Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript lists approximately sixty methods to document image binarization, including multithresholding algorithms and deep learning techniques. They explain in detail some of these methodologies. However, they describe most of them in just one sentence by giving a general idea of the tool. There is no comprehensive analysis of the selected techniques as required in review papers. A comparative study of advantages and drawbacks should be included, with suggestions on when to use each method.

The authors compare the performance of 22 techniques using well-known quality measures, without explanation for the omission of the rest. These results are partially presented in referenced works such as Kumar et al. and Zhaos et al. Besides, these methodologies are applied only to a data set that can yield partial or biased conclusions.

The authors have done an extensive information relay, which can be included in a complete introduction section. I consider they should propose a deeper study in the available literature. They could present fewer techniques to help a better reader's selection.

Finally, the authors should carefully revise the use of math type within the text and the punctuation of equations by considering them as part of a sentence.

Comments on the Quality of English Language

No comments.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The authors of this manuscript offer an in-depth review of document binarization, including an analysis of the main techniques implemented and a discussion of new challenges and trends.

The topic is a perfect match for the proposed special issue of the journal Electronics.

The manuscript is well written. The literature review is well structured overall.

However, there are a few points to improve before final acceptance of this submission.

I therefore propose the following minor revisions.

1. The title of the manuscript deserves further consideration. I propose the following title: "A Review of Document Binarization: Main Techniques, New Challenges and Trends".

2. The abstract needs to be more thorough. Although the background and the authors' contributions are correctly presented, a quantitative analysis of the survey results is lacking.

3. The first paragraph of the introduction should be expanded. The authors explain the challenges of binarization, including text detection in scenes and medical image analysis. I would like to see articles cited to give more force to what is explained. In the field of medical imaging, with which I am very familiar, there are tried and tested image segmentation techniques. The following article is a good example: https://doi.org/10.3390/cancers14184399 (article to be cited). In relation to this example, what are the remaining challenges in the literature review that the authors propose to carry out?

4. Same comments as above for the second paragraph of the introduction. Thanks to the authors for citing articles to give more force to what is explained about deep learning algorithms.

5. In the title of Figure 1, please indicate the source of the images.

6. When the authors review the various binarization techniques, they summarize the results of their readings in tables. However, in each table, they simply cite each author and describe the method itself. I would like them to add information on the advantages and drawbacks of each method.

7. As far as the results are concerned (see Section 4), the authors use a number of measures. However, their use is completely unjustified. Please correct this in the revised version of the manuscript.

8. Table 7 is the main result of the authors' work. I would, however, like some of the results to be in bold to indicate to the reader which techniques are most effective.

9. In this manuscript, the authors present a detailed review of supervised techniques based on deep learning. What about unsupervised techniques? Unsupervised techniques have a wide range of applications, for example in biometrics: https://doi.org/10.1109/JSEN.2021.3100151.

10. In terms of document formatting:

10.1. In section titles, make sure that the first letter of each word is capitalized.

10.2. Be careful when managing spaces between an author's name and an article quote. For example, write "T. Lelore et al. [ 32 ]" instead of "T. Lelore et al.[ 32 ]".

10.3. Be careful when using mathematical symbols. Remember that the cross represents the simple product, while the star represents the convolution product. On that note, I would like to thank the authors for reviewing equations (2), (3), (7) and (10).

10.4. The revised version of the manuscript must be proofread by a native English speaker.

Comments on the Quality of English Language

Minor editing of English language required.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

In this work the authors provide an extensive review of the methods available for document binarization. The methods reviewed cover the literature well and they provide some experimental comparisons based on numerical performance measures.

I am not very confident about the literature in document binarization as it is not my direct area of research, the papers that seem important and relevant seem to be covered in this review article. While the literature review in the field seems strong and potentially useful for researchers in this field, I have a few comments that hopefully will help the authors improve their paper:

Major comments:
1-  I didn't find clear indication in the text showing if the numerical comparisons were taken from the original papers or were replicated. If replicated, please provide code/github implementations. 

2- there are a few review papers on this topic that I couldn't find in your citations. Please provide proper citation and distinguish your difference to previous reviews. 

Binarization of Document Images: A Comprehensive Review

Wan Azani Mustafa1, Mohamed Mydin M. Abdul Kader

https://iopscience.iop.org/article/10.1088/1742-6596/1019/1/012023/pdf

Document Image Binarization

 

• Showmik Bhowmik 

https://link.springer.com/chapter/10.1007/978-981-99-4277-0_2

2

Document Image Binarization

Konstantinos Ntirogiannis

https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=feeacdc02b027f0d1d6b6395f57fab8b685c78f5

 

3- The other review articles provide a lot of visual examples. Please provide both success and failure examples of different methods.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have submitted a novel version of the manuscript with considerable improvements. I recommend this version for publication in the journal Electronics.