Laplacian Regularized Spatial-Aware Collaborative Graph for Discriminant Analysis of Hyperspectral Imagery

Round 1

Reviewer 1 Report

Dear Authors

 

This paper presents an interesting study to extract low-dimensional features from Hyperspectral Images. I consider the topic important and worthwhile revisiting taking into account the new possibilities for objects detection and extraction from Hyperspectral Images. The method has multiple applications and would be of interest to different professionals. The paper is well presented, the English is clear, and the overall methodology well explained and properly referenced. The authors demonstrate a deep knowledge of mathematical methods to extract features from Hyperspectral Images and explain them in a clear way. However, the current manuscript requires prior revisions to make the presented work publishable. Some suggestions to achieve this:

The description of datasets has to be clearly separated from the analysis procedure and the obtained results. The current mix is hard to comprehend.

I did not see a section of accuracy assessment that should be added to this manuscript and that’s mixed by experimental section. The accuracy assessment clearly needs to distinguish between true and false positives as well as false negatives.       

I do believe this work deserves to be published but it should first undergo prior revisions as specified above before it will be suitable for publication.

Author Response

Many thanks for your constructive suggestions and comments. Below, we answered your questions one by one.

 

Q1: The description of datasets has to be clearly separated from the analysis procedure and the obtained results. The current mix is hard to comprehend.

 

Response: The section 4.1 has been re-written by adding some descriptions and figures w.r.t. the HSIs data used in our paper.

 

 

Q2: I did not see a section of accuracy assessment that should be added to this manuscript and that’s mixed by experimental section. The accuracy assessment clearly needs to distinguish between true and false positives as well as false negatives.      

 

Response: In addition to Overall Accuracy (OA), the Average Accuracy (AA), Kappa Coefficient (KC) as well as the True Positive Rate (TPR) and False Positive Rates (FPR) have been added in our experiments, which can be found in Table 3, 6 and 9.

Author Response File: Author Response.pdf

Reviewer 2 Report

PLease, provide the original images for the simulation results to observe the diferrences between the original and the classification maps of diferent models 

Author Response

Thank you very much for your invaluable comments.  Our responses to your questions are listed below item by item.

 

Q1: Please, provide the original images for the simulation results to observe the differences between the original and the classification maps of different models.

 

Response: The section 4.1 has been re-written by adding some descriptions and figures w.r.t. the HSIs data used in our paper, and the classification maps also have been updated in Figure 6, 9, and 12.

Round 2

Reviewer 2 Report

The revised paper has been improved according with the reviewers comments