Fourfold Bounce Scattering-Based Reconstruction of Building Backs Using Airborne Array TomoSAR Point Clouds
, Fubo Zhang 1, Xingdong Liang 1,2,*, Yanlei Li 1, Qichang Guo 1, Yangliang Wan 1, Xiangxi Bu 1 and Yunlong Liu 1Round 1
Reviewer 1 Report
The proposed approach provided a method for building-back 3D reconstruction by using TomoSAR four-bounce scattering point(FB).The major innovation of this paper is an algorithm which used the distribution characteristics of FB to estimate the coordinate of the building-back.The perspective of this paper is somewhat innovative.However,the applicable condition of the method has not been sufficiently discussed.Following are a few question I have regarding the content presented.
1.The main contribution of this paper is an FB-based algorithm for the reconstruction of the back of a building. The article is somewhat scientific and answers the research question. However, the method proposed in the article appears to be narrowly adapted and does not provide much validation of the method
2.A specific description of the scope of application of the algorithm is missing in the authors' conclusion.
3.In the part of Methodology,the method use several threshold to detect region growing seed and evaluate adjacency of scattering point.However,the paper have not discussed how to calculate threshold in different dataset.So the method is difficult to use in the real scenes.
4.The article only compares the reconstruction accuracy of one building in the results section, which is not generalizable and lacks statistical hypothesis testing
5.The authors list a variety of methods in the related work, but only one simple method is compared in the experimental section, so why not compare more 3D reconstruction methods
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
This paper tries to reconstruct the backs of buildings which cannot be completely solved by multi-view observation. To this end, this paper suggests a fourfold bounce scattering point-based reconstruction method. The authors derived the fourfold bounce scattering model which gives the distribution law of fourfold bounce scattering points. The suggested method is evaluated on both simulation and real data.
Major comments
One of the main contributions of this paper is that the distribution of fourfold bounce scattering points between combined buildings may imply information on the backs of buildings. However, the distribution described in Figure 5 (b) looks similar to Figure 4 in R. Cheng et al 2019. (Multiple-Bounce Scattering of Tomo-SAR in Single-Pass Mode for Building Reconstructions). If the distribution is not a unique phenomenon that appears in fourfold bounce scattering points, it would be good to clarify the contribution related to distribution and add the aforementioned paper as a reference.
The proposed method and traditional method are compared and evaluated based on the ground-truth value in the experimental results and analysis section. However, the results could not fully support the improvement due to the missing details. First of all, there is no reference for the comparison method (the traditional roof point-based method). Second, there is no explanation of how the true value is acquired in real experiments. Therefore, the author needs to add more details of experiments to verify the results.
Minor comments
Typo:
(-r cos a + lk, r sin a) -> (-r sin a + lk, r cos a) at line 122
(2 x2 cos a) -> (2 x2 sin a) at eq (9) and (12)
Reference 13 article title
Duplicated sentence:
‘In fact, some sparse roof points may be ignored with a high probability.' at line 357~359’
Reference 11 and 20 are the same.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The proposed approach provided a method for building-back 3D reconstruction by using TomoSAR four-bounce scattering point(FB).The major innovation of this paper is an algorithm which used the distribution characteristics of FB to estimate the coordinate of the building-back.The perspective of this paper is somewhat innovative. The manuscript can be accepted for publication after minor revision. Following are a few question I have regarding the content presented.
Line 149. Form(14) should be Equation(14). The similar problem should be revised.
Line 230 “2Te2,” should be a error. All errors should be revised.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
This paper tries to reconstruct the backs of buildings with the fourfold bounce scattering point-based reconstruction method. the fourfold bounce scattering model appears when multiple buildings are closely located. The suggested method is compared with the facade-fusion method and neural network-based reconstruction method. Compared methods are based on single-view TomoSAR. The suggested method is evaluated on both simulation and real data.
The revised version addresses most of the comments from the previous review. The paper has improved in most parts. However, the detail of the results brought a minor question.
Minor comment
Multi-view TomoSAR was introduced as a method to acquire the backs of buildings in Paragraph 1, Lines 39-43. However, the author addresses multi-view observation methods are weak in reconstructing the backs of buildings in Paragraph 1, Lines 52-53. Subsequently, single-view TomoSAR based methods are compared without the detail of the weakness of the multi-view TomoSAR method. If there are any known reasons or examples related to the weakness of the multi-view TomoSAR, it would be good to clarify the weakness of the multi-view TomoSAR to persuade others about the results.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
