Perceptual Quality Assessment of Pan-Sharpened Images
and Alan Bovik 3,†Round 1
Reviewer 1 Report
Please see the attached file.
Comments for author File: 
Comments.pdf
Author Response
Please see the attached pdf file with the answers to the comments
Author Response File: Author Response.pdf
Reviewer 2 Report
The evaluation of the results is a very important and challenge task in the literature of pansharpening. Based on the statistics of natural images, the paper present a novel image quality assessment measure, which supports the visual qualitative analysis of pansharpening results. Although it is a very interesting and well-written work, there are some comments to be addressed before its publication.
1. It is stated in Line 67, Page 2 that "Even though these approaches process the images at the native scale, they are biased by the definition of the indexes", Why the biases happened? In addition, can you validate your proposed image quality measure, i.e. the opinion aware and opinion unaware quality analyzers, at the reduced scale or for some simulated image? since the ground truth is known, it is easy to validate the proposed image quality measure.
2. the motivation for using the natural scene statistics to develop the human perceptual quality assessment is not clear, please add some statements in the introduction section for this and also introduce some work for perceptual quality assessment.
3. The organization of subsection 2.3 can be improved. For example, a algorithm flowchart can be given in this subsection. In addition, how to get the pristine model? it said in Line 194, Page 6 that " ... is constructed from 20 images...", while in Line 242, Page 7, it said " The pristine model was composed of 80 original MS images....".
4. Please add more details about the Opinion Aware Image Quality Analyzer in subsection 2.4. It is not clear that how to train the opinion-aware quality model, for example, how to get the quality aware-feature space?
5. In fact, the experiments are not sufficient enough since only one kind of satellite data set is used to illustrate the effectiveness of the proposed two analyzers.
Author Response
Please the attached pdf file with the answers to your comments.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have significantly revised the paper. I have some additional suggestions.
In some tables, it will be good to use bold numbers to highlight the best performing methods.
I am surprised to see that PCA is considered as a high performing algorithm in several tables. This is contrary to those findings in Vivone et al. paper. Can the authors offer some explanations for this observation?
The GS or GSA method was not included. In my opinion, GSA achieves a good compromise between computational burden and performance. Will the authors include GSA in the revised paper?
Author Response
Thank you for the second review of our paper. We have answered your questions in the attached pdf file.
Author Response File: Author Response.pdf
Reviewer 2 Report
Most of problems I concerned have been solved, although not perfectly. The submission has been greatly improved and is worthy of publication.
Author Response
Thank you for the evaluation of our paper. Your comments and suggestions were very useful to improve the quality of our work.
Round 3
Reviewer 1 Report
The authors have addressed my comments. The paper is acceptable now.
