Multi-Feature Patch-Based Segmentation Technique in the Gray-Centered RGB Color Space for Improved Apple Target Recognition

Round 1

Reviewer 1 Report

The proposed method is interesting and fits into the scientific space represented by journal “Agriculture”. However, it requires a few clarifications, additions and possible corrections:

• On what basis was color considered to be a representative feature in the process of identifying information encoded in the form of digital images of apples? After all, for this purpose, descriptors such as: shape or texture indicators ...
• What is the advantage of the proposed identification technique in relation to the methods of neural image analysis?
• Indicate the source of the empirical data used
• Define the basic agrotechnical parameters of the tested apple images (variety, location of the plantation, etc.)
• Complete the description of the quantities used in the formulas, e.g. (1), (3).

Author Response

Dear Reviewer:

On behalf of my co-authors, we thank you very much for giving us an opportunity to revise our manuscript, we appreciate you very much for your positive and constructive comments on our manuscript entitled “Multi-feature Patch-based Segmentation Technique in the Gray-centered RGB Color Space for Improved Apple Target Recognition”. (ID: agriculture-1136311).

Author Response File: Author Response.docx

Reviewer 2 Report

This article purposes a new image segmentation technique applied in the identification of apple targets to be integrated in agricultural fruit-picking equipment (robot picking). According to the authors the method can effectively deal with the effects of illumination and shadows, under the complex natural environment in orchards  where changing illumination conditions are observed.  

This is a topic of special importance towards intelligent agricultural automation processes. However, the article needs to be improved in some aspects:

Major concerns:

1- Although in an abbreviated form, the authors should mention how the information of apple detection in the image coordinate system can be used to obtain the location of the fruit in 3D world coordinates, allowing fruit picking by the robot. This task is not trivial and some light should be shed on the process of inferring the position of the apple in the real world based on his image location.

2 - To evaluate the proposed method, 300 images of red apples were collected. In the abstract, they state that 180 images were used for testing. What criteria were used to choose the images used in the testing? However, in section 3, line 305, it is written that 100 images were used to verify the validity and reliability of the proposed apple image segmentation method. The authors must explain this apparent inconsistency in the number of images used in the tests. Finally, the collection of images in the orchard is an easy task and so the dataset must be increased to give reliability in the results.

Specific questions:

1- Line 158: What information are you referring in the sentence?

2 – Line 163: What is the meaning of “main subjects”? Do you mean “regions”?

3 - Is the graph of figure 3 obtained in any of the four cases of the apple surface appearance?

4 – Line 200: What adjusting do you refer? This sentence is unclear.

5 – Line 204: What is the meaning of “image after offsetting the shadow COI”?

6 - Line 223-244: Why and how the 10 COI were chosen?

7 - Line 234-235: The sentence is unclear?

8 – Line 294: How the threshold value B is determined? Is it a constant value?

9 – Line 394 : Is the image transformed to the grayscale? In the method description only RGB to HSV space transformation is referred.

Author Response

Dear Reviewer:

On behalf of my co-authors, we thank you very much for giving us an opportunity to revise our manuscript, we appreciate you very much for your positive and constructive comments on our manuscript entitled “Multi-feature Patch-based Segmentation Technique in the Gray-centered RGB Color Space for Improved Apple Target Recognition”. (ID: agriculture-1136311).

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors answered the main questions and provided clarifications to better understand the options taken in conducting the experiments. The article quality has improved and constitutes an important scientific contribution in the field.