Gradient Weakly Sensitive Multi-Source Sensor Image Registration Method

Round 1

Reviewer 1 Report

This paper presented gradient weakly sensitive multi-source sensor image registration method to supress the noise and alignment differences. The authors conducted relevant research and proposed the method for accurate estimation of the multi-source sensor image using image registration concept. The method mainly consisting of four parts: Feature point extraction, Chunked Harris feature point extraction, Feature point description, and Feature matching and outlier removal for detail restoration. However, in my opinion the paper has some shortcomings in regards to the proposed algorithm.

1)      Introduction need to modify and reference in the introduction not properly cited: Chen et al [] proposed a Partial Intensity Invariant Feature Descriptor (the Partial Intensity Invariant Feature Descriptor, PIIFD) for multi-source retinal image alignment, which has intensity and rotation invariance.

2)      Include the key contribution of the proposed method.

3)      The rest of this article is divided into the following sections also include in the introduction.

4)      In Figure 1 include the reference images and sizes also for the justification of the method.

5)      The initial module of the presented work necessitates a discussion on mathematical analysis, focusing on the different stages inclusion of Gabor convolution sequence. The 3×3 convolution is commonly employed due to its effectiveness in capturing local features and patterns within an image while keeping computational complexity relatively low. On the other hand, the 5×5 convolution offers a larger receptive field, enabling it to capture more global information and intricate details within the data.  Discuss the different sizes and processes that impact on the results. As convolution is the part of the feature extraction for construction process.

6)      The description of the mathematical equation requires further elaboration. To ensure a better understanding of concept, it is essential to discuss terminologies clearly. Additionally, improvements are needed in the clarity and arrangement of mathematical equations.

7)      Figure 5: Schematic diagram of transforming the coordinates near the feature point using the main phase angle. In this figure the axis is not clearly marked include into it.

8)      The significance of the design presented in this manuscript should be articulated more effectively when compared to other important studies published in the same field. I strongly recommend that the authors review recently developed works to better illustration.

9)      Section 5: Comparison and experimental results in place of experiences. Discuss more elaborately quantitative and quantitative analysis including reference numbers.

10)  "Discussion" section should be added in a more highlighting, argumentative way. The author should analyze the reasons behind the obtained results and provide a more robust and comprehensive discussion.

 The readability and presentation of the study require further improvement as the paper suffers from language-related issues.

Author Response

Dear reviewer:

Thank you for reviewing our paper "Gradient Weakly Sensitive Multi-Source Sensor Image Alignment Method".

We would like to respond to your review comments as follows.

For question 1: Introduction need to modify and reference in the introduction not properly cited: Chen et al [] proposed a Partial Intensity Invariant Feature Descriptor (the Partial Intensity Invariant Feature Descriptor, PIIFD) for multi-source retinal image alignment, which has intensity and rotation invariance.

We responded: The chen paper was re-cited and checked for full citations. Page 2, line 66.

For question 2: Include the key contribution of the proposed method.

We responded: The contribution and research significance of this paper has been added in the last natural paragraph in the introduction. Page 2, line 57. Page 2, line 61 to 64. Page 2, line 68.

For question 3: The rest of this article is divided into the following sections also include in the introduction.

We responded: For the introduction section redundancies were eliminated and research implications were added. Page 2, line 72 to 80.

For question 4: In Figure 1 include the reference images and sizes also for the justification of the method.

We responded: The reference image size under the algorithm in this paper can be any size larger than 96*96 and is illustrated below the image. Page 3, line 95 to 97.

For question 5: The initial module of the presented work necessitates a discussion on mathematical analysis, focusing on the different stages inclusion of Gabor convolution sequence. The 3×3 convolution is commonly employed due to its effectiveness in capturing local features and patterns within an image while keeping computational complexity relatively low. On the other hand, the 5×5 convolution offers a larger receptive field, enabling it to capture more global information and intricate details within the data.  Discuss the different sizes and processes that impact on the results. As convolution is the part of the feature extraction for construction process.

We responded: In 2.2, the 2D Log-Gabor filter size of 3 × 3 is introduced. and refinements are introduced from the 2D Log-Gabor filter to the 2D Log-Gabor convolution sequence, which leads to the construction of the PAMI map. Page 4, line 116. Page 8, line 219 to 223.

For question 6: The description of the mathematical equation requires further elaboration. To ensure a better understanding of concept, it is essential to discuss terminologies clearly. Additionally, improvements are needed in the clarity and arrangement of mathematical equations.

We responded: The parameters of equations (1 to 4) were introduced with refinements and the equations were re-edited. Page 7, line 205. Page 3, line 106 to 111.

For question 7: Figure 5: Schematic diagram of transforming the coordinates near the feature point using the main phase angle. In this figure the axis is not clearly marked include into it.

We responded: Updated description of axes at the bottom of Figure 5. Page 9, line 241 to 250.

For question 8: The significance of the design presented in this manuscript should be articulated more effectively when compared to other important studies published in the same field. I strongly recommend that the authors review recently developed works to better illustration.

We responded: The limitations of analyzing other studies in the same field have been added in the introduction and the significance of one's own research has been added in the last paragraph. Comparison of other studies in the same field is also added in the qualitative and quantitative experimental sections. The success of the algorithm of this paper is presented in terms of robustness and resistance to timing differences. Page 2, line 57, Page 2, line 61 to 64. Page 2, line 68. Page 2, line 72 to 80. Page 13, line 337 to 343. Page 14, line 351 to 359. Page 15, line 363 to 373.

For question 9: Section 5: Comparison and experimental results in place of experiences. Discuss more elaborately quantitative and quantitative analysis including reference numbers.

We responded: Discussions have been added to the original experiments to analyze the advantages of robustness and the disadvantages of poor resistance to scale differences, and Nc correctly matched point statistics experiments have been added to augment the quantitative analyses. Page 13, line 337 to 343. Page 14, line 351 to 359. Page 15, line 363 to 373.

For question 10: "Discussion" section should be added in a more highlighting, argumentative way. The author should analyze the reasons behind the obtained results and provide a more robust and comprehensive discussion.

We responded: The concluding section was rewritten and the reasons behind the results obtained were analyzed and discussed. Page 15, line 375 to 392.

Best regards,

Ronghua Li

School of Mechanical Engineering, Dalian Jiaotong University, Dalian 116028, China

[+86 15840612150]

Email: [email protected]

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper presents a comprehensive approach to address the challenging task of aligning heterogeneous remote sensing images. I have a few comments and suggestions, as follows: 1. The abstract could benefit from clearer and more concise language. Some sentences are overly complex and may be difficult to understand at first read. The abstract jumps between different steps of the proposed method without clear transitions, making it slightly challenging to follow the logical flow of the approach. While the abstract mentions qualitative experiments, it lacks specific quantitative results such as improvement percentages or statistical significance, which would strengthen the validation of the proposed method. 2. The introduction could benefit from clearer organization and structure. It jumps between discussing existing methods, identifying challenges, and introducing the proposed method without clear transitions. The introduction briefly mentions the limitations of existing algorithms but could further elaborate on how these limitations motivate the need for the proposed method. Providing a more cohesive narrative would enhance the understanding of the research context. 3. While the introduction references several studies to support the discussion, some citations are missing or incomplete (e.g., "Chen et al []"). Providing complete and accurate citations would strengthen the credibility of the discussion. 4. The feature point extraction section contains complex mathematical equations and technical terms that might be challenging for readers without a strong background in signal processing or image analysis. Providing simpler explanations or additional clarifications could improve accessibility. The transition between subsections 2.1 and 2.2 could be smoother. Specifically, it would benefit from a clearer link between phase congruence transformation and chunked Harris feature point extraction, as they are related but distinct processes. Additionally, in section 2, the figures are helpful, the section could benefit from a more detailed explanation of each figure's significance. Providing captions that explicitly state what each figure illustrates and how it relates to the text would enhance clarity. 5. Some of the mathematical equations introduced in Section 3 are complex and may be difficult for readers to grasp without a strong mathematical background. Simplifying the equations or providing additional explanations could improve clarity. The transition between subsections 3.1, 3.2, and 3.3 could be smoother. Each subsection introduces a distinct aspect of feature point extraction, but clearer transitions between them would improve the overall coherence of the section. While the figures are helpful, the section could benefit from more detailed explanations of the information presented in each figure. Providing captions that explicitly state the significance of each figure's content and how it relates to the text would enhance clarity. 6. Section 3 lacks a discussion on the computational complexity of the proposed method. Providing insights into the computational requirements and potential scalability issues would be beneficial for readers evaluating the practical feasibility of implementing the method. 7. Equation (23) contains a typographical error ("In Equation (22):"). Correcting this error to refer to the correct equation number would improve clarity. Additionally, Equation (23) could benefit from a brief explanation of each variable's significance to aid comprehension. 8. While Section 4 describes the outlier removal technique, a more detailed explanation of the rationale behind selecting 4 feature points and calculating the area ratio of triangles would enhance understanding. Providing a step-by-step breakdown of the outlier removal process could clarify the method further. Providing insights into the computational requirements and potential scalability issues of the method would be beneficial for readers evaluating its practical feasibility. While the section discusses counting the number of inliers to assess the confidence of the affine model, it could provide more clarity on how this measure is used to iteratively refine the model. Expanding on this aspect would improve understanding of the iterative loop process mentioned. 9. While the experimental section provides a detailed description of the experimental setup and results, some parts could benefit from clearer explanations. For instance, the explanation of the outlier removal process could be elaborated further to improve understanding, particularly for readers less familiar with the methodology. Although the section discusses the alignment results and performance metrics, the inclusion of visual aids such as plots or diagrams could enhance the clarity and interpretation of the findings. Visual representations of CMR and RMSE metrics across different methods and datasets would provide a more intuitive understanding of the results. While the section highlights the advantages of the proposed method, it could also address its limitations or potential areas for improvement. Acknowledging limitations helps in providing a balanced evaluation and suggests directions for future research or refinement of the method. While qualitative observations are valuable, incorporating statistical analysis to support the qualitative findings could strengthen the experimental section. Statistical tests or confidence intervals could provide more robust evidence of the method's performance compared to existing algorithms. 11. While the conclusion mentions qualitative verification of the algorithm's advantages, it could benefit from including more quantitative analysis and statistical measures to support the claims made. Providing specific numerical results or statistical significance tests would enhance the credibility of the conclusions drawn. A discussion of potential limitations or challenges faced by the proposed method would provide a more balanced perspective. Acknowledging limitations helps to contextualize the findings and indicates avenues for future research or improvement. 12. Including a brief section on future directions or areas for further exploration would add depth to the conclusion. This could involve discussing potential extensions of the proposed method or addressing specific challenges that remain unresolved. 13. There are some grammatical errors and awkward sentence structures that could be improved for better readability and comprehension.

The quality of English language in the manuscript is generally good, with clear and understandable sentences throughout. However, there are instances of grammatical errors, awkward phrasing, and punctuation issues that detract from the clarity of the writing. Some sentences are overly complex and could be simplified for better readability. Additionally, there are inconsistencies in terminology and usage, which could be addressed for coherence. Overall, while the language is sufficient to convey the ideas, careful proofreading and editing are needed to improve the fluency and consistency of the manuscript.

Author Response

Dear reviewer:

Thank you for reviewing our paper "Gradient Weakly Sensitive Multi-Source Sensor Image Alignment Method".

We would like to respond to your review comments as follows.

For question 1: The abstract could benefit from clearer and more concise language. Some sentences are overly complex and may be difficult to understand at first read. The abstract jumps between different steps of the proposed method without clear transitions, making it slightly challenging to follow the logical flow of the approach. While the abstract mentions qualitative experiments, it lacks specific quantitative results such as improvement percentages or statistical significance, which would strengthen the validation of the proposed method.

We responded: A phased account of the abstract has been provided to strengthen the links between the phases and to complement the quantitative experiments in this paper and to strengthen the validation. Page 1, lines 12 through 24

For question 2: The introduction could benefit from clearer organization and structure. It jumps between discussing existing methods, identifying challenges, and introducing the proposed method without clear transitions. The introduction briefly mentions the limitations of existing algorithms but could further elaborate on how these limitations motivate the need for the proposed method. Providing a more cohesive narrative would enhance the understanding of the research context.

We responded: The limitations of the current algorithms in the introduction are supplemented and the problem addressed by the algorithms in this paper is presented by introducing the challenges of poor generalization, low accuracy and no rotational invariance of the current algorithms. Page 2, lines 72 to 80

For question 3: While the introduction references several studies to support the discussion, some citations are missing or incomplete (e.g., "Chen et al []"). Providing complete and accurate citations would strengthen the credibility of the discussion.

We responded: Re-inserted the "chen" citation and checked the other citations in this article. Page 2, line 66.

For question 4: The feature point extraction section contains complex mathematical equations and technical terms that might be challenging for readers without a strong background in signal processing or image analysis. Providing simpler explanations or additional clarifications could improve accessibility. The transition between subsections 2.1 and 2.2 could be smoother. Specifically, it would benefit from a clearer link between phase congruence transformation and chunked Harris feature point extraction, as they are related but distinct processes. Additionally, in section 2, the figures are helpful, the section could benefit from a more detailed explanation of each figure's significance. Providing captions that explicitly state what each figure illustrates and how it relates to the text would enhance clarity.

We responded: Additional explanatory notes were added for formulas that lacked explanations. Page 3, lines 106 to 111. Rewrite the title of section 2.2 to make it more relevant to the content of the section. Page 4, lines 130. Below the graphs an explanation of what each graph represents was given and it was pointed out that the green color in the middle is a characteristic point. Explanations were made to re-edit the mathematical formulas and add explanations to their parameters. Page 6, lines 154 to 156.

For question 5: Some of the mathematical equations introduced in Section 3 are complex and may be difficult for readers to grasp without a strong mathematical background. Simplifying the equations or providing additional explanations could improve clarity. The transition between subsections 3.1, 3.2, and 3.3 could be smoother. Each subsection introduces a distinct aspect of feature point extraction, but clearer transitions between them would improve the overall coherence of the section. While the figures are helpful, the section could benefit from more detailed explanations of the information presented in each figure. Providing captions that explicitly state the significance of each figure's content and how it relates to the text would enhance clarity.

We responded: Additions to the meaning of the parameters. Page 8, lines 206 to 209. The explanation of the diagrams has been refined below Figures 3-5, Figure 6 has been optimized to make it easier for the reader to understand its expression, and the scalars represented by the horizontal and vertical coordinates are presented below. Page 7, lines 195 to 203. Page 8, lines 219 to 223. Page 9, lines 241 to 250. Page 10, lines 258 to 266.

For question 6: Section 3 lacks a discussion on the computational complexity of the proposed method. Providing insights into the computational requirements and potential scalability issues would be beneficial for readers evaluating the practical feasibility of implementing the method.

We responded: The mathematical formula (19) was re-edited to add explanations to its parameters. Page 7, line 205 to Page 8, line 209.

For question 7: Equation (23) contains a typographical error ("In Equation (22):"). Correcting this error to refer to the correct equation number would improve clarity. Additionally, Equation (23) could benefit from a brief explanation of each variable's significance to aid comprehension.

We responded: Equation (23) has been changed and explained below the equation. Page 11, line 272 to 274.

For question 8: While Section 4 describes the outlier removal technique, a more detailed explanation of the rationale behind selecting 4 feature points and calculating the area ratio of triangles would enhance understanding. Providing a step-by-step breakdown of the outlier removal process could clarify the method further. Providing insights into the computational requirements and potential scalability issues of the method would be beneficial for readers evaluating its practical feasibility. While the section discusses counting the number of inliers to assess the confidence of the affine model, it could provide more clarity on how this measure is used to iteratively refine the model. Expanding on this aspect would improve understanding of the iterative loop process mentioned.

We responded: The outlier removal process is presented in a hierarchical manner with the meaning of interior points and finally the confidence level of each model is determined by estimating the number of interior points under the corresponding affine model. Page 11, line 272 to 274.

For question 9: While the experimental section provides a detailed description of the experimental setup and results, some parts could benefit from clearer explanations. For instance, the explanation of the outlier removal process could be elaborated further to improve understanding, particularly for readers less familiar with the methodology. Although the section discusses the alignment results and performance metrics, the inclusion of visual aids such as plots or diagrams could enhance the clarity and interpretation of the findings. Visual representations of CMR and RMSE metrics across different methods and datasets would provide a more intuitive understanding of the results. While the section highlights the advantages of the proposed method, it could also address its limitations or potential areas for improvement. Acknowledging limitations helps in providing a balanced evaluation and suggests directions for future research or refinement of the method. While qualitative observations are valuable, incorporating statistical analysis to support the qualitative findings could strengthen the experimental section. Statistical tests or confidence intervals could provide more robust evidence of the method's performance compared to existing algorithms.

We responded: A detailed explanation of outlier removal was added. Page 11, line 272 to 274. Section 7 Foresight has been added and directions for improvement have been suggested. Page 15, line 394 to 400. Complemented the quantitative experiments to count the number of correctly matched points, thus providing stronger evidence for this paper. Page 15, line 363 to 373.

For question 11: While the conclusion mentions qualitative verification of the algorithm's advantages, it could benefit from including more quantitative analysis and statistical measures to support the claims made. Providing specific numerical results or statistical significance tests would enhance the credibility of the conclusions drawn. A discussion of potential limitations or challenges faced by the proposed method would provide a more balanced perspective. Acknowledging limitations helps to contextualize the findings and indicates avenues for future research or improvement.

We responded: Section 7 Foresight has been added and directions for improvement have been suggested. Page 15, line 394 to 400.

For question 12: Including a brief section on future directions or areas for further exploration would add depth to the conclusion. This could involve discussing potential extensions of the proposed method or addressing specific challenges that remain unresolved.

We responded: Section 7 Foresight has been added and directions for improvement have been suggested. Page 15, line 394 to 400.

For question 13: There are some grammatical errors and awkward sentence structures that could be improved for better readability and comprehension.

We responded: Grammatical errors and sentence structure of this article have been improved

Best regards,

Ronghua Li

School of Mechanical Engineering, Dalian Jiaotong University, Dalian 116028, China

[+86 15840612150]

Email: [email protected]

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors meticulously integrated all the comments provided by the reviewers into the current version of the paper.

Reviewer 2 Report

My previous concerns are now addressed. I have no further comments.