A Deconvolutional Deblurring Algorithm Based on Dual-Channel Images
Round 1
Reviewer 1 Report
This paper proposes an algorithm for deblurring images by combining the characteristics of multichannel image restoration and single channel blind deconvolution. Simulation and experimental results show the improved deblurring achieved by the proposed algorithm. Following are some minor comments:-
- Please define the abbreviation TV.
- Please improve the quality of text, especially equations in Fig. 2
- In Fig. 3, what is 'Compration method'?
- The manuscript needs grammatical improvement in some places.
Author Response
Cover Letter
Dear Sir/Madam,
Thank you very much for your letter and advice. We have revised the paper“A Deconvolutional Deblurring Algorithm Based on the Dual-channel Images, and all the mentioned were highlighted in red in the revised manuscript. We hope that the revision is acceptable, and I look forward to hearing from you soon.
Thank you and best regards.
With best wishes,
Yours sincerely,
Yang bai
Corresponding author:
Name: Min Huang
E-mail: [email protected]
Author Response File: 
Author Response.docx
Reviewer 2 Report
In this paper, a dual-channel image deblurring method based on the idea of block aggregation by studying the imaging principles and existing algorithms is presented. From my view, this paper is not well organized and the proposed method is not valuable for this research filed. After reviewed this paper, there are some questions and suggestions as follows.
- Article innovation is low.
- All figures need to be enhanced in terms of quality and resolution.
- The literature review is poor. You must review all significant similar works that have been done. Also, review some of the good recent works that have been done in this area and are more similar to your paper.
- The comparison section is relatively weak. The proposed method should be compared with at least 5 other methods.
- It is necessary to experimentally analyze the proposed algorithm in terms of time consumed and compare with other algorithms.
- What are the advantages and disadvantages of this study compared to the existing studies in this area?
- There are many grammatical mistakes and typo errors.
- Write a pseudocode in standard format for the proposed algorithm.
Author Response
Cover Letter
Dear Sir/Madam,
Thank you very much for your letter and advice. We have revised the paper“A Deconvolutional Deblurring Algorithm Based on the Dual-channel Images, and all the mentioned were highlighted in red in the revised manuscript. We hope that the revision is acceptable, and I look forward to hearing from you soon.
Thank you and best regards.
With best wishes,
Yours sincerely,
Yang bai
Corresponding author:
Name: Min Huang
E-mail: [email protected]
Author Response File: Author Response.docx
Reviewer 3 Report
The submitted article presents an algorithm to remove noise from images that experience motion blurring. It appears that the article has undergone a significant review from a past iteration, and could be improved with a number of changes suggested below.
In general, the article requires additional proof reading and editing to reach publication standards. It is suggested that the authors seek a native English speaker to thoroughly read through the article and make edits. In particular, the use of clearly outlined sub sections would be of significant aid for readability. In the current state, there is a lack of consistency throughout the article with the formatting of sub sections. Figure captions also require significant detail. In their current form, they provide little to no description as to what the figure is attempting to explain to the reader.
Page 1, Section 1. Introduction (Third line from bottom of page):
In the parentheses, “(ex: shape and detail )”, do the authors mean “for example…”? There is also an additional unnecessary line space after the word detail.
Page 3, Figure 1:
The caption for this figure needs significant expansion on the description. The current caption provides no detail about the figure.
Page 3, Sub Section “De-noise”:
The sentence “Therefore, the parameter settings…”, is not clear. Can the authors please clarify what they mean in this sentence? Particularly the statement “… rather than protected the image texture.”.
Page 3, Sub Section “Block”:
Can the authors comment on the block sizes? Particularly if the blocks are asymmetrical in size, how does this algorithm handle these cases?
Page 4, Sub Section “Part 3”:
The final sentence of this sub section mentions a “weighted average”, but there is no detail on how this weighted average is selected. Please add this detail.
Page 4, Section 3, second paragraph from bottom of page:
The authors denote lambda(x) and lambda(h) regularization coefficients, and again there is no detail about how these coefficients are selected.
Page 5, Paragraphs 3 and 4:
An estimation of how many iterations are required for convergence should be mentioned here, along with how the algorithm achieves it’s convergence condition.
Page 5, bottom paragraph of page:
The “regularization term” “norm_2” should be italicized, given it is a variable.
Page 6, Section 3.2:
The abbreviation “EM” is not previously defined before being used. Please define it.
Page 6, Figure 2:
Again the caption for this figure provides no detail about the figure itself. The figure also does not add anything to the article, as it simply reuses the equations already shown in-line in the text of the article. The authors should consider how to present this figure more carefully and it’s utility.
Page 7, paragraph 4, preceding Table 1:
This paragraph is unclear and needs to be clarified.
Page 7, Table 1:
Again, the caption provides no detail as to what the table is presenting to the reader. “Nom_2” is not consistent with the use of “norm_2” in text. Please ensure this is consistent. The following line where this variable is used is again not consistent, and uses a dash, rather than an underscore “norm-2”, not “norm_2”.
Page 8, bottom line:
Typo, “he data”, “the data…”?
Page 9, first line:
The authors state that 90 iterations are used, why this number of iterations? Detail needs to be provided here as to why this number of iterations is used.
Page 9, Figure 3:
The plot of the SSIM on the right shows little to no significant difference at the 90 iteration point. Again the caption needs detail. The comparation method used also should be highlighted here, and the legend of the plot modified accordingly.
Page 9, Section 4, paragraph 2:
The authors discuss using the “mean-shift clustering algorithm” for kernel classification. Additional detail needs to be given as to how the algorithm deals with outlier cases.
Page 9, Section 4, paragraph 3:
The sentence “Second, the rest of the area…” does not make sense. Please clarify.
Page 10:
The outlining of the steps used needs reformatting to help with readability. The authors could consider using a figure here.
Page 10, Section 5, paragraph 1, line 3:
There appears to be another typo here “stimulation data”, do the authors mean “simulation data”?
Page 11, Sub Section “Group 1”, paragraph 1:
The sentence “The comparisons obtained by the literature…” has no context and needs clarification.
Page 11, Sub Section “Group 1”, paragraph 2, sentence 2:
The sentence “From a subjective visual point…” has grammatical issues. Please clarify.
Page 11, Table 2:
Again caption has no detail. This table also spans across from the bottom of page 11 to page 12, please either add a page break or ensure the table is kept together on a single page.
Page 12, Figure 5:
Again, caption has little detail.
Page 12, Tables 3 and 4:
Again, caption has little to no detail. A line separator between the proposed algorithm and results from literature is needed to separate the results for readability. Also, why is there no comparison with Reference 11? Please include the results from this algorithm as well.
Page 13, Figure 6:
The results of deblurring the vehicle image in the top right for results from References 11 and 14 appear to swapped around. Why is this the case? It is also subjectively arguable that the imaging results from deblurring the number plate shows more detail and is sharper from the results of Reference 11 than the proposed result (i.e. the letter “B” (fifth character) is clearly visible in the results of Reference 11, but is not the case in the proposed result). Additionally, the figure caption requires editing for clarity.
Page 13, Figure 7, part (a):
The outlines for the magnified images are not consistent with figure 6. Please ensure line widths and colors match. Again, the two parts of the image should be presented on a single page, rather than spanning across two pages, for readability. Also, part (b) of the figure should appear before part (a).
Page 14, paragraph 1:
The discussion of the results from the figures references the wrong figures. Please correct. As commented above, the results are arguable subjectively, particularly when magnified. This section requires more discussion.
Page 15, Table 7,
Again, there is no detail in the caption.
Page 15, Figure 8:
Please ensure the caption does not span across to another page. Why is there no result from Reference 11? Please include this result for comparison. The caption also requires clarification. In part (d) of the caption, the sentence abruptly stops.
Page 16, Figure 9:
Again results from Reference 11 need to be presented. A discussion on the proposed kernel result is needed. Please explain why there 12 blocks, and the differences in the kernels obtained.
Page 16 and 17, Figure 10:
Part (a) is not defined. The plot shown appears to be above 1 for both plots at a spatial frequency of 0.2. Please clarify. Again, a result from Reference 11 is needed for comparison here. Overall, why is this figure separate to Figure 8?
Page 17:
The discussion would benefit from the addition of a number of points.
Is the code for the presented algorithm publicly available, if so, where? If not, why?
The presented results only focus on monochrome images, how does it perform with color images?
There is no discussion on the speed or computation requirements of the proposed algorithm. If the algorithm takes significant processing power/time over algorithms from References 11 and 14, the amount of improvement presented would lack significance.
Following on from the point above, if the computational requirements are low, what are the proposed applications?
Author Response
Cover Letter
Dear Sir/Madam,
Thank you very much for your letter and advice. We have revised the paper“A Deconvolutional Deblurring Algorithm Based on the Dual-channel Images”, and all the mentioned were highlighted in red in the revised manuscript. We hope that the revision is acceptable, and I look forward to hearing from you soon.
Thank you and best regards.
With best wishes,
Yours sincerely,
Yang bai
Corresponding author:
Name: Min Huang
E-mail: [email protected]
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Good revisions have been made in the paper and the revised version has the necessary qualities for acceptance compared to the previous version. In my opinion, the article is acceptable in its current form.
Author Response
Dear Sir/Madam
Thank you very much for your recognition of my work. I have made great progress with my paper as a result of your advice. I have again proofread the grammar and data in the new version of the paper.
Once again, thank you very much for your valuable comments.
Yours Y.B
Author Response File: Author Response.docx
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
The authors present an algorithm that combines previously published work with the aim to improve deblurring of images (single and dual channel).
General
The paper seems relevant and original, with possible impact. However, the presentation is far from ideal. The writing is very confusing, many grammar errors, and the style also needs work--some sentences lack scientific style e.g. line 72, 154.
The paper uses past tense incorrectly throughout. For example, in line 273 the text reads: "The experimental results were shown in Fig. 2". It should read "The experimental results are shown in Fig. 2" Authors need to distinguish between when they speak of the experiments performed (past tense) and when this are presented in the text.
The article can not be publish as it is, it requires major work in order to make it readable. Find attached a pdf with annotations to help tackle some of the many issues
Specifics
Do the numbers in tables 1, 2 and 3 correspond only to the examples shown un Fig. 1, 2 and 3, respectively? If so, then the paper lacks enough experiments. Can't prove the superiority of the proposed algorithms with only four images.
The References require to be checked, as the format does not look correct. For example, where was reference 8 published?
Citation of references in the text also require to be reviewed. There are et al. missing, surnames in the text don't match the one in the Reference section.
Comments for author File: 
Comments.pdf
Reviewer 2 Report
The authors propose a deblurring algorithm (DCBCD) combining the advantages of tra- ditional multi-channel restoration and the single-channel blind deconvolution restoration algorithm. The estimation of the blur kernel is improved and alleviates the problems of inconsistency in the blur kernel by using the image correlation by slices. Although the idea sounds interesting, the material reported does not reflect the method’s advantages. In general, the material is not well reported; there is no formal presentation of the mathematical calculations. The document is not well written. The method improves the state-of-the-art deblurring results, but I do not recognize the novelty of the approach to be published; even more, the gain is up to 1 dB in the three groups of data reported, which in my opinion, is not enough to be significantly. With these considerations, my recommendation is to reject the manuscript. Other comments regarding specific lines in the manuscript are presented in the attached file.
Comments for author File: Comments.pdf
Reviewer 3 Report
-There are too many examples of incorrect sentences or typos, spaces, and inappropriate use of capitalization in the manuscript. Thorough English proofreading is a must. Also, there are many expressions that are not generally used in the manuscript, and it seems that the submission was made hastily without sufficient revision work before submission.
-Eq(10) subscript j is missing for del y’
- Please report the standard deviation of the quantitative measures in Table 2 & 3. Or is it just based single example shown in Fig 2 & 3?
- It is necessary to clarify why the following two papers were selected for performance comparison. Both papers are very old papers, so they do not seem appropriate for comparison. It is necessary to mention and compare the performance of more recent dual-channel image deblurring methods.
- Sroubek F.; Milanfar P . Robust Multichannel Blind Deconvolution via Fast Alternating Minimization[J]. IEEE Transactions 388 on Image Processing, 2012, 21(4):1687-1700. 389
- Z.Xu.; Z.Fugen.; X.Z. Bei. Blind deconvolution using a nondimensional Gaussianity measure. 2014 IEEE International 394 Conference on Image Processing. IEEE, 2014.
