An Improved Ball Pivot Algorithm-Based Ground Filtering Mechanism for LiDAR Data

Round 1

Reviewer 1 Report

Please refer to the attachment.

Comments for author File: Comments.pdf

Author Response

The authors gratefully acknowledge the editors and the anonymous reviewers for their constructive comments, which have greatly helped us to improve the quality and presentation of our manuscript. In light of the comments, we have carefully revised our paper. For more details, please refer to the item-by-item response. Thank you for your time.

Author Response File: Author Response.pdf

Reviewer 2 Report

Novelty: use of alpha shapes in a different way is interesting, so I find it novel and interesting.

significance: in my opinion, the state-of-art methods in open source and commercial tools are already efficient, and community do not ask much about filtering problem 

quality of presentation: I find it low and give some remarks below.

interest to readers: in my opinion, although the problem was much focused n the past, LIDAR point data processing solutions find audience.

overall merit: the paper needs to be rewritten in terms of writing style. It has novel idea, worth to be published after improvement.

General remarks:

** The authors developed a method for filtering of LIDAR point cloud. Their method is point based and follows the idea of TIN densification method with some modification by integration of alpha shapes, to define the points which should be eliminated.I find the idea is interesting, and looks easy to implement. 

**In my knowledge, alpha shapes should not contain any point inside the circle. How this feature is used? This is not clear in the statements.

**Authors say 'The threshold of elevation angle is predefined as 80˚ in this paper.' why 80? e.g. 50 is already high for ground surface?

**But my objection is about presentation of the results. Figure 4 is good to show the workflow but, I would expect to see some attractive images for each step to show the interactions in the data-set.

**Secondly, specs of data-set is given but there should be some figures which show also the regions on aerial images, to show the test surfaces in better way, and to let the readers to reproduce the results.

**The errors should be explained much better. In remote sensing, omission and commission errors are clear what they are. Commission relates to correctness, and omission refers to completeness, here how do you describe? 

**

Author Response

The authors gratefully acknowledge the editors and the anonymous reviewers for their constructive comments, which have greatly helped us to improve the quality and presentation of our manuscript. In light of the comments, we have carefully revised our paper. For more details, please refer to the item-by-item response. Thank you for your time.

Author Response File: Author Response.pdf

Reviewer 3 Report

The reviewer didn't have enough time to thoroughly go through this manuscript. However, the manuscript have a good enough material to be published. The reviewer recommends more emphasis on Section 2.3, which describes the core methodology by the authors. OItherwise one may discard this manuscript because, in the first look, this is very similar to the original ball pivot paper by Bernardini et al (2002, 1999).

Author Response

The authors gratefully acknowledge the editors and the anonymous reviewers for their constructive comments, which have greatly helped us to improve the quality and presentation of our manuscript. In light of the comments, we have carefully revised our paper. For more details, please refer to the item-by-item response. Thank you for your time.

Author Response File: Author Response.pdf

Reviewer 4 Report

This paper proposed a ground filtering method based on improved Ball Pivot Algorithm for the requirement of Digital Elevation Model, especially extraction and classification of ground points from the Light Detection and Ranging technology data. It combines the 3D α-shape algorithm and spatial sorting method in order for applying the filter to different environment with efficiency and accuracy. An improved BPA is proposed to avoid the reconstruction failure from damage of TIN construction in the bare-ground region.

Advancing the knowledge on these topics is relevant for the remote sensing research community. In general, the manuscript gives a good description of literature review, and results and conclusions are supported by the empirical data analyzed in the study. However, I believe the manuscript needs to be revised. The quality of the collected data is very good but I think the analysis presented in the manuscript is somehow limited and needs to be extended, e.g. about Table 4. It is noted that the manuscript needs careful English editing paying attention to English grammar, so that the goals and results of the study are clear to the reader.

Some details are listed below:

Line 132 sentence ‘…a novel filtering method is proposed to enhance the efficient and accuracy of…’, ‘efficient’ should be ‘efficiency’.

Line 148 ‘…applicability of the method is proved in in Benchmark dataset provided…’ there are 2 times ‘in’

Lines 143-145, ‘…the proposed method can directly apply to other various environment and terrains, as it was more effectively and feasibly to ground filtering, which broke the limitation of setting intricate parameters or thresholds….’ ‘was more effectively and feasibly’ should be ‘is more effective and feasible’, which is similar to Line 29.

Line 146-148 ‘…An improved BPA is designed in the proposed filtering model, which directly extract…solve the problem of output loss in bare ground region’ Here ‘extract’ should be ‘extracts’ and ‘solve’ should be ‘solves’

In figure 3, ‘the lower point (A)’ is not visible.

Line 233, ‘…In 3D scene, the ball-pivoting algorithm is usually used to reconstruct…’ ,should ‘ball-pivoting algorithm’ be the same as ‘BPA’?

Line 240 the words ‘as follow’ should be revised.

Line 240, ‘Identify the lowest point by lowest height or select points with less distance than a threshold to a base surface in the set…add these points to the initial points collection G‘ What is the characteristics of initial point collection G, e.g. the number of points? ‘Identify the lowest point’ or ‘Identify the lowest points’?

It seems that Table 4 has not been explained in detail. In Table 4, r1=2.5 and 5, but in the title of table 4, it is indicated r1=30m.

Comments for author File: Comments.pdf

Author Response

The authors gratefully acknowledge the editors and the anonymous reviewers for their constructive comments, which have greatly helped us to improve the quality and presentation of our manuscript. In light of the comments, we have carefully revised our paper. For more details, please refer to the item-by-item response. Thank you for your time.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The manuscript entitled “An improved Ball Pivot Algorithm-Based Automatic Ground Filtering Mechanism for LiDAR data” has been improved a lot in the revised version. With more details, the description of the proposed algorithm is much clearer. This version can be accepted.

Some minor comments：

A suggestion for Figure 6 (L399): what I mean in the last comment is to add the reference hill-shade maps that are generated from the ground points of the ISPRS standard datasets, i.e., put the reference mesh and the correct mesh side by side. In this way, the readers can have a clearer impression about the performance of your algorithm.

L45: “the ones that doesn’t”-> do not

L155: remove “as it was more effective”, this statement is too arbitrary with no evidence.

L181: improving -> applying? Then applying to what? To the TIN surface?

     What does it mean by “large scale”? please specify this (e.g., radius?)

L183-184: “each point” from the original point cloud?

L186: those starting points -> those new seeding points

L190: which was treated … and was designed to

L195: are became to three -> become three

L341: 5m in Sample 51-71