An Efficient Document Skew Detection Method Using Probability Model and Q Test
Round 1
Reviewer 1 Report
This article is about skew detection in electronic documents. In the proposed method, the authors use, among others two randomly selected CCs and Dixon's Q test statistics. The presented method was used to determine skew in 900 documents with different angle of inclination. In addition, accuracy and the time needed to process the document make this method useful in practice.
The article is carefully written. The list of references is sufficient and helpful in understanding the subject.
Below is a list of minor editorial notes:
1) some compound sentences (e.g. lines 77-80) should be divided into parts; this note also applies to other parts of the text,
2) table 1: it may be worth inserting specific methods and algorithms in the form of references described in part 2 (Related Work),
3) line 104: sentences should not be started with [24-26] [6],
4) line 153: the abbreviation SE is not explained,
5) line 186: the sentence should be without intend and start with a lowercase letter,
6) line 260: it is better not to start sentences with And,
7) line 262: no spaces after S_optm,
8) the bulleted list on page 10 should be moved closer to the left margin,
9) table 6: since the authors have no results for the Boudra method, it is better to omit this column.
After considering the above minor editing errors, I suggest publishing the article.
Author Response
Dear Reviewer：
Thanks for your comments. Those comments are valuable and very helpful for improving our paper. We have read your comments carefully and revised them carefully. We will response your comments point by point. The responds are marked in red in this paper.
1) some compound sentences (e.g. lines 77-80) should be divided into parts; this note also applies to other parts of the text,
Response: Some compound setences have been modified, e.g. lines 81-82, lines 86-89 of the paper. (Note 2,3)
2) table 1: it may be worth inserting specific methods and algorithms in the form of references described in part 2 (Related Work),
Response: I have added the references to table 1, please see line 108 of the paper. (Note 5)
3) line 104: sentences should not be started with [24-26] [6],
Response: It has been modified, please see lines 116-118 of the paper. (Note 6)
4) line 153: the abbreviation SE is not explained,
Response: It has been modified, please see line 166 of the paper.(Note 7)
5) line 186: the sentence should be without intend and start with a lowercase letter,
Response: This sentence has been modified to start with a lowercase letter, please see line 199 of the paper.(Note 8)
6) line 260: it is better not to start sentences with And,
Response: I have deleted the word "And", please see line 276 of the paper.(Note 10)
7) line 262: no spaces after S_optm,
Response: I have added a spece after S_optm, please see line 278 of the paper.(Note 11)
8) the bulleted list on page 10 should be moved closer to the left margin,
Response:The bulleted list on page 10 has been moved closer to the left margin, please see lines 321-329 of page 10 of the paper.(Note 12)
9) table 6: since the authors have no results for the Boudra method, it is better to omit this column.
Response: I have omitted the column about Boudra method, please see Table 6 and line 397 of the paper.(Note 13)
Special thanks to you for your good comments.
Best regards!
Reviewer 2 Report
I understand the topic of this paper very easily (thanks to the good writing), and I will just quote the sentences in the abstract as summary here " Document skew detection is one of the key technologies in most of the document analysis systems. However, existing skew detection methods either have low accuracy or require large amount of computation. To achieve a good tradeoff between efficiency and performance, this paper proposes a novel skew detection approach based on bounding boxes, probability model and Dixon’s Q test.". These sentences summarize the paper very well.
The paper is already in a very good condition, just some opinions to share with you:
Please add a chapter called "background" and introduce necessary terms such as Q test. Thus I do not have to go to citation to read another paper. In your introduction, you mentioned a lot about the importance of skew detection, but there is no reason saying why it has to be fast. It is important to convince people that efficiency is important. If you have to use some magic number, either for heuristic or simple reasons, please explain yourself. Like line 220, if you "choose a value of eight times the number of ECCs as the threshold value", then remember to say why 'eight'Overall, this paper is in very high quality. Thank you for the good reading experience.
Author Response
Dear Reviewer:
Thanks for your comments very much. Those comments are all valuable and vey helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in re in here. The main corrections in the paper and the responds to your comments are as following:
1.Please add a chapter called "background" and introduce necessary terms such as Q test. Thus I do not have to go to citation to read another paper.
Response: After our discussing, in order to avoid affecting the structure of the paper, we add some introduction of Q test in the chapter called "Related Work". In which we describe the principle and steps of "Q test" method in detail, so that readers can understand this method more easily. Please see lines 98-107 of the paper.(Note 4)
2. In your introduction, you mentioned a lot about the importance of skew detection, but there is no reason saying why it has to be fast. It is important to convince people that efficiency is important.
Response: We add some content in the "Introduction" section (section 1) to explain why angle detection needs to be fast. There two reasons: (a)it can improve the sacnning speed of the printer.(b) meet the requirements of some real-time system. Please see lines 35-38 of the paper.(Note 1)
3. If you have to use some magic number, either for heuristic or simple reasons, please explain yourself. Like line 220, if you "choose a value of eight times the number of ECCs as the threshold value", then remember to say why 'eight'
Response: We add some content in section 3.3 to explain why we choose a value of eight times the number of ECCs as the threshold value. This value of "eight" is derived from our experiments after considering both speed and accuracy of our method. Please see lines 232-236 of page 7.(Note 9)
Special thanks to you for your good opinions!
Yours sincerely!
