A Composite Vision-Based Method for Post-Assembly Dimensional Inspection of Engine Oil Seals
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsA novel vision-based method streamlines engine oil seal inspection, merging 2D imaging with 3D reconstruction. A circular ROI mask, refined through progressive template matching, eliminates interference, while phase-shift striping and M-RANSAC enhance height accuracy. Outperforming traditional point cloud methods, it achieves 98.16% success, far surpassing manual measurement. With precision within 0.3323%–1.4461% and stable repeatability, it ensures reliable, efficient, and automated assembly inspection. Figure 1: it does not scale or dimensions.
All references are from China, please make international effort. For instance, the idea of monitoring is in some works by https://doi.org/10.1080/10426914.2016.1244838 with that the work will see a wider scope. One of the biggest developments of this century is the incorporation of Artificial Intelligence (AI) in the machine vision technology that has seconded the accuracy, flexibility, and utility of manufacturing activities. Or https://doi.org/10.1016/j.jmapro.2023.07.028
- Figure 9 too dark
- Figure 19 shows the best results, please explain why this and not 17 and 18.
- The algorithms is ok, pleas define the 1-2 adaptations specially developed for your case.
- Figure 3: too small, please check in addition the work: Tool wear monitoring of high-speed broaching process with carbide tools to reduce production errors, Mechanical Systems and Signal Processing 172, 109003 you must try to propose more international references, as the visual and other sensors approaches in Early detection of tool wear in electromechanical broaching machines by monitoring main stroke servomotors, Mechanical Systems and Signal Processing 204, 110773
- Include a better discussion of what is measured, in figure 1: diameter? Other?
- Table 1 shows times..is it so critical..is not precison or fault positives more important?
Make a better version.
Author Response
Please see the attachment. And our modifications/additions to the original are given in yellow text in the submitted manuscript.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper aims to develop a composite vision-based inspection method for measuring the dimensions of engine oil seals after assembly, addressing the limitations of traditional manual and contact-based measurement techniques. The study includes the introduction of a multi-stage template matching approach to extract regions of interest, the combination of structured light 3D reconstruction with image-based template matching for precise dimensional measurement, and the application of an improved RANSAC algorithm to enhance plane fitting accuracy. The proposed method significantly improves measurement precision, reduces interference from non-target regions, and enhances computational efficiency compared to existing techniques. Experimental validation demonstrates that the method achieves a high success rate (98.16%) in an industrial setting, surpassing traditional approaches in terms of reliability, repeatability, and real-time applicability.
The title is clear and appropriate. It accurately reflects the study's focus on composite vision techniques for post-assembly inspection of engine oil seals.
The Abstract effectively presents the research problem, proposed method, and key findings. It could be clearer in explaining the novelty of the method compared to existing approaches. Additionally, some phrases repeat similar points, and numerical results could be more concisely presented. The practical significance of the proposed method for real-world applications should be better highlighted.
The Introduction clearly defines the problem of traditional dimensional measurement methods for oil seals after assembly and highlights the need for automated visual inspection. The literature review is quite detailed and includes too many technical details about existing methods, which could be summarized or moved to the Principles section. The main research hypothesis is not explicitly stated. The final paragraph presents the new method, but it should better emphasize how it differs from previous approaches. A short section linking the problem to industrial requirements, explaining the specific challenges of manual inspection in manufacturing could be included.
The Principles section is well-structured and provides a detailed explanation of the theoretical foundations behind the proposed method, including structured light measurement, template matching, and the M-RANSAC algorithm.
The description of PSP is technically accurate. It is suggested to include a brief comparison with other structured light methods to justify the choice of PSP.
The explanation before the equations 1 and 2 could be slightly reworded to clarify the physical meaning of each term.
The reason for using progressive template matching should be emphasized more (explain why is it superior to single-template matching).
The process of obtaining the elliptical region from template matching is well-described, but the impact of possible variations in lighting and reflections should be briefly discussed.
M-RANSAC algorithm: The improvements over standard RANSAC should be more explicitly stated and the choice of parameters (Table 2) should be briefly justified (why were these specific values selected).
Distance calculation: The explanation of how height differences are extracted could be clarified by explicitly stating the assumptions made (e.g., whether the reference plane is assumed to be perfectly flat); the relationship between measurement uncertainty and sensor precision should be briefly mentioned.
The Experiments and Results section presents the experimental setup, comparative evaluations, and performance analysis of the proposed method. The methodology is well-structured, and the results are comprehensive.
The experimental setup should briefly explain why specific parameters (e.g., resolution, focal length) were chosen.
The ROI extraction parameter table (Table 5) needs a short explanation of how values were determined.
The comparison of methods should summarize the weaknesses of competing methods in a concise table.
The repeatability analysis (Figures 13-14, Tables 6-7) should discuss factors that could affect repeatability.
The manual vs. automatic measurement comparison (Figures 16-19) should include additional statistical validation and describe how manual measurements were conducted.
The overall performance evaluation should briefly analyze the failure cases (2%), what caused them, and how can the method be improved.
The Conclusion effectively summarizes the study’s key findings and highlights the advantages of the proposed method.
It is suggested to summarize the results more concisely.
The industrial applicability statement should briefly mention potential implementation challenges (e.g., cost, scalability).
Acknowledge the limitations of the study and add a future work paragraph to suggest possible improvements.
The references cited in the manuscript are appropriate and relevant to this research, but it is suggested to include more recent studies and clarify how the cited works compare to the proposed approach.
Note: The authors are kindly requested to include exact line numbers in the manuscript corresponding to the added clarifications for each response to the comments (in addition to highlighting the text in a different color, which is assumed).
Comments on the Quality of English LanguageThe English in the manuscript is understandable, but there are some grammar and syntax issues that require revision. The text needs sentence restructuring and proofreading. Moderate editing is suggested.
Author Response
Please see the attachment. And our modifications/additions to the original are given in yellow text in the submitted manuscript.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe authors proposed a vision-based method for detecting the assembly size of engine oil seals. To achieve accurate positioning of the inner and outer ring regions of the oil seals, the process begins with obtaining the center point and the major/minor axes through ellipse fitting. By combining the three-frequency four-step phase-shifting profilometry with an improved RANSAC algorithm for plane fitting, the final calculation of the height difference between the inner and outer rings of the oil seals. Experimental analysis demonstrated their feasibility.
Overall, the submission is written well and the innovation parts are just fit. I have only some minor suggestions as below:
- Template matching is time consuming and results are prone to noise. How about edge detection combined with circle Hough tranform?
- The proposed method belongs to traditional approach and may take lot of computation time. Please discuss the whole time spent for all the measurement D1~D3 and whether superior than manual inspection.
- Please make a comparisons with the referred works [10-16] regarding to the performance metrics.
Author Response
Please see the attachment. And our modifications/additions to the original are given in yellow text in the submitted manuscript.
Author Response File: Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for AuthorsThe manuscript proposed a composite vision-based method for detecting the assembly size of engine oil seals. The main contribution of this paper is the non-contact measurement of the dimensions of oil seals post-assembly on the engine. However, some observations and suggestions need to be solved to improved the paper.
– The introduction section needs to be improved, the authors should highlight the main contribution.
– In the sentence ‘’ an increasing number of scholars’’ ; ‘’many scholars have proposed the met’’, This reviewer suggests using ‘’researchers’’ instead of scholars.
– Figure 1 should describe it in depth, the authors could explain in three parts a),b),c) for better illustration. The size of the letter is too small to be readable.
– RANSAC and ROI need to be defined in the abstract section or introduction.
– Please use the equation in text for the ‘’obtained as D=D2-D1’’.
– Figure 7 is too small to read and the quality of the image needs to be enhanced. The axis labels of Figure 16 and Figure 19 are too small (increment the size font).
– In the sentence ‘’where (𝑥𝑝, 𝑦𝑝) is the pixel coordinate’’ lacks of function phi (φ
). φ(𝑥𝑝, 𝑦𝑝).
– Table 4, please capitalize ‘’mean error’’.
– The description of Figure 11 could be enhanced if the authors explain in a paragraph.
– Please standarize ‘’method II’’ or ‘’Method II’’ to be uniform for the main concepts of the manuscript.
– The quality of Figure 14 is poor. It must be enhanced.
– Please correct the grammatical error in the sentence ‘’measurement is It is more stable’’.
– This reference did not appear in the manuscript ‘’Shui, W.Y.; Zhou, M.Q.; Maddock, S.; He, T.P.; Wang, X.C.; Deng, Q.Q. A PCA-Based method for determining craniofacial relationship and sexual dimorphism of facial shapes. Comput. Biol. Med. 2017, 90, 33-49’’.
Author Response
Please see the attachment. And our modifications/additions to the original are given in yellow text in the submitted manuscript.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsPaper is OK
Reviewer 4 Report
Comments and Suggestions for AuthorsThe authors have enhanced the manuscript as sugested by this reviewer.