Detection of Defective Features in Cerasus Humilis Fruit Based on Hyperspectral Imaging Technology
Round 1
Reviewer 1 Report
Please see the comments on the attached file.
Comments for author File: 
Comments.pdf
Author Response
Dear Reviewer,
We have made a point-by-point response based on your comments. Please see the attachment for details of the revise.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Comments on the work were sent in a PDF file.
Comments for author File: Comments.pdf
Author Response
Dear Reviewer,
We have made a point-by-point response based on your comments. Please see the attachment for details of the revise.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Review round 2
Detection of Defective Features in Cerasus Humilis Fruit Based on Hyperspectral Imaging Technology
Thank you for the answers to my questions and the corrections made to the article. The authors asked for suggestions for question 4 in my previous review.
Point 4: Line 406 - The authors summarize the work of the algorithm as follows:
For insect damage samples, the insect damage area was too small and reflective, so it was not completely identified. Two sound samples were mis-takenly detected as a rust spot fruit, the possible reason was the reflective and uneven colors of the skin of the fruit. Furthermore, the classification accuracy of cracked Cerasus Humilis fruits was the lowest, this was because the crack region in Cerasus Humilis fruits was rotten.
This raises the question for the authors about how to correctly change the imaging (image acquisition) system to detect defects related to insect damage and small cracks. I understand that an off-the-shelf solution from Zolix was used, but please suggest how to rebuild it.
Response 4: We appreciate the reviewer for the comment. At present, we have realized this problem. In view of this problem, our team is trying to adjust the image acquisition system, but we encountered some problems during the adjustment process, which led to no breakthrough in the research. Can we ask the reviewer to provide some guidance suggestions.
Reviewer Suggestions
I think you should start by specifying the imaging resolution. It is not a parameter related to the resolution of the matrix. This is an image-related parameter. It tells you what area of the real object is projected onto one image pixel. The imaging resolution is given in mm/pixel.
If the whole real surface defect is imageable on one or two pixels, we are not able to detect it well. In my experience, the defect should be visible on at least 3 pixels. Then the disturbances associated with blurring the edges of the defect do not hide the defect in the image. A better solution is the visibility of the smallest sought defect on at least 5 pixels. Then we are sure that the defect will be visible in the image.
Example resolution calculations can be found in the article https://www.mdpi.com/1999-4907/9/1/30
The quality of the article would be improved if information about the image resolution was added to the information about the image and how many millimeters of the fruit's surface correspond to 1 pixel of the image.
Author Response
Dear Reviewers,
We have replied point-by-point according to the reviewer's comments. Please see the attachment.
Author Response File: Author Response.pdf
