Multi-Scale Cyclic Image Deblurring Based on PVC-Resnet

Round 1

Reviewer 1 Report

This study provides image deblurring method based on deep learning. The structure of the proposed paper is clear which is easy to read and understand, but there are still some points which should be improved.

1、        Check English, especially typos and grammatical errors, which should be improved.

2、        Some provided figures are too small e.g. fig5, and high resolution figures are suggested.

3、        Lines 114-116, you mentioned that small targets in blurred image can be deblurred by using the proposed method. Only small targets can be deblurred? What about other targets in the blurred image?

4、        Line 198: The multi-scale feature extraction module designed in this paper is used to extract shallow features of objects of different sizes in blurred images. Explain how to achieve multi-scale feature extraction?

5、        The format of the model and evaluation indicators in Tables 3 and 4 is incorrect. Check and correct it.

6、        Why do you use the calibration board data set to verify image deblurring?

7、        You mentioned ringing artifact in the study. Explain it.

 

 

 

This study provides image deblurring method based on deep learning. The structure of the proposed paper is clear which is easy to read and understand, but there are still some points which should be improved.

1、        Check English, especially typos and grammatical errors, which should be improved.

2、        Some provided figures are too small e.g. fig5, and high resolution figures are suggested.

3、        Lines 114-116, you mentioned that small targets in blurred image can be deblurred by using the proposed method. Only small targets can be deblurred? What about other targets in the blurred image?

4、        Line 198: The multi-scale feature extraction module designed in this paper is used to extract shallow features of objects of different sizes in blurred images. Explain how to achieve multi-scale feature extraction?

5、        The format of the model and evaluation indicators in Tables 3 and 4 is incorrect. Check and correct it.

6、        Why do you use the calibration board data set to verify image deblurring?

7、        You mentioned ringing artifact in the study. Explain it.

 

Author Response

Please refer to the attached document.

Author Response File: Author Response.pdf

Reviewer 2 Report

I have the following comments.

1. It is known that in addition to the SSIM and PSNR metrics, MSE is used to compare images. It is also known that to restore blurred images it is better to use the Stat-SSIM metric, which gives better results compared to SSIM.

2. There is no need to quote metrics (4), (5), (6), because these are known and standard equations.

3. Image blurring results in noise. Noise can be additive or multiplicative. How does your model remove this nature of noise and how will it work on low contrast blurred images.

4. Comparison with different deblurring models shows a slight improvement. Therefore, for comparison, it would be interesting to evaluate the benefits in terms of runtime.

5. In References, the works of R.Vel, Gu S.H, K.Zang and others should be mentioned and supplemented for 2022-2023.

 Minor editing of English language required

Author Response

Please refer to the attached document.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

I am satisfied with the responses to my concerns. The changes and additions made have improved the perception of the article.

Minor editing of English language required