Polarization Orientation Method Based on Remote Sensing Image in Cloudy Weather
Round 1
Reviewer 1 Report
This paper based on the previous research on the bionic compound eye, proposes an atmospheric polarization state measurement platform based on a multi-channel camera array with different polarization degrees.
The paper proposes an alternative approach of the mainstream GPS navigation, geomagnetic navigation, inertial navigation and other technologies.
Despite the coherence of paper flow reading, in my opinion some flaws are presented, namely in section 2.3 where proposed model is not suficiently justified and must be improved, in order to robust the paper conclusions. The equations (8) to (16) must be justified, taking into account the generality of the process.
In my opinion this explanation is central in order to may assess the paper results.
Major recomendations:
1- Section 2.3 must be justified with clear references and justification of the model.
Minor recomendations:
The quality of the figures must be improved: Figs. 2,4,5,6,7, 9, 10,11,12 .
Abstract: NMS is not defined;
Section 2.2.: RPN is not refined. A bibliography reference should be interesting
Line 197 Roi Pooling algorithm must be refered in bibliography
Line 198; The RCNN method must be explainded and Refered in bibliography;
Line 204; The NMS method must be explainded and Refered in bibliography;
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Dear Authors:
There are several serious concerns which need to be addressed:
(1) There should be more detailed explanation on the experimental data. How many data did you collect? What is the size and resolution of the remote sensing data?
(2) The data used for experimental studies are very small. The weather data are only collected in September 2022. Some of the related experimental results and discussion about the robustness and navigation accuracy of the polarization navigation in different weather conditions (or the different seasons) should be included and compared.
(3) The line 148 “I Solar azimuth acquisition in the carrier coordinate system” is rather cryptic. The letter “I” is typo? Please rewrite.
(4) The proposed framework extracts the feature map based on the DOP image using the threshold segmentation method. Which threshold segmentation method did you use in this study? In equation (4), ??ℎ represents the feature threshold. How can you set up the parameter ??ℎ ? The author should cite the relative references of the threshold segmentation method.
(5) The region of interest of the feature map are extracted based on SRPN and RRois. There should be more detailed discussions on the choice of SRPN approachs. Have the authors tried different ROI extraction methods, and this one turned out to be better? Authors should cite the relative references of SRPN and RRoi methods. In equation (5), what is the meaning of IOU? Please also explain the figure 4 in a detail.
(6) The line 184 “B is the real label window provided by the training set”. Please elaborately describe training set that you use in this study. In equation (6), ?∗ is the training parameter, how can you set up this training parameter? ?∗(?) is the objective function, please elaborately describe this objective function.
(7) The paper includes several abbreviations. Please give the full name as a remark when the abbreviation appears for the first time.
(8) Correct all Equations in the manuscript to the appropriate format. Adjust the font of the variables and functions represented by the mathematical formulas.
(9) There are several grammatical errors in this manuscript. Please do a careful proofread and fix them all.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
In this manuscript authors propose a polarization orientation method based on the solar meridian measurement. The introduction has a nice touch to it. I quite liked it with the little facts about insects. Although the rest of the paper can improve from additional reference citations at least in the parts 2.1-2.3. The topic of the paper is well matched to the journal so I can recommend to accept it after addressing some points.
Line 64. Why do authors capitalize the angle in “polarization Angle”. I think it’s unnecessary. Especially since degree of polarization wasn’t capitalized.
Line 113. The abbreviations for SPRN, RRois and RCNN should be given not only in the abstract, but in the manuscript text too.
Line 127 Usually Stokes vector is determined by 4 parameters, seems like authors excluded the last parameter V (S3) which corresponds to chirality of the light polarization. I think it needs to be mentioned that authors consider only the linearly polarized part of the radiation.
Figure 2 is unreadable. I suggest authors to rework it, since shows the proposed algorithm. Also increase font size and provide shaper images.
Figure 3 Same as with Figure 2.
I suggest authors provide some references in the part 2.3 from where the equations 8–16 were taken from.
Figure 9 is just too much. Even at 200% scaling it’s barely recognizable. I suggest authors reduce it to a few specific sets of images to show the rotation between acquisition images.
Lines 323-4 “According to the previous research and analysis, the abrupt change of Angle is 323 caused by the ambiguity of the solar meridian.” What is the ambiguity, authors could cite a relevant refence here.
Authors conclude that their method is correct using fitting, but one of the main points of the algorithm is the solar meridian calculation, what was the error and robustness?
Please cite and compare, if possible, to “Bionic Orientation and Visual Enhancement With a Novel Polarization Camera” by Yujie Wang et al.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors responded to the considerations made in the first review.
Reviewer 2 Report
Thank you for revising the paper and providing adequate information about my concerns.
