Review Reports

Round 1

Reviewer 1 Report

I still have an essential issue with ground truth. The dataset contains visibile images at 3600x3600 and hyperspectral fibre at 300x300x86 resolution. I assume that the fused data should have 3600x3600x86 resolution, and the same for ground truth. But how do you obtain ground truth? Apparently, you have no data at that resolution.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The manuscript presents a very detailed issue, full of scientific information, but I don't understand how the author has put yellow oil in so many places in the text. What is your purpose? Or is it incomplete? On page 9 you have a presentation error.
As far as I see the analysis of your results almost gives a very vague picture, does not give a specific number or a method, or a specific factor to evaluate, I think it will be better if you add more.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

I consider the submitted paper is suitable for publication, if some improvements  are considered. I am not providing a full feedback review since I consider this paper does not fit the scope of this special issue. This special issue is on active learning, and the paper does not apply any active learning method.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

The Earth's surface is continually imaged by remote sensors operating in a wide range of spectral bands. Images from different spectral bands have very different characteristics and there is often a need to produce composite images to produce new images for particular applications. This paper examines the problem of fusing hyperspectral and panchromatic images using a SSRN model.

The paper is sensibly structured, giving appropriate background and points to a large amount of work already done in this field. It describes the proposed method in detail and gives some details of the three datasets used testing and evaluation. Promising results are presented. There are numerous very minor errors in English grammar, but the meaning is always clear.

I have a few specific concerns - most of them can be addressed very easily.

line 29 - I think that reference [1] is the wrong reference here.

line 34 - You need to explain exactly what you mean by the terms "hyperspectral" and "panchromatic" in terms of spectral bands, etc.

page 2 and throughout - introducing more paragraphs and sub-sections would make the paper very much easier to read.

section 1 - it would be good to hear something about the deficiencies of existing methods in order to justify the introduction of a new approach. Or just say that we are trying to produce something that is "better" (and explain what "better" means!).

line 245 - careful with formatting

line 363 - in section 3.1 you need to describe the datasets a little more in detail. Spectral details for panchromatic? You have three groups - how many images of each type in each group? Are these datasets publicly available?

line 423 - section 3.3 - I'm a little concerned about the subjective nature of the evaluation here. It would be good to start this section by explaining how the evaluation is going to be done using both subjective and objective approaches - the more objective analysis appears later.

line 462 - delectable??

line 473 - this is still the analysis of the results so not really the discussion. And is it possible to say just a little about the statistical reliability of the results?

It would be good to end the paper with a brief discussion about future work.

Check through the references - for example, reference 35 should be H. Li, "Deep Learning for Natural Language Processing: Advantages and Challenges," National Science Review, 2018.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 4 Report

Many improvements have been made to this paper and most of my original concerns have been addressed. I'm still a little concerned that there are still many English grammar mistakes - as a result the paper is still rather hard to follow in places. But I'm not sure how important this is in practice.

A little more has been added to analyse the deficiencies of existing methods and hence to provide further justification for this study. I would have liked to see just a little more.

It's a shame that the testing datasets aren't publicly available - it makes it much more difficult for other researchers to perform comparative studies.