Real-Time Multi-Scale Barcode Image Deblurring Based on Edge Feature Guidance

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper proposes a EGMSNet for barcode image restoration. While the method shows promising results in improving barcode clarity and decoding accuracy, there are several issues that need to be addressed to strengthen the scientific clarity of the work.
(1)Lines 59-67 - The motivation and contribution statements are relatively weak. The authors should better justify why existing deblurring methods are insufficient for barcode images specifically.
(2)Lines 78-92 - The literature review of traditional deblurring methods needs more depth and critical analysis rather than just listing different approaches.
(3)The overall architecture description lacks sufficient technical details about key components like the encoder-decoder structure and cross-scale feature fusion.
(4)Lines 161-171 - The edge branch network description is unclear about how exactly the edge features are extracted and integrated with encoder features.
(5)Lines 180-188 - The mathematical formulation of the feature filtering mechanism needs more rigorous derivation and explanation.
(6)Lines 236-239 - The datasets used for evaluation appear limited. More diverse real-world barcode images should be included for comprehensive validation.
(7)Lines 252-266 - The ablation studies lack statistical significance tests to validate the improvements from different components.
(8)Lines 289-298 - The comparison with state-of-the-art methods needs more thorough analysis beyond just reporting metrics.
(9)The discussion section is too brief and does not adequately address limitations or potential improvements.
(10)Several grammatical errors and unclear sentences need to be corrected for better readability.
(11)The authors may add more state-of-art cv articles in various fields (3D vision technologies for a self-developed structural external crack damage recognition robot; Automation in Construction. Deep Neural Remote Sensing and Sentinel-2 Satellite Image Processing of Kirkuk City, Iraq for Sustainable Prospective; Journal of Optics and Photonics Research. ).

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The author proposes a multi-scale real-time deblurring method guided by edge features, which significantly improves barcode clarity and scanning accuracy, particularly in noisy environments. However, there are still several areas in the paper that require improvement.

1. There are many references on image deblurring, but the author does not analyze the issues present in the methods discussed in these references. The paper should address which problems it can solve.
2. In Section 3.1, the author proposes an edge-guided multi-scale fusion network architecture, EGMSNet. What are the advantages of this architecture, and why is it effective in enhancing the clarity of image edges?
3. In the ablation study, there should be a comparison between the deblurring performance of EGMSNet and the effect of introducing FFM into EGMSNet.
4. Compare the time complexity and computational complexity of the barcode image deblurring method proposed by the author with all the comparative methods.
5. The discussion section should highlight the advantages and limitations of the proposed method.
6. The conclusion section also seems to rush to the end. The authors will have to demonstrate the impact and insights of the research. The authors need to rewrite the entire conclusion section with focus on both impact and insights of the manuscript. Clearly state your unique research contributions in the conclusion section. Provide some future directions.

Comments on the Quality of English Language

The quality of the English language does not affect my understanding of the paper. Overall, the language quality is acceptable.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

1. The EGMSNet architecture diagram in Figure 1 is not fully explained in the main text. The sizes of the three scales S1, S2, and S3 are not defined, and there is a lack of specific description of the cross-scale information propagation process, including how it is transmitted and how this transmission enhances the deblurring effect. Moreover, although the text mentions that each ResBlock has two 3*3 convolution layers, the specific parameters of these convolutions are not further explained. It is suggested to add more specific details.
2. Is using the Sobel operator to extract edges the best choice? The traditional Sobel operator does not have adaptive capabilities. Has any other technology been introduced to enhance the performance of the Sobel operator? The text mentions the sensitivity of the Sobel operator, but it does not explain whether noise will interfere with the experiment. The method is not comparedwith other edge detection methods. It is suggested to add comparison content to explain why the Sobel operator is more suitable for processing images.
3. The design of the FMM module includes the spatial selection module (SSM) and the frequency selection module (FSM), but the necessity of these modules is not explained in the experimental part. It is suggested to conduct ablation experiments to further illustrate their necessity.
4. The weight α=0.1 of the wavelet reconstruction loss performs optimally in the current experiment, but it is still necessary to further explore whether it is applicable to all image types and degrees of blurring. It is suggested to consider dynamically adjusting the value of αbased on the degree of blurring of the image.
5. There is a spelling error in line 382, "deblur-ring" should be "deblurring". It is suggested to carefully proofread the paper.
6. The discussion section mentions that the performance of the method in this paper "still has room for improvement under extreme blurring or low contrast conditions", but no detailed results are provided. It does not specifically explain how these limitations are manifested. It is suggested to add an analysis of the reasons and improvement directions.
7. Some of the references are outdated, and the latest related research has not been cited.

Comments on the Quality of English Language

The quality of English language is acceptable, but it is suggested to carefully proofread the whole paper.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

Dear Authors,

This study's primary question explores proposing a multi-scale real-time deblurring method based on edge feature guidance. There is a designed network that integrated an Edge Feature Fusion Module (EFFM) to restore image edges better. It offers a Feature Filtering Mechanism (FFM), which effectively suppresses noise interference by precisely filtering and enhancing critical signal features. The different modules, experiments and their comparisons are used throughout the study, illustrating the findings with data sets, Tables, Figures and interpretations. There is commendable research conducted. It presents a valuable opportunity for this method's implementation and adaptation in practical applications in different industrial environments and fields, including automatic identification and data collection systems, where barcode technologies are used.

Certain sections of this article require a few revisions based on the information available before publication. I have highlighted the main details in the general and specific comments. Please refer to the attached document, which outlines the minor issues identified in the manuscript.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 5 Report

Comments and Suggestions for Authors

In this paper, the authors propose a real-time and multi-scale network for de-blurring barcode images. The EGMSNet model aims to recover image edges better and details using EFFM and FFM. The model is specifically designed to improve the readability of barcodes and performs in real time. The motivation for the work is clearly stated. The negative effects of blurred barcode images on barcode recognition systems are emphasized in detail. The need for this problem and the limitations of existing methods are clearly stated in the introduction of the paper. The results of the study are satisfactory; however, I would kindly offer the authors the following gentle suggestions:
1- The study uniquely integrates EFFM and FFM by developing multiscale networks, which are commonly used in the literature. However, the elements of novelty need to be stated more clearly comparatively and the differences with respect to existing methods should be emphasized more strongly.

2- more extensive testing on real data sets should be presented in the discussion section.
3- The mathematical background of the study should be given.
4- The interpretability of the results should be improved by adding ROC curves or additional explanatory visualizations.
5- The novelty of the study should be emphasized more strongly in the comparative literature.

Comments on the Quality of English Language

In this paper, the authors propose a real-time and multi-scale network for de-blurring barcode images. The EGMSNet model aims to recover image edges better and details using EFFM and FFM. The model is specifically designed to improve the readability of barcodes and performs in real time. The motivation for the work is clearly stated. The negative effects of blurred barcode images on barcode recognition systems are emphasized in detail. The need for this problem and the limitations of existing methods are clearly stated in the introduction of the paper. The results of the study are satisfactory; however, I would kindly offer the authors the following gentle suggestions:
1- The study uniquely integrates EFFM and FFM by developing multiscale networks, which are commonly used in the literature. However, the elements of novelty need to be stated more clearly comparatively, and the differences with respect to existing methods should be emphasized more strongly.

2- more extensive testing on real data sets should be presented in the discussion section.
3- The mathematical background of the study should be given.
4- The interpretability of the results should be improved by adding ROC curves or additional explanatory visualizations.
5- The novelty of the study should be emphasized more strongly in the comparative literature.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The author has made careful revisions based on the reviewer's comments and agrees to accept the manuscript.

Comments on the Quality of English Language

We have reviewed the manuscript, and while the quality of the English language may not be perfect, it does not significantly impact the understanding of the article.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have improved the paper based on the comments of the reviewer.

Comments on the Quality of English Language

The quality of English language is acceptable.

Reviewer 5 Report

Comments and Suggestions for Authors

The authors have made the corrections I suggested and responded to the comments; in this sense, the manuscript is satisfactory to me, and I kindly suggest that the manuscript be accepted.