Effect of Image Segmentation Thresholding on Droplet Size Measurement
Round 1
Reviewer 1 Report
The particle size of droplets is one of the main factors affecting the deposition and drift of pesticide spraying. Reasonable droplet size selection can help to improve the utilization rate of pesticides, reduce the use of pesticides and improve the control effect of crop diseases and insect pests. Therefore, the Image Segmentation Thresholding method proposed in this paper is very meaningful to test the particle size of droplets.
but there are still some aspects of this article that need to be revised before official publication.
Some terms need to be unified，such as, Line45, the authors use "droplet size" while Line 50,51 use "drop size", Line 42 "drop size spectrum" while Line 54 " droplet size spectra". Check the similar all through the manuscript.
Line76,77: "...but could not be used to quantify the amount of spray deposits in most field applications."，Such a statement is inaccurate and imprecise.
In "the image acquisition system", a photo of the complete test bed should be provided, and each part should be named inthe photo.
In test spraying, the flow rate should be given. Which type nozzle was used ? How to control the nozzle move speed ?
"...is managed by a suitable software",which software? Commercial or self-made software ？
The average calibration factor ?? was 5.33 um/ pixel-1. How to get it ?
Trial design details need to be clearer.
Line203: Subsequently, the threshold values TV?, chosen for this segmentation process, were correlated to the average gray level (AGL) of the corresponding images. This allowed to obtain a regression line highly significant (R2 = 0.922, p-level < 0.001), suggesting the possibility to automate the selection of the threshold in function of the average gray level of the images, at least for the image acquisition system used in this study. What is the regression line? What do you meas "suggesting the possibility to automate the selection ..."?
Relative Span Factor， I think it is Relative Droplet Spectral Width.
Line254: R fucntion, it is suggested to supplement the function.
Line271: the RSF did not show a increasing trend as from1.07 at 1.0 Mpa to 1.15 at 1.5 Mpa. This sentence is wrongly descriped. Please explain the possible reasons.
All pictures in the text need to be replaced with higher-resolution ones.
Author Response
Please see the attachment.
Reviewer 2 Report
Review
Manuscript ID: agronomy-1752613
Title: Effect of Image Thresholding on Drop Size Measurement
Authors: Emanuele Cerruto, Giuseppe Manetto, Salvatore Privitera, Rita Papa, Domenico Longo
General comments
The authors of the manuscript extracted the features of spray drops generated by an air-induction nozzle from the image of these drops immersed in the silicone oil. Such extraction is used to measure the size of drops, and thus evaluate and classify the spray quality. The accuracy of the drop size measurements based on the image segmentation depends on proper selection of the segmentation thresholds. The this study the authors attempted to compute the image segmentation threshold based on the relationship between the average gray level of the image and the reference threshold value determined by the operator and considered “optimal”. Such approach to threshold computation can facilitate the automation of the image segmentation process, and hence ensure obtaining more objective results of drop size measurements. Testing this approach was an actual objective of the reported study.
By using proper methodology of droplet collection, sampling, image acquiring, and by analysing the obtained data with an interesting statistical approach the authors showed that the variations in the volumetric and numeric diameter of droplets generated at different pressures (resulting in different drop size classes) for thresholds varied in the range of +/-5% of the computed threshold value were so low that their effect on the drop size measurement error may be neglected. They reasonably conclude that the robust threshold value computation based on the average gray level of the image can allow automation of the image segmentation and ensure better objectivity of the drop size measurements. The objective of the research was thus met, valuably contributing to methodology of the drop size measurement.
The manuscript is written in a clear language, properly structured and the findings are well supported statistically. The minor specific comments and suggestions are given below.
I do recommend publishing the manuscript.
Specific comments and suggestions [line number]
[2] Title: I am also fan of short titles but suggest adding just one word to read: “Effect of Image Segmentation Thresholding on Drop Size Measurement”.
[26; 121, 123] Replace “pulverization” (referring to powders) by “atomization” (referring to spray).
[110-113] “In this research (… ). Images of drops (…). ” – the information in these two sentences is in M&M section where it belongs. Delete them to avoid unnecessary repetitions.
[124] “The test liquid (…) is taken and sent to the nozzle …” – suggest the rewording: “The spray liquid (…) is pumped to the nozzle …”.
[127] Suggest replacing “pressure regulator” by “pressure valve”.
[131] Suggest replacing “flow transducer” by “flowmeter”.
[154] The mentioned “average calibration factor Cf” is worth one or two explanatory sentences.
[173] “… touching the edges.” – of what? (image?, Petri dish?)
[182-184] “A similar result ...” – this belongs to the section DISCUSSION and actually it is there [in 350]. Delete the sentence.
[224] Table 1: Complete “Spray diameter …” by “Spray droplet diameter …”.
[227] Replace “… drop pulverization also increased” by “… drop size decreased”
[230] Replace “numeric” by “number”
[399] “Error was affected by pressure values” – the real factor affecting the error was drop size which in this particular study was modified by pressure, and which might also be altered by nozzle type, nozzle size or spray liquid properties (adjuvants). Suggest replacing by “Error was affected by drop size modified by spray liquid pressure”.
[409] The statement on “statistical significance” here is inconsistent with the one in 316-317 about “most cases (10 out of 16) (…) not statistically significant” ?
Also:
Suggest using brackets [] for units: [MPa], [pixel] in tables, figures, 211 and 212, and leave parentheses () for supplemental information, equations, years of references, …
Suggest using a scientific SI notation of units: m s-1 [in139, 159]; g L-1 [in 141] or µm per pixel [in 155].
Author Response
Please, see the attached file
Author Response File: 
Author Response.pdf
Reviewer 3 Report
1. Droplets size is an important index for estimating the performance of equipment for crop protection. Not only the specific substance ( in this paper, clean water with soluble colouring agent Poinceau Red, Novema Srl, Turin, Italy, at the concentration of 2 g/L) used in the spray system affect the droplet size, the main factors affecting the size of the droplet are the nozzle type, spray pressure，flow rate and and spray shape. Only four pressures at a fixed speed is not sufficient to illustrate the actual situation, and the choice of pressure levels is not given.
2. The distance and height of sampling will also affect the droplet size, only use the fixed distance ( 0.5 m between nozzle and Petri dishes) is not sufficient.
3. Reasons for choosing intrusive techniques and liquid immersion method should be stated clearly.
4. Average gray level and reference threshold value are classical and commonly used binarization methods, through increasing the number of iterations to improve segmentation accuracy is not an algorithmic innovation.
Author Response
Please, see the attached file
Author Response File: Author Response.pdf
Reviewer 4 Report
Good scientific approach to the problem with adequate conclusions.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 5 Report
General
The conclusions are interesting for future research and practical activities!
The manuscript could bring interesting contributions for integrated approach technology-machinery-environment, for all activities regarding plant protection, and I agree on its publication.
However, there are few issues that must be addressed to necessary improve the overall communication of author`s work:
· The methodology and results could be seriously improved because it seems to be confused and not at all synchronized.
· Please, use the SI units and consequence in concepts.
· Regarding contributions, for increasing of validity it is necessary to correct and synchronize the methodology and results, also title, objectives, discussions, and conclusions.
Abstract must include small ideas, correlated, about objective and method, results, and discussions.
Introduction
Must be consequent and correlated to method and results.
Material and methods
The chapter must be seriously reviewed and correlated with the results, title and objectives (also, discussions, conclusions and key words).
Keep in mind that the readers should find here detailed description of all your procedures and method (author, process, conditions, etc.) used, in correlation with results and discussion. Anybody must be able to repeat your methods and obtain the same results.
Also, starting at these reasons, I supplementary suggest next:
- Fig 1 – there are necessarily supplementary explanations regarding parts, and ensembles, together with functional description
- Row 147-150, please, use SI units
- Too many justifications regarding adopted methodological solutions, especially in cap. 2.2
- Row 240, why was choose „Feret’s diameters ratio greater than 1.50 were ignored”?
- In equation (4), something is wrong (I suppose is cubic root), (row 253).
Results
Please, respect the consequence in expression of measurement units (example Fig. 4, etc.).
Also, attention to the quality of figures!
I repeat methodology, results, title, and objectives must be concordant.
Figures must be more clear for readers.
Discussion
Must be focused on title, objectives, methodology and results, and must compare own results with others similarly. Sometimes seems to be confusion between discussion and introduction.
Must be reorganized. Based on results, please show more self-criticism to your methods, discuss all limitations of your results
Conclusions
OK, clearly indicate whether the research hypothesis tends to be confirmed or not, highlight/quantify the industrial/environmental significance.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
The main factors affecting the size of the droplet are the nozzle type, spray pressure flow rate and spray shape. Only four pressures at a fixed speed is not sufficient to illustrate the actual situation. Even if the main objective of the manuscript is to assess how the choice of the threshold used for image binarization affects the results in terms of numeric and volumetric diameters computation. The discussion has practical scientific significance only in the context of well-established experiments.
During pesticide application, the ideal distance between nozzles and target is around 0.5 m, but the distance deviation is large due to the growth of the crop itself in actual situation. Therefore, distance should also be considered in the experimental setup.
This research assessed how much variations of about ±5% in the choice of the segmenting threshold affect the computation of volumetric and numeric diameters. Now there are many mature binary segmentation methods, there has no reference and no comparison in this paper.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
