Detection and Imaging of Underground Objects for Distinguishing Explosives by Using a Fluxgate Sensor Array

Round 1

Reviewer 1 Report

1) Some grammatical errors (e.g. incorrect tense) are throughout the document.  These should be corrected prior to final acceptance. Some examples include:

 

Line 2: Change to ... broadcasting a signal underground Line 6: devices instead of device? Line 9: Change classification to classifying. Line 13:  eliminate "the" Line 26:  This sentence makes no sense -    In terms of safety, systems that classify all detected objects as explosives will reduce the speed of the operator [1]. Line 44: Change technique to techniques  and mass and trace analysis to mass analysis and trace analysis Line 47:  eliminate use; ergonomic? Line 54: MA acronym not defined until later in paper.

2) What is meant by "Low classification memory capacity in Line 31?

3)  Line 37:  Change BED's broadcast ... to BED system's broadcast...

4) Line 59:  How do distinguish between ferrous ores and man-made items with passive methods?

5) Line 69:  The authors need to elaborate on the k-nearest neighborhood (kNN) classification study.  Specifically, what is it, why is it relevant, how is it applied, and how do they know it gives consistently good and applicable results.  References should also be supplied.

6) Line 73:  Elaborate on what is meant by "negatively" in this sentence.  Is the author saying that they get different results based on the positioning of the sensor?  The authors state that a solution has been "proposed" in Line 74.  Why did they use the word proposed instead of presented?  Do they have a solution or just a proposal for a solution?

7) Line 76:  Should the acronym be (SA & IC) instead of (SN & IC)?

8) Line 78:  Remove "the" in front of Figure 1

9) Line 83:  Add in what the sensitivity in micro-Teslas the TE 100 sensors have.

10) Line 89:  What is meant by "...move in the land..."?

11) Line 91: "... a buried object.

12) Figures need more initial explanation in the text when they are first presented.  Subsequent discussion and elaboration is good.  For instance, what is the purpose of the circuit diagrams?  What do they do?  Where did they come from?

13)  In Lines 99 - 101, the authors claim that by looking at Figures 1, 4, and 5 the "sensor array is designed."   Where did this design come from?  How do they know it works?   What requirements does it satisfy (e.g. what sensitivity is needed to properly discriminate the buried IEDs)?  What about false alarms?  Not enough is said about the requirements and supporting analysis that the shown circuit achieves what is intended.

14) In Line 136:  what is meant by "certain but unclassified features"?

15) In Line 137:  How are the previously determined features obtained with respect to  explosives in various types of soils and other potential clutter such as ferromagnetic natural materials?  What is the expected variation in the Earth's magnetic field when these "clutter" items are present in the data set?   How do results change as a function of different soil types, temperatures, ground conditions, and clutter items present?

16) In Line 138:  What "small k-values" are being discussed?  This section needs references.

17) Line 145:  processing speed instead of processor speed?

18) Do the authors have analytical expressions for the Reliability Coefficient and probability of false alarm?

19) Line 148:  "firstly" to first

20) Line 149:  Delete, In the other hand,

21) Line 153:  Elaborate on what data goes into the "data matrices"

22) Line 154:  Ten out of thirty-three instead of 10 out of 33 ...

23) Line 162:  Two-dimensional  instead of 2D at the start of the sentence.

24) Line 173:  What is in this 33 by 25 and later 33 by 2 data matrix?

25) Line 181 - 185:  Are there mathematical descriptions for these success parameters?  These are just listed without any elaboration or description.

26) Line 186:  Space between Table and 1.  How are the chosen k values determined?

27) Line 189:  How was the "good reliability" of greater than 0.6 determined?

28) Why do you think you are getting better results with the reduced dimensionality from 32 x 5 to 32 by 2?

 

Author Response

Thank you for yor review.

Desired changes have been made.

Best regards.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript reports on detection and classification of underground ferromagnetic objects. The detection is based on measuring distortion to earth’s magnetic field introduced by the objects and classification of the objects kNN algorithm is used. This work is similar to previous work [38] on the same topic. The authors of this contribution used larger number (32) of fluxgate sensors but only one algorithm for classification (k-Nearest Neighbor). This approach showed 9% improvement in classification of the object compering to reported in Ref [38].  

The authors should improve the manuscript by addressing the following points to  to be consider for publication:

Explain what is it novelty compering to work reported in the reference [38] Indicate why the authors use for object classification only kNN algorithm. Compare their approach for the objects detection and classification to that presented in [38]. Separate the Results and conclusions section into Results and discussion, and Conclusions.

 

Moreover, the manuscript text must be improved by:

Shortening paragraphs and improving unity of theme and coherence, especially in Introduction and Results and Discussion. Correcting sentence:

Line 53 „It is possible detect”  add to

Line 54 Define „MA” at first use

Line 61 „ linear magnetic field lines”

Line 76 Use „and” instead of „&”

Line 149 „In the other hand” use On

Line 154 „different amount of ferromagnetic properties..”

Author Response

Thank you for your review.

Desired changes have been made.

Best regards.

Round 2

Reviewer 2 Report

The authors did not supply point-by point response to the issues I have raised in the review report.

Author Response

Thank you for your review.

Desired changes have been made.

Best regards.

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

Still there missing clear indication of using only kNN algorithm and comparison of your approach to that in Ref 38.