Visual Detection of Surface Defects Based on Self-Feature Comparison in Robot 3-D Printing
Round 1
Reviewer 1 Report
The paper presents a self-feature extraction method of shape defect detection of robot 3D printing products, which distinguish the defects and their own characteristics by comparing the theoretical projection contours and the experimental contours of products. That is, the different and their characteristics are identified by comparing the parameters of contours in the picture captured from the FDM platform and the picture projected from the theoretical model. The proposal itself is interesting, although the content of the article should be reinforced in many points that I detail below.
Point 1. The introduction is very short: it is limited to presenting the work developed, when an introduction should talk about the state of the art. I recommend that the authors provide previous results or research related to the work carried out. This will result in a greater number of references: it is striking that there are only seven. I also invite the authors to explain the need to do this control, that is, its possible application to quality control processes, for example.
L 31-32 There is a repetition between the abstract and the Introduction: “However, there are many defects that may affect the surface quality, accuracy, or even cause the collapse of the parts in the printing process”
L 33: The authors speak of "our team", but do not explain which team it is. I recommend explaining it before or writing it in another way.
Point 2: I recommend the authors to make a more detailed description of the software and hardware used in the research.
Point 3:
L 104: “According to the principle of industrial CCD camera imaging, the coordinates of projection point is determined by the coordinates of the original point and the coordinates and parameters of the camera.” Regarding this, the authors could provide an appointment to justify it.
Point 4:
Given the importance of the formula used to evaluate the contours similarity, I invite the authors to justify the reason for that particular algorithm, that is, justify its use.
In Table 3, the similarity of contours parameters could be expresed in %, in order to be more understable for the reader/user. Also, in table 3, the authors wrote “the red Font means…” ans there is no red fonts in table 3. The authors could also write about the accuracy range (in millimeters) that the system (hardware, software and comparison algorithm) is able to detect.
Point 5: The conclusions are not conclusions themselves. In the conclusions the authors limit themselves to describing the work done, without providing conclusions based on results. I recommend that authors completely rewrite this point and provide concrete conclusions based on data analysis, results of the application of a methodology. The authors should provide, in turn, data on the reliability of the methodology carried out. They could also provide possible future applications.
Once these recommendations have been taken into account, I propose a change in the structure of the article in order to follow a structure of Introduction, methodology, results, data analysis and conclusions. Thus, I propose to change the title of point 2 and call it "Methodology", where to make a brief description of the methodology carried out. Then, make two sub-headings 2.1 and 2.1 with "Identifying defects and their own characteristics" and "Self-feature extraction of model itself based on the location relation of points cloud".
Point 4 would be like that point 3: Evaluation of contour similarity based on empirical formula. And the conclusions would be point 4, where I propose to add "Conclusions and future work" or "Conclusions and fields of application"
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The topic of the article is relevant. 3D printing is widely used in many fields, from diffraction optics (A method for the diffractive microrelief formation using the layered photoresist growth / A.V. Volkov, et al // Optics and Lasers in Engineering, 1998, Vol.29 (4), 281-288) to medicine (Engineering anatomically shaped human bone grafts / W. L. Grayson, et al // PNAS - 2010 February 23. – 107(8). – P. 3299–3304; Development of Mesh‑Defect Removal Algorithm to Enhance the Fitting of 3D‑Printed Parts for Comminuted Bone Fractures / Irwansyah Idram, et al. // Journal of Medical and Biological Engineering (2019) 39:855–873, https://doi.org/10.1007/s40846-019-00477-8).
Unfortunately, the list of references does not prove the relevance of the topic of the article, since from recent publications the article contains only the article of the authors of 2018.
The article contains some shortcomings and inaccuracies that need to be corrected.
It is necessary to correct the formula (1): it is necessary to enter the multiplication sign ( * ) after ?n (since the same signs are inside the brackets) and make the final closing bracket of the same height and vertical arrangement as the opening one located after ?n.
The designations of formula (1) (lines 137-139) need to be corrected. Notations x. y, r, A are absent in the formula (1). In formula (1) there are notations xi. yi, ri, Ai (i=n, m), this should be specified in lines 137-139.
Formula (4-1) in article there is no, in it there is formula (1), need correct string 143.
It is necessary to expand the too short list of References to include recent publications:
Fei Chang et al 2019 Meas. Sci. Technol. 30 125905
Zhen Liu, et al. A Novel Stereo Vision Measurement System Using Both Line Scan Camera and Frame Camera. IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 68, No. 10, OCTOBER 2019, p. 3563.
Mingjun Ren, et al. Fast Defect Inspection Based on Data-Driven Photometric Stereo. IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 68, NO. 4, APRIL 2019, p. 1148.
Zakharov AA, Zhiznyakov AL, Titov VS. A method for feature matching in images using descriptor structures. Computer Optics 2019; 43(5): 810-817. DOI: 10.18287/2412-6179-2019-43-5-810-817
Chervyakov NI, Lyakhov PA, Orazaev AR. Two methods of adaptive median filtering of impulse noise in images. Computer Optics 2018; 42(4): 667-678. DOI: 10.18287/2412-6179-2018-42-4-667-678.
Abul’khanov S.R., et al. Information Pattern in Imaging of a Rough Surface. IOP Conf. Series: Materials Science and Engineering, 302 (2018) 012068 doi:10.1088/1757-899X/302/1/012068.
Domen Taberni, et al. Segmentation-based d eep-learning a pproach f or surface-defect detection. Journal of Intelligent Manufacturing, 2019, https://doi.org/10.1007/s10845-019-01476-x
Qinbang Zhou, et al. An Automatic Surface Defect Inspection System for Automobiles Using Machine Vision Methods. Sensors, 2019, 19, 644; doi:10.3390/s19030644.
Javier Villalba-Diez, et al. Deep Learning for Industrial Computer Vision Quality Control in the Printing Industry 4.0. Sensors, 2019, 19, 3987; doi:10.3390/s19183987.
Weiyi Lin, et al. Online quality monitoring in material extrusion additive manufacturing processes based on laser scanning technology. Precision Engineering, Volume 60, November 2019, Pages 76-84.
Zhiyuan Wang, et al. An accurate detection method for surface defects of complex components based on support vector machine and spreading algorithm. Measurement, Volume 147, December 2019, 106886.
Defu Zhang, et al. Unified detection method of aluminium profile surface defects: Common and rare defect categories. Optics and Lasers in Engineering, Volume 126, March 2020, 105936.
It is necessary to analyze recent publications and show the advantages of the proposed approach, for which tasks it is more effective.
Comments for author File: Comments.pdf
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have carried out almost all the recommendations I did in my review. The article has improved substantially. The introduction presents recent works related to the one carried out. The description of the hardware and software has been conveniently completed. Figure 5 has been introduced to explain the principle of industrial CCD camera imaging. Also, the results in% facilitate their understanding by the reader. Finally, changes in the structure of the article help to understand the sequence of work carried out. But I still miss that, in the conclusions, the authors provide quantitative results of the tests carried out that serve as an example to validate the methodology applied.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
