Application of Digital Image Analysis to the Prediction of Chlorophyll Content in Astragalus Seeds

Round 1

Reviewer 1 Report

The paper describes the way how image analysis techniques can be used to predict the chlorophyll content in Astragalus using both MLR and MLP. The procedure and the results are presented well. However, it seems that the paper is a lack of justifiable claims for academic contribution in both theoretical and applicational senses. Virtually there is no theoretical advances in terms of technics applied in the research. It seems that the paper might have some advances in terms of the application of the existing approaches to a new field. However, the way the paper is described is too plain without detailed and justifiable reasons behind. For example, it is critically important for the authors to provide the reasons why R and g values are more correlated to the chlorophyll content in Astragalus. It may be because chlorophyll is more reactive or sensitive to a certain wave-length of light spectrum (like the filters used in Jalink, etc (1998) (18) in the reference list))? Also it requires to provide more relevant data to claim why R and G are more correlated to chlorophyll content compared to other colour bands or colour indices. The paper can be  much improved with more data and more detailed justifications.    

Page 3: Line 02: Why the seeds are divided to 5 groups based on only R and G values? Is there any reasons behind? Are these two colours related to the maturity of the seeds? Explain the reasons why you have used only R and G to divide the seeds in different groups.

 

Page 3 Line 19: the author mentioned “quality of hard seeds”. What is the exact meaning of “QUALITY” in this context? Is it related to the germination rate or what else? Explain it.

Page 11 line 99: how differentiate immature and high-mature seeds? Is it based on the time-lapse or by visual inspection? Explain.

Page 11: According to the results presented in the paper, fundamentally, the proposed scheme will not work for mature seeds. How the user can make a decision whether the seeds are mature since the proposed scheme will work only for immature seeds. Also as a practical issue, in actual agricultural field, mature seeds are more commonly used than the immature seeds for germination. Is that correct? If so, how the authors will justify the proposed scheme since the scheme will work only immature seeds (it will not work for mature seeds, so the proposed scheme cannot be used for the practical use). One more, what this “highly” matured seeds mean(line 99 page 11)? What would be different from “highly” matured seeds and “low” matured seeds. How different this “low” matured seeds from “immature seeds”? Clear explanation on how to differentiate one from the other.

Also it is also necessary to have a graph like Figure 1 in the following paper to show the relation between the maturity and G or R values.

“Possibility use of digital image analysis for the estimation of the rapeseed maturity stage”, Małgorzata Tańskaa, Marta Ambrosewicz-Walacikb, Krzysztof Jankowskic, and Daniela Rotkiewicza, INTERNATIONAL JOURNAL OF FOOD PROPERTIES 2017, VOL. 20, NO. S3, S2379–S2394 https://www.tandfonline.com/doi/pdf/10.1080/10942912.2017.1371188

 

Also, the author describes that there are a strong linear relationship between R and G values with chlorophyll content with high R^2 value. How about between other colour components, for example H (in HSV model) or a, b (in Lab) model? Have the authors tried normalised rgb or colour indices like Difference Green-red index (DGRI), Normalised Difference Index (NDI), Excessive Green Index (EGI), Modified Excessive Index (MEI), etc . (There are many other colour indices)?

One additional graph is required to show the level of R value and the status of germination of seeds. It seems that there should be a clear boundary line between germinated seeds and non-germinated seeds in terms of their respective R-values.

One additional minor comments: 

Page 1, It would be better to remove the statements in Line 34-39. The effects of Astragalus are not relevant to the main study of this paper. So it is better to be removed and focus on the approaches and results obtained from the main research activities conducted in this specific research.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This paper proposes to apply machine vision to measure the chlorophyll content of seeds instead of Chlorophyll fluorescence. Although the problem is interesting, the methodologies are already widely applied to image processing problems. A few suggestions are provided for reference. 1. The use of "machine vision technology" in the title is not appropriate. Because the image is captured by Scanmaker i360/i460 scanner, there is no emphasis on the use of relevant machine vision technology, such as light sources and lens. 2. Need to supplement more articles about image-based plant identification or measurement. 3. Lines 102-106 need to be analyzed by data or graphs to illustrate the results of using R and G into five groups. 4. There are many ways to reduce the dimensionality; why choose PCA and compare it with other methods 5. Please define all feature variables to quantify the original image and why these feature variables are chosen. 6. Comparing the results of building the model directly without performing PCA? 7. Table 2 provides the p-values for the Durbin-Watson test? 8. Must add the residual analysis results for the multiple linear regression 9. In Table 4, the results are not described in detail, and more information is needed on the meaning and comparison of the data. 10. The model created is the image captured by the Scanmaker i360/i460 scanner; please verify that if you use other scanners, the model can also use it.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

It has improved as suggested.

Reviewer 2 Report

The author responds to the suggestions to the original article and has made corrections and additions to the content. The quality and readability of the article have been greatly improved, and I recommend this article for publication.