Color Classification and Texture Recognition System of Solid Wood Panels

Round 1

Reviewer 1 Report

Comments in the PDF file

Comments for author File: Comments.pdf

Author Response

 

Point 1: This info is not wrong. But, it is misleading. All solid wood panels contain no formaldehyde? be specific.

Response: According to the reviewer’s suggestion, we have modified this part.

 

Point 2: I am not a native speaker. But to me, spliced refers to ropes. Perhaps, apply a different choice of words.

Response: According to the reviewer’s suggestion, our article has been polished by native English experts. And referring to other literatures, it is found that relevant words occur in similar application occasion.

 

Point 3: "cannot meet societies needs in terms of processing quality". This is misleading. For many years, the industry relies on this type of observation. Perhaps, the author could change the choice of words.

Response: According to the reviewer’s suggestion, relevant words have been changed in line 49.

 

Point 4: Line 49: Change “In a study...” into “In study…”

Response: According to the reviewer’s suggestion, we have modified this part.

 

Point 5: Double check if all of these letters have previously been defined.

Response: According to the reviewer’s suggestion, we have been defined these letters in lines 92-93.

 

Point 6: Lines 77-81: Reword this sentence.

Response: According to the reviewer’s suggestion, we have been reworded this sentence in lines 91-95 and 98-101.

 

Point 7: Crucial abspect missing. Which species are the authors using?

Response: According to the reviewer’s suggestion, we added usage of species in lines 23 and 109.

 

Point 8: Briefly describe the transmission belt, role of control panel. They were missing in the paragraph below.

Response: According to the reviewer’s suggestion, we described the transmission belt and role of control panel in lines 133-138.

 

Point 9: Line 106: Missing light source information

Response: According to the reviewer’s suggestion, we added missing information in lines 131-132.

 

Point 10: Some references here are needed. Mainly about canny method. The opencv python library has some useful info, I believe. In addition, it is very useful if the author provide the arguments used to procees the image.

Response: According to the reviewer’s suggestion, we added some references to canny method in lines 150-152. According to the reviewer’s suggestion, we provided the arguments used to process the images in lines 147-150.

 

Point 11: Not sure if this was defined. I know it means red, green, blue.  But keep it consistent.

Response: According to the reviewer’s suggestion, we have been defined these letters in lines 92-93.

 

Point 12: More information is needed on how the feature have been extracted and what they look like. Only saying that 27 features were obtained is not enough to replicate and explain the results and conclusions.

Response: According to the reviewer’s suggestion, we have been described the method to extract color features in lines 164-168 and Figure 3 described what they look like(Color histogram of beech panel in RGB, HSV and Lab color space.).

 

Point 13: This score is based on Rousseuw 1987 or 1986. I am not sure. But a citation is needed here.

Response: According to the reviewer’s suggestion, we added the reference to the core index of the contour coefficient method in line 230.

 

Point 14: Why not the elbow method?

Response: Both the contour coefficient method and the elbow method can be used to determine the K value of K-means clustering problems. These two methods have the same function and goal. Contour coefficient method, which combines two factors of cohesion and degree of separation, is an evaluation method of clustering effect. In this study, the contour coefficient method was selected to determine the K value of K-means clustering of beech wood panels, and good results were obtained.

 

Point 15: K-means is one of the many clustering techniques for clustering. Why did the authors chose specifically K-means? Birch technique can provide the number of cluster in an unsupervised manner.

Response: The K-means clustering method is widely used and is stable and effective. K-means clustering algorithm has the advantages of easy implementation, fast convergence speed and excellent clustering effect. In addition, the main parameter that needs to be adjusted in the experimental process is only the number of clusters K. In this study, not only has color classification been completed, but also texture recognition has been realized. The experimental results verified the effectiveness of K-means clustering algorithm in beech wood surface color classification. Next, we will further study and analyze how birch technique determines the number of cluster centers in unsupervised learning.

 

Point 16: But, this should happen. Why not try to reduce the amount of features with PCA, then compare that with T-SNE, and then cluster to see each one can separate bettter?

Response: We use the three color features mentioned in this paper to extract feature vectors for unsupervised clustering analysis and realize data dimensionality reduction. Through comparative experimental analysis, we can find which color features are suitable for color classification and which color features are conducive to texture recognition. Using this method, not only the computational difficulty can be reduced, but also the color features suitable for color classification and texture recognition can be found.

PCA is generally not directly used for feature extraction, but for dimensionality reduction of feature matrix. PCA looks for the principal axis direction used to effectively represent the common characteristics of the same kind of samples, which is very effective for representing the common characteristics of the same kind of data samples, but it is not suitable for distinguishing different sample classes. The idea of PCA is to map n-dimensional features to k-dimension (k < n), which is a new orthogonal feature. The new main components obtained by PCA cannot be explained, and its meaning is often vague, it is not convenient to determine the composition applicable to color or texture.

T-SNE is too slow and not suitable for large-scale computing or big data. For a dataset with 1800 high pixel images, the processing speed will be very slow. The results of t-SNE have some randomness. In addition, the robustness and consistency of t-SNE are not very good.

Our purpose is to extract features, realize data dimensionality reduction, clarify which features are suitable for color classification or texture recognition, and finally realize sorting. Based on the above analysis, the scheme proposed in this paper is more conducive to the color sorting of solid wood panels.

 

Point 17: Another reason for trying to reduce the dataset dimension.

Response: According to the reviewer’s suggestion, we described the reasons for trying to reduce the dataset dimension in lines 406-407.

 

Point 18: Line 281: The evaluation of our work is ”Well, yes...”.

Response: Thank you very much for your affirmation of my work.

 

Author Response File: Author Response.docx

Reviewer 2 Report

I had the opportunity to review the manuscript "Color Classification and Texture Recognition System of Solid Wood Panels"

The topic and the manuscript is very interesting, the manuscript needs some revisions before acceptance:

Abstract: Please be more specific about your results.

Introduction:

Line 30: Please reformulate the sentence, it is not completely true (for example formaldehyde is a naturally occurring chemical in wood).

Lines 32-34:  Please reformulate this sentence, there are other excellent properties more important than these.

Lines 34-36: you are talking about esthetical attributes of wood, please modify it (there are plenty of other "surface parameters" like roughness, ...)

Lines 43-46: There is plenty of research dealing with a machine vision system for the color of the wood and wood-based panels, please check for example here:

https://ieeexplore.ieee.org/abstract/document/668534

https://doi.org/10.1007/11558484_20

Lines 46-49: This part should be moved to Materials and methods or as an Aim at the end of the Introduction section.

Lines 72: Please add also results of research using K-means clustering in case of wood color, check for example here:

doi.org/10.3390/app10228113

doi.org/10.1007/978-3-030-51965-0_41

10.1109/NGCT.2015.7375241.

doi.org/10.3390/app10196816

Material and Methods: Please start with materials (samples) and then methods used 

Lines 136-138: This part belongs to the Introduction section.

Lines 144-148: The aim should be mentioned at the end of the Introduction section.

Lines 348-354: Please add more discussion with other research in this area ( machine learning area - to classify wood, for example using a binary pattern, Hough transform method, wavelet and local binary pattern method, etc.).

Conclusions: Please add limitations of your research and direction of further research.

 

 

 

 

 

Author Response

Point 1: Abstract: Please be more specific about your results.

Response: According to the reviewer’s suggestion, we revised the description of the results in the abstract section in order to be clear and specific.

 

Point 2: Line 30: Please reformulate the sentence, it is not completely true (for example formaldehyde is a naturally occurring chemical in wood).

Response: According to the reviewer’s suggestion, we revised this sentence in lines 30-33.

 

Point 3: Lines 32-34:  Please reformulate this sentence, there are other excellent properties more important than these.

Response: According to the reviewer’s suggestion, we revised this sentence and choose other words to describe the properties of wood panels.

 

Point 4: Lines 34-36: you are talking about esthetical attributes of wood, please modify it (there are plenty of other "surface parameters" like roughness, ...)

Response: According to the reviewer’s suggestion, we added more surface parameters to describe the surface properties of wood panels more comprehensively.

 

Point 5: Lines 43-46: There is plenty of research dealing with a machine vision system for the color of the wood and wood-based panels, please check for example here:

https://ieeexplore.ieee.org/abstract/document/668534

https://doi.org/10.1007/11558484_20

Response: According to the reviewer’s suggestion, we added other research papers about wood color in lines 51-56.

 

Point 6: Lines 46-49: This part should be moved to Materials and methods or as an Aim at the end of the Introduction section.

Response: According to the reviewer’s suggestion, we moved this part into the end of the Introduction section as an aim.

 

Point 7: Lines 72: Please add also results of research using K-means clustering in case of wood color, check for example here:

doi.org/10.3390/app10228113

doi.org/10.1007/978-3-030-51965-0_41

10.1109/NGCT.2015.7375241.

doi.org/10.3390/app10196816

Response: According to the reviewer’s suggestion, we added some references to K-means clustering in lines 80-86.

 

Point 8: Material and Methods: Please start with materials (samples) and then methods used. 

Response: The steps of this experimental study are as follows: 1. Use the image acquisition system to collect the image of solid wood panels; 2. Remove the background of the collected solid wood panel images; 3. After analysis, select the appropriate color spaces and color features to extract the feature vectors; 4. Extract feature vectors and complete clustering using k-means. The Material and Methods section follows this logical order.

 

Point 9: Lines 136-138: This part belongs to the Introduction section.

Response: According to the reviewer’s suggestion, we moved this part into the Introduction section in lines 95-97.

 

Point 10: Lines 144-148: The aim should be mentioned at the end of the Introduction section.

Response: According to the reviewer’s suggestion, we mentioned our aim at the end of the Introduction section as suggested.

 

Point 11: Lines 348-354: Please add more discussion with other research in this area ( machine learning area - to classify wood, for example using a binary pattern, Hough transform method, wavelet and local binary pattern method, etc.).

Response: According to the reviewer’s suggestion, we added more discussion with other research in machine learning area at the end of Discussion section as suggested.

 

Point 12: Conclusions: Please add limitations of your research and direction of further research.

Response: According to the reviewer’s suggestion, we added limitations of our research and direction of further research at the end of Conclusions section as suggested.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

I recommend including a clear overall objective, hypothesis, and specific objectives. You may say, yeah, we include that in lines 150 to 152 or even before that. If that is the case, the overall objective, hypothesis, and specific objectives of this research remain totally unclear. 

When I read the materials and methods, I read MATERIALS. We don't have that. We jump straight to methods (Imaging).

The authors added the common name of the species used in this work. But no other information is given. If the information is given, it is buried somewhere in the text that makes the reading difficult to follow.  

Describe the solid wood panel that you talk about in lines 155-156. What's the moisture content of the "solid wood panels?" Are they commercially available? What are the dimensions? Panels mean a lot of different things. 

Author Response

Point 1: I recommend including a clear overall objective, hypothesis, and specific objectives. You may say, yeah, we include that in lines 150 to 152 or even before that. If that is the case, the overall objective, hypothesis, and specific objectives of this research remain totally unclear.

Response: According to the reviewer’s suggestion, we have supplemented this part in lines 107-126.

 

Point 2: When I read the materials and methods, I read MATERIALS. We don't have that. We jump straight to methods (Imaging).

Response: According to the reviewer’s suggestion, we have added the description to the materials in lines 128-136. 

 

Point 3: The authors added the common name of the species used in this work. But no other information is given. If the information is given, it is buried somewhere in the text that makes the reading difficult to follow.

Response: According to the reviewer’s suggestion, we have supplemented the other information about the species used in this work in lines 128-136.

 

Point 4: Describe the solid wood panel that you talk about in lines 155-156. What's the moisture content of the "solid wood panels?" Are they commercially available? What are the dimensions? Panels mean a lot of different things.

Response: According to the reviewer’s suggestion, we have supplemented this part in lines 128-136, added these information about the materials used in this work.

Author Response File: Author Response.docx

Reviewer 2 Report

Dear Authors, thank you for the revisions based on my suggestions. 

Please check some more suggestions:

Line 30-31: This sentence is too general, I suggest omitting it.

Line 36: degradability - you have meant durability?

Lines 97-102: Please also explain with one sentence how the K-means clustering algorithm mathematically works (clustering the data into K-clusters based on K-centroids, etc.), please check here:

doi.org/10.3390/app10228113

Materials and methods:

Please add chapter 2.1 Materials and be precise about samples: source of wood, wood species used, samples cutting, the orientation of samples, surface finishing?, samples climatization, etc.

Lines 495-510: Please compare your research method and results with other research provided in this area (scientific discussion).

 

 

 

Author Response

Point 1: Line 30-31: This sentence is too general, I suggest omitting it.

Response: According to the reviewer’s suggestion, we have omitted this sentence.

 

Point 2: Line 36: degradability - you have meant durability?

Response: According to the reviewer’s suggestion, we found the word inappropriate and have modified this part.

 

Point 3: Lines 97-102: Please also explain with one sentence how the K-means clustering algorithm mathematically works (clustering the data into K-clusters based on K-centroids, etc.), please check here:

doi.org/10.3390/app10228113

Response: According to the reviewer’s suggestion, we added other research paper about how the K-means clustering algorithm works in lines 101-103.

 

Point 4: Please add chapter 2.1 Materials and be precise about samples: source of wood, wood species used, samples cutting, the orientation of samples, surface finishing?, samples climatization, etc.

Response: According to the reviewer’s suggestion, we have supplemented this part in chapter 2.1, added these information about the materials used in this work.

 

Point 5: Lines 495-510: Please compare your research method and results with other research provided in this area (scientific discussion).

Response: According to the reviewer’s suggestion, we made comparison at the end of Discussion section.

 

Author Response File: Author Response.docx

Round 3

Reviewer 1 Report

My hope is that the authors have gained some experience with this review. Now, it is clear to me why the authors pursued this area. 

Great work!

Reviewer 2 Report

The manuscript was improved. I suggest to accept it.