Assessment of the Segmentation of RGB Remote Sensing Images: A Subjective Approach
, Tadas Limba 2 and Mindaugas Kiskis 2
Round 1
Reviewer 1 Report
The overall structure of the paper is chaotic, and the content is repeated and complicated. Innovation and priorities are not highlighted. Is MOS innovative? The novle approach should be highlighted and readable. In addition, a detailed technology roadmap should be provided.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper deals with a traditional topic of color image segmentation but it treats the problem from an interesting and new viewpoint. The correlations (both traditional and rank) are calculated for different parameters (metrics) and segmentation results provided by humans. I like this approach and the paper contains a lot of new results in this sense. Meanwhile, to my opinion, the paper can be improved. There are both essential an minor comments. The essential comments are the following:
1) it is worth clarifying what types of images are considered: only color, preudo-color, multispectral presented in color form? If only color are considered, then is the approach applicable to more complex situations?
2) The authors state that humans are able to carry out image segmentation almost perfectly. I can afree with this statement but only under condition that humans have got some training and instructions, especially if we deal with remote sensing color images that differ a little from color images in a common sense;
3) is there information how many images fall into each interval of MOS values? These nunbers can influence cofrrelation factors determined and presented in Tables 10-12;
4) is it possible to explain why such metrics as PSNR, UQI, and SSIM perform well (according to data in Table 12) whilst advanced visual quality metrics "fail"?
5) according to data in Table 12, is it possible to create some combined metric?
Minor ones are the following:
1) Is GT relates to ground truth in line 57?
2) I dislike sentence in Line 82.
3) line 382 - replace tables by Tables
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper presents an interesting study on the existing metrics for stalleite image segmentation.
The study is original and the approach is well described.
There are no evident errors in the thinking or in the presentation.
Only minor corrections are suggested:
- line 193 - the first sentence should have a predicate.
- before each equation from pages 6 and 7 please use ':' instead '.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have answered all the comments well. I recommend this paper for publication in Remote Sensing.
Author Response
No remarks received.
