Review Reports

Round 1

Reviewer 1 Report

eq. 2 lack of definition for L ;

eq. 3 lack o definition for k 

For more comfortable reading, the subsection 2.1-2.4 probably could be described as original SIFT algorithm (in conventional sense).

Then authors may show their algorithm also as algorithm with highlighting their new parts.  

Line 151: "In order to reduce the error and matching time, a stabilizer P is added, and then ...". Probably P should be here with some definition. 

Line 156: (eq 12) This equation need to have some justification or reference for it. I'm wondering how this expression was obtained.

Lines 158-161: The Authors need to review the estimations of computational complexity for conventional SIFT 

Lines 161-171: The Authors should calculate computational complexity for their variant of SIFT.

Lines 163-166: I recommend splitting the complex sentence (and this remark related to  the all paper)

Line Fig.11 this fugues should have axes labels. 

section 4: probably it would be better to introduce table with results for traditional and proposed implementations. 

section 4:  If authors talking about speeding up time, then they need to compare both algorithm working time.

Legion Y7000p is a laptop? Clarify this.

Conclusion should contain some numerical results. 

 

 

  

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper proposes an improved SIFT algorithm that allows you to increase the stability factor when constructing a scale space, eliminate the points of coincidence of unstable points and speed up image processing. The article has a scientific novelty and is of practical interest for a number of tasks, including tasks related to computer vision. The article is well written. As a remark, I would like the authors to pay more attention to the description of the data on which this algorithm was tested. It is not clear why the current dataset was chosen and how the model will work with other images. In addition, I would like to see the rationale for the metrics that were used to test the quality of this algorithm.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper represents a scale-invariant feature transform (SIFT) depending on the stability factor for image feature matching.

The authors firstly change the scale space calculation method, add a stability factor, and reduce the matching error and calculation time. Then, by establishing the feature descriptor of the

cross-shaped partition, the dimension of the descriptor is reduced from 128 to 96, which

reduces the calculation amount of the matching and shortens the matching time.

 

Strength:

1. Important topic for research.

 

Weakness and comments:

1. In abstract: Typos and what is SIFT? “dimension, Reduce”

2. In Introduction: This sentence does not make sense, “But the real-time problem of SIFT algorithm is still a research hotspot.”

3. DoG was never defined.

4. Missing references for Equations 1 and 2.

5. This paper missed some key references, “Hossein-Nejad, Z., Agahi, H. & Mahmoodzadeh, A. Image matching based on the adaptive redundant keypoint elimination method in the SIFT algorithm. Pattern Anal Applic 24, 669–683 (2021). https://doi.org/10.1007/s10044-020-00938-w.”

6. This paper must be compared with the state-of-the-arts, for example, “Hossein-Nejad, Z., Agahi, H. & Mahmoodzadeh, A. Image matching based on the adaptive redundant keypoint elimination method in the SIFT algorithm. Pattern Anal Applic 24, 669–683 (2021). https://doi.org/10.1007/s10044-020-00938-w.”

7. This paper must be checked for typos and grammatical errors.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

The revision has addressed all my comments.