1.18 cm, close to the…”

Line 404: “Our method works well for trees with relatively round trunks but is not suitable for trees with trunks that deviate substantially from being round.”

Line 409: “The main objective of this study was to analyze the application of SfM…”

 

Line 453-457: Replace lines 453-457 with the following to make your main conclusions clearer: “If trees in a forest have relatively round trunks, DBH can be accurately estimated using images from one side of the tree, saving substantial time and effort, but the method is unsuitable for areas where tree trunks have a substantial deviation from roundness.”

Line 466: “Therefore many useless points…”

Line 471: “Even though we took some measures….

Author Response

Please see the attachment.

Author Response File: Author Response.docx

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The manuscript entitled “Automatic forest DBH measurement based on Structure from Motion” presents a cost-effective and automatic method measuring BDH of trees using SFM.  

The proposed method is not new and actually an integration of SFM and DBH extraction steps are considered the main outcome the research. The manuscript is not written well and it is full of writing mistakes which makes the reading difficult.

It terms of the concept of the research, it can be found that photogrammetry solution is not mentioned and only sfm is considered as the whole solution. The reference of the photogrammetry has been neglected because that sfm methods cannot provide metric measurements without control points.  

Some of mistakes are as follow:

Line 13: researchs should be changed to "research works"

Line 19, 27, 69: sentence starts with "And"

Line 24: relative absolute means???

Line 42: DHB or DBH?

Line 62-64: it is not readable

Line 85: in formal writing, "As far as I know", it should be rewritten as "based on the authors knowledge"

Line 150: diameter tape??

Line 170: building mesh should be replaced by "generate point cloud"

Line 171: cloud point -> point cloud

….

Line 425: rewrite the sentence please.

 

Reviewer 2 Report

please find in the attachment

Comments for author File: Comments.pdf