Method for 2D-3D Registration under Inverse Depth and Structural Semantic Constraints for Digital Twin City

Round 1

Reviewer 1 Report

1. Overview and general recommendation:

The authors of the article have undertaken an interesting topic involving the problem of vision sensor registration in the  virtual digital twin city scene.

 The paper presents the 2D-3D registration method under the inverse  depth and structural semantic constraints. The perspective and inverse depth images of the  virtual scene are obtained by using the perspective view and inverse depth nascent technology, and   the structural semantic features are extracted by the 2-lines minimal solution set method.

 The research problem was formulated correctly, and the researcher's methodology was also precisely presented. I have no comments on the scientific part.

 The presented material corresponds to the profile of the Journal " Applied Sciences ". The scientific value of the submitted material qualifies the article for publication in this Journal.

2. Minor comments:

 Please consider changing the layout of the manuscript by introducing the main points:

1.     Introduction

2.     Related Works

3.     Materials and methods

4.     Results and discussion

5.     Conclusions

Author Response

Point 1: Please consider changing the layout of the manuscript by introducing the main points:

1. Introduction
2. Related Works
3. Materials and methods
4. Results and discussion
5. Conclusions.

Response 1: Dear Professor, thanks very much for your comments. We have changed the layout of the paper according to your opinion.

Author Response File: Author Response.docx

Reviewer 2 Report

This study proposes the 2D-3D registration method under the inverse depth and structural semantic constraints for digital twin cities.  

There are two main contributions of this study:

1. Authors created the 2D-3D registration method with constraints of structural semantic and inverse depth coordinate. The method effectively solves the problem of large error caused by uneven distribution of feature points in the vertical direction. Also, the method has high accuracy registration of monocular images in the digital twin scene.

2. The method seamlessly integrates with existing digital twin platforms.

  The 2D-3D registration consists of two parts, camera pose estimation and object positioning and the authors reviewed the literature for these procedures. Methods of the paper consist from several algorithms: Perspective View and Inverse Depth Image Nascent Technology (PDNT), Structural Semantic Feature Extraction, EDLines line feature detection, 2-lines MSS plumb line extraction with prior constraint, Matching and Pose Estimation under Inverse Depth and Structural Semantic Constraint, Cross-validation of Dynamic Adaptive Threshold, Fast P5P Method and finally Error Function Combined with Structural Semantic Constraint.   The paper also contains a detailed experiment that confirms the advantages of this approach.   The paper described all the steps in detail and the authors presented the existing algorithms they used as a basis for developing their approach.  

Finally, I think that it is a very interesting and good paper I suggest the Editorial Board to accept it for publication.

Comments for author File: Comments.pdf

Author Response

Dear Professor,

Thanks for your comments and support.