Vision-Based System for Black Rubber Roller Surface Inspection

Round 1

Reviewer 1 Report

Related to “Vision-Based System for Black Rubber Roller Surface Inspection”

# Comments

Rubber rollers are used in many manufacturing processes to accomplish various mechanical functions and create a uniform pressure zone. Some rollers are used to apply paint or printing inks.

The paper has good quality and recommends for publication after major revision

 

v  The introduction and its continuation have no references. Authors should include relevant and up-to-date references.

v  The defects are categorized into six main categories, which include damage, corrosion, material lacking at the chamfer, grind lacking, scratch, and porosity, but in figure.1, 11 types of defects are presented. The authors should have clarified appropriately.

v  Visual inspection challenges should be described and extended after figure.1. Present in a table is highly recommended.

v  According to the previous comment, the conventional inspection methods should be described and extended after figure.1. Present in a table is highly recommended.

v  Relevant and recent references must be cited in lines (47-79).

v  I recommend Section 2 title, “Related work” be changed to “Discussion”

v  I recommend that the advantages and disadvantages of discussed approaches in section 2, be summarized in a table.

v  Is the developed system in this work cost-effective? Please specify the novelty and advantages of this method and compare it with other methods.

v  Please numbered all subtitles, such as Laplacian Filters, Image Segmentation: Thresholding, Morphological Image Processing, and…

v  The captions of the 15, 16, 17, and 18 figures need additional data and must be extended.

 

 

 

Minor editing of the English language required

Author Response

We have submitted the response to Reviewer. Could you please find the file for our answers

Author Response File: Author Response.docx

Reviewer 2 Report

The authors claim that the article proposes a machine vision system for monitoring the surface of black rubber rollers in production processes. The system is aimed at improving the surface quality of the rollers by detecting and classifying defects. Two algorithms are proposed for detecting defects: the traditional method and the method based on deep learning. The authors claim that the first one is fast, but limited to detecting surface defects, while the second one is slower, but is able to detect and classify defects.

The authors show that the accuracy of the algorithms is verified by experiments, with the traditional method achieving an almost perfect accuracy of approximately 98% for detecting defects, and the method based on deep learning achieves an accuracy of approximately 95.2% for detecting defects and 96% for classifying defects. The authors claim that the proposed machine vision system can significantly improve the control of the surface of black rubber rollers, thereby ensuring high quality production.

Applying two methods, the authors came to the conclusion that each method has its advantages and disadvantages, and both methods give good results in identifying defects and can be applied in industry.

The work deserves to be published in the journal of  Applied Sciences.

Comments for author File: Comments.docx

Minor editing of English language required

Author Response

Authors would like to thank you so much for your comments. We really appreciate these comments. Based on your suggestion, the authors have checked carefully  English edting. The revision have shown more details in our Manuscripts.

Reviewer 3 Report

This manuscript describes two proposed surface defect detection methods based on traditional and deep-learning ways respectively. Authors thoroughly went through the background, experimental setup and result analysis, and eventually showed the high accuracy of traditional way and classification capability of deep-learning based algo. Though it is detailedly written, there are some questions/issues needed to be addressed:

1. Fig.1 shows different categories of surface defects. Please use higher resolution and contrast graphs to illustrate them, especially for f, h, k which have defects right on the rubber.

2. How the input data is determined in table 1? Would results be different if initial input changes?

3. Better to keep the whole 3x3 grids of fig. 7 in the same page without breaking rows.

4. Line272-281 describes how morphological filters work to improve the quality (fig. 10). Please explain more on the erosion and dilation filters by using related references or listing the formula.

5. In line365, gamma factor value is determined by experiments. Please explain a bit more on what types of experiments have been done and how it is determined.

6. Line410, section index should be "4.1.1" instead of "4.4.1"

7. Line490, section index should be "6" instead of "5"

8. In the manuscript, authors mentioned different lighting systems have been compared in section 4 multiple times (Line412, Line489). I didn't find any detailed comparisons with data and analysis. Please add them.

9. Given that both traditional method and deep-learning one have their drawbacks, how these 2 methods should be used together to achieve high accuracy and high classification capability?

10. Please provide some comparisons between methods described in the manuscript and other algorithms currently being used.

Author Response

We have submitted the response to Reviewer. Could you please find the file for our answers

Author Response File: Author Response.docx

Reviewer 4 Report

The machine vision system for the surface examination of black rubber rollers used in manufacturing operations is proposed in this study. The system's goal is to improve the rollers' surface quality by identifying and categorizing flaws. A lighting setup is put in place to draw attention to surface flaws. A traditional-based algorithm and a deep learning-based algorithm are the two methods that are suggested for flaw identification.

The proposed research is interesting and found within the scope of this journal.

The following recommendations need to be incorporated into the current manuscript before acceptance:

1)     Extensive English language corrections are required. Avoid using pronunciations.

2)     Abstract should be more informatics. Add important results.

3)     Introductions seemed to be general and no references were cited in the whole text.

4)     From where you have drawn Figure 1. (give the literature citation)

5)     Literature review must be re-written, adding the critical review with adding more than one author and similar findings together. Write the literature gap, motivation, and novelty of this work.

6)     In Figure 2 and Figure 3, labeling of mechanical structural parts and lighting system parts is required.

7)     Provide the essential units of terms mentioned in all Tables used in the manuscript.

8)     Cite the references for all equations used in the manuscript.

9)     Actual physics of both systems used in this study should be compared. Also, add agreement and disagreement considering research papers available in the literature.

 

10)  Add some comparative performance results in the Conclusion. Also, the future scope should be written separately at last. 

Extensive editing of the English language required

Author Response

We have submitted the response to Reviewer. Could you please find the file for our answers

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Accept in present form

Author Response

Authors would like to thank you so much for your accept. In additional, we have updated more literature reviews and checked carefully English as show in manuscript. 

Your Sincerely,

Authors

Author Response File: Author Response.docx

Reviewer 3 Report

Authors addressed some of the comments but not all. Need some minor modifications before being accepted:

1. Figure 1 (f) is still not clear. Recommend to use a different contrast to show the defect shape.

2. How erosion and dilation filters being implemented is still not added to the manuscript.

Author Response

Authors would like to thank you so much for your comments. We really appreciate these comments. Based on your suggestion, the authors added some modifications and details:

1. Figure 1 (f) is still not clear. Recommend to use a different contrast to show the defect shape.                                                                               

We have updated the Figure 1 as your comments

2. How erosion and dilation filters being implemented is still not added to the manuscript.

We have done by following steps

Morphological processing:

- First use Erosion aka image erosion to:

+ Remove isolated noise pixels.

+ Removes noisy pixels around the object to make the border of the object smoother.

- Then use Dilation which is the inverse operation with Erosion to expand the original image to:

+ With broken images, it can help to connect the image

+ Fill in the gaps and smooth the boundaries of the object

The results have shown in manuscript. In additional we have updated the literature reviews and grammar English in this papers.

Your Sincerely,

Authors

Author Response File: Author Response.docx

Reviewer 4 Report

The authors haven't considered the reviewer's comments response completely. However, It is required to modify the introduction part as follows:

Add more recent works, Write critical observations, Write research gaps, and Write the objectives of this research, Also small paragraphs should be merged.

Minor revision is required.

I don't find any changes in the English language. Extensive editing is required. 

Author Response

Authors would like to thank you so much for your comments. We really appreciate these comments. Based on your suggestion, the authors added some modifications by updating more literature reviews and grammar English in this papers. The results have shown in manuscript.

Your Sincerely,

Authors

Author Response File: Author Response.docx