3D Reconstruction from 2D Plans Exemplified by Bridge Structures

Round 1

Reviewer 1 Report

The manuscript at hand considers a highly practical problem: 3D Reconstruction from 2D Plans of historic bridges. While they did some interesting work, there are also several flaws which could be fixed.

1.line 74. if the authors could add a comment  on why the Douglas-Peucker algorithm is used, and how it compares to other algorithms

2.Relevant references may need to be added

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

The text presents a methodology for the extraction of vertices (edges) of bridge drawings for their vectorisation and reconstruction in 3D space. The subject is interesting, but there are some aspects that should be explored in more detail.

First of all, the introduction is extremely concise, some more explanation should be given in order to contextualise the research, and some assertive phrases such as "Printed or drawn maps and plans are usually digitized at considerable expense", describing a process that tends to remain in digital space (the drawing reported start from digital space). In the same way, the state of the art also needs to be explained. The authors, however, only deal with the topic of digital 'automatic translation' of drawings, without taking into account that the examples of drawings are often the basis of 3D models or extracted directly from them (BIM) and are not design sketches. Furthermore, if we focus on the plan representation part of these artefacts, there are numerous examples of bridges drawn (especially from the last century) but never represented in 3D space, with increasing levels of complexity in terms of both drawings and geometric shapes represented. 

As far as the research aspect is concerned, the idea is interesting. There is a logical leap, which should be better explained, between the study of sections and the extraction of slices with their development in space (especially when this is curved). 

A final observation concerns the comparative study with the survey of real objects versus remodelled ones. The lack of this step weakens the whole process, bearing in mind that the authors speak of as-built by referring to their 3D models, which is deeply uncorrect. The as-built should refer to the existing 3D object, whereas this one definitely has differences to the one that was actually built the first time (for construction deformation and time component). So it is almost uncorrect to use the terms "monitoring" using these 3D models. The lack of distance between the ideal model and the real one in this sense is crucial. Finally, it is not clear why the authors go into so much detail about the point cloud extracted from the reconstructed model, as they do not currently compare it with any reality-based point cloud. It should be better explained, because in geometric terms a shape (especially if it is regular and not spatially curved) resampled at different levels of detail remains the same, even if this implies a different time from the point of view of data processing. 

The text is clear and well readable. 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors proposed a method to obtain three-dimensional (3D) geometry of the bridge structured from two-dimensional (2D) diagrams. The method involves identifying the corner points from a 2D plan and estimating the 3D geometry. The 3D geometries of the various objects are then fused together. The authors used the proposed method to estimate the 3D geometry of three bridge structures. The article has sufficient novelty. The authors should address the following queries:

1.    The proposed method assumes that the 2D plan is available which is not the case with a laser scanner. This is a limitation of the proposed method. The authors should discuss it.

2.    The authors should proofread the article there are a few typographical errors.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The problems raised have been corrected and it is recommended to publish.

Author Response

Thank you for the recommendation.

Reviewer 2 Report

The authors responded to the reviewer's comments in a timely manner. The article therefore no longer has any logical leaps or gaps in terms of methodological development or explanation. It remains that the authors did not go into the points highlighted in depth, but merely gave a punctual response, which means that the level of the article in general terms remains unchanged. The article is publishable, although a significative improvement and deepening is possible, which perhaps the authors will reserve for future research.

Author Response

Thank you for your recommendations.
A further research into the consideration of the difference between the captured as-built data and the generated 3D point cloud from the plans is currently ongoing. The results will be published in a later publications as mentioned.