Integrated UAV-Based Real-Time Mapping for Security Applications
Round 1
Reviewer 1 Report
Overall, the writing and presentation of the paper is acceptable, the technical aspects are mostly well described and accurate, and the results will interest for remote sensing scientists and esearchers and managers interested in UAV industries. However, there are several issues that should be addressed prior to publication as listed as follows:
First, I found it is uncommon to write manuscript where the result and discussion sections are located/described early than the materials and methods section. Thus, I suggest the authors to rearrange the sections of this by writing the materials and methods section first then followed by the results and discussions sections, accordingly
Second, although the novelties of technology are well presented in this paper, but I was wondering how or whether this Integrated UAV-Based Real-Time Mapping can be used for monitoring wide area of disaster e.g. forest fires, especially in developing countries where 4G technology is rarely available.
Third, how much the cost to build the UAV system?
Author Response
Thank you very much for your review and your comments!
Regarding the order of sections: we strongly followed the MDPI latex template, which provides the result section before methodology section. We missed the IJGI author instructions, which provide a different order of these sections. We will fix this issue and consider your further comments.
Regarding your second note: 4G is only one option, we also examine the integration of a proprietary IP based downlink. Furthermore, the presented demonstration system provides a non-stop operation time of up to 90 minutes per battery charge with a typical cruise speed of about 80km/h - which allows to capture even larger areas. It was tested already for scenarios mapping areas in the range of up to few square kilometres. However, forest fires were not yet a considered application, we focus on disaster affected areas after e.g. earth quakes, tsunamis or floods.
To answer your last question: the carrier itself with the mentioned data downlink is in the range of 80.000 Euros, the hardware components of the camera system are in the range of 35.000 Euros. Please keep in mind, this is a development and demonstration system, so focus is on developing and demonstration of technology for security applications and customers.
Reviewer 2 Report
The authors presented clearly a UAV-based real-time processing pipeline and did a great job demonstrating the details of this pipeline system. I would suggest the authors reconsider the paper's title regarding "security applications" since the paper did not include any aspect of security applications. I would also like to suggest the authors restructure the paper's outline by bringing up the materials and methodologies earlier in the paper, earlier than 2. Results, because this will give the readers a better idea of the components of this system. Other than these two points, I believe this paper has great merit and should be published as soon as possible.
Author Response
Thank you very much for your review and your comments!
Regarding the order of sections: we strongly followed the MDPI latex template, which provides the result sections before methodology section. We missed the IJGI instruction, which provide a different order of these sections. We will fix this issue and consider your further comment regarding "security applications" in the title.
Reviewer 3 Report
The study aim is to present the Integrated UAV-Based Real-Time Mapping for Security Applications. The manuscript is presented in a clear and nice way. However, there are a few suggestions.
1. In the introduction, some citation missing in line 21 – 26
2. Why “Materials and Methods” in section 4. Generally, it added to section 2
3. What is the limitation of this method? Better to discuss it
4. All figures need to arrange as Figure 2 (a); (b)
5. Better to add conclusion section
Author Response
Thank you very much for your review and your comments!
We will consider all your issues and upload a revised version as soon as possible!
Regarding the order of sections: we strongly followed the MDPI latex template, which provides the result sections before methodology section. We missed the IJGI instruction, which provide a different order of these sections. Of course we will fix this major issue.
Reviewer 4 Report
Review report for ijgi-457242
Summary
The paper proposed a real-time mapping model for UAV image applications. The topic is interesting to readers and suitable to the journal. However, writing structure of the paper and incomplete paper are not suitable to an international journal. I listed my suggestions for the authors to improve their research.
Major issues
The writing structure is generally arranged in this sequence: Introduction, Study Area and Methods or Materials and Methods, Results, Discussion, Conclusion, Acknowledgement, References.
More minor issues
L298, revise “uAV” into UAV here.
1.L93-95, please explain limitations of the proposed UAV system, e.g. weight, flying height and others. 2.In Fig. 12, please adjust the texts position in the legend.Opinion
In general, the manuscript is acceptable to the journal after major issues revised. I suggest rejecting this incomplete version but encouraging the authors to re-submit the manuscript after major revision.
Author Response
Thank you very much for your review and your comments!
Regarding the order of sections: we strongly followed the MDPI latex template, which provides the result sections before methodology section. We missed the IJGI instruction, which provide a different order of these sections. We will fix this issue and consider your further mentioned minor issues.
Reviewer 5 Report
This paper presents a recently developed system for the quick generation and transmittion of aerial maps. The proposed approach is based on the use of the TAC clipping method previously proposed by some of the authors.
The proposed method has the clear advantage of being fast. In order to rpoduce the map it is required the availability of some prior information on the geometry of the scene (a DEM). The DEM exploited by the authors has a quite low resolution (90m x 90m, with a vertical error up to 15m): I suggest to the authors to provide an analysis of the influence of DEM error on the system performance.
Furthermore, since the fundamental step in such map generation is the use of the TAC clipping, I suggest to the authors to provide a short summary of the method in order to ease the comprehension of the map generation method to the readers.
In my opinion, a more detailed validation of the mapping accuracy of the method should also be provided by the authors.
Despite details about the TAC clipping method has not been provided here, according to the short description done by the authors (projective mapping) it seems like lens distortion has not been corrected. I guess this should have a negative effect on the mapping error.
pag 11, line 284: it is not clear how the TAC projection has a computational complexity independent of the number of pixels. This shall be true only if the image pixels are actually not remapped to their "correct" coordinates (in such case the computational complexity is obvsiously linear with respect to the numer of pixels), but they are only associated to the computed homography. Actually, even just setting the association between a subset of the original image to be used in the map along with the computed homography is an operation with linear complexity with respect to the number of involved pixels. The authors should clarify such aspect providing a more clear explanation (for instance, the computational time may be linear in the number of pixels but usually negligible with respect to the other processing steps given the simple operations involved).
Finally, I have some other minor suggestions:
- pag 1, last line "a remote operator" instead of "an..."
- pag 2, "spatial resolution" instead of "spacial..."
- pag 3, line 69: how long was the experiment?
- pag 4, it is not so clear to me why inserting a Discussion section at this point of the paper instead than at the end...
- pag 12, equation afer line 291: it would be better to provide an upper bound than a lower bound , if possible.
Author Response
Dear Sir or Madam,
thank you for your comprehensive comments and recommendations of your review! Unfortunately you obviously did not received the last version of our draft, which was uploaded on 15th of march, 2019. However, we discussed and took into account all of your comments. These are the main changes of the revised version:
- chapter 1, line 32: fixed typo "an remote operator" > "a remote operator"
- chapter 1, line 45: fixed typo "spacial" -> "spatial"
- chapter 2, line 59-61: added statement regarding distortion correction
- chapter 2.2, lines: 87-89: added qualitative statement regarding usage of DEM with higher resolution and its impact on accuracy
- chapter 2.8, line 231-233: added statement about experienced additional loads (compared to lower boundary)
- chapter 3, line 286-287: added experiment time
Additional comments on your feedback:
- quality of DEM, resolution & error: we added a qualitative statement in section 2.2;
- section order (position of Discussion section): as already mentioned, the section order was already fixed in the version uploaded on 15th of march. The wrong sequence of section was based on a misunderstanding of the original MDPI template (which proposes the discussion section to be placed in the beginning).
- lower & upper bound of computational effort: we added a statement of practical experiences. There is no reasonable (in mathematical terms) upper bound for the computational load. The lower bound may be exceeded due to the non-linear movements of the aircraft, e.g. movements around its vertical axis or during a turning manoeuvre. Theoretical limit meets the number of all captured pixels per second, which is not relevant for practical applications.
- short summary of TAC: a short summary and main principle of TAC is described in section 2.2 Terrain Aware Image Clipping (TAC). Figure 3 illustrates a sample result of TAC. TAC is described in detail in https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/IV-1/61/2018/isprs-annals-IV-1-61-2018.pdf, as referenced in this section.
- mapping accuracy: chapter 3, lines 291-296 roughly address investigations of mapping accuracy. Reference 39 ("A Real-Time Remote Sensing System for Maritime Security Applications", CEAS Space Journal 2018.) given as additional reference for this issue.
- distortion correction: it is correct, we do not apply distortion correction in the presented chain in order to save computational efforts, in particular on small embedded systems. There are mainly two reasons that enable quite good results without an undistortion stage: firstly, application of optics with a low distortion and secondly, due to application of TAC only inner image areas are used, which are generally less affected by distortion effects than the outer areas. However, the approach may be extended to enable distortion correction support, but this has to be implemented at at least two stages: 1. TAC stage: overlapping analysis has to take into account the distortion model for correct mapping between world and sensor coordinate system and 2. after radiometric correction stage, a new undistorted image has to be rendered with application of distortion model.
- computational complexity of TAC: Actually, TAC has constant complexity, as described here: https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/IV-1/61/2018/isprs-annals-IV-1-61-2018.pdf. The algorithm consists of eight steps, each single step with constant complexity. Sensor's resolution is only relevant for forth and backward mapping between sensor and world coordinate system. There is no loop or propagation along any number of pixels, the algorithm would even work in a metric sensor coordinate system without any knowledge about sensor's number of pixels.
Please find attached the revised version of our contribution.
Round 2
Reviewer 4 Report
A conclusion section needed for a formal journal paper.
Author Response
Dear Sir or Madam,
thank you very much for your remark regarding the missing conclusions chapter! The authors instruction site of the IJGI journal https://0-www-mdpi-com.brum.beds.ac.uk/journal/ijgi/instructions lists the conclusions chapter to be not mandatory, but applicable, if the discussion chapter is too long or complex. We discussed this issue and we think, the discussion chapter of our contribution covers all relevant topics from discussion of results, the findings and their implications, its interpretation and relevance for applications in practice, and we outline some potential future research directions. Furthermore, the discussion chapter is not too long, but an extra conclusion chapter would be comparatively short.
So we would like to ask you, if we can leave the conclusions to be included in the discussion chapter? Otherwise we would split the discussion chapter into a discussion and a conclusions chapter.
Thanks in advance!
Reviewer 5 Report
Authors have answered to my questions (probably part of the changes that I suggested have already been applied in a more recent version of the paper with respect to the one that I received).
Author Response
Thank you very much for your review!
