Next Article in Journal
Remote Sensing and Social Sensing Data Reveal Scale-Dependent and System-Specific Strengths of Urban Heat Island Determinants
Next Article in Special Issue
Exploring TanDEM-X Interferometric Products for Crop-Type Mapping
Previous Article in Journal
RADet: Refine Feature Pyramid Network and Multi-Layer Attention Network for Arbitrary-Oriented Object Detection of Remote Sensing Images
Previous Article in Special Issue
An End-to-End Local-Global-Fusion Feature Extraction Network for Remote Sensing Image Scene Classification
Open AccessArticle
Peer-Review Record

Urban Land-Cover Classification Using Side-View Information from Oblique Images

Remote Sens. 2020, 12(3), 390; https://doi.org/10.3390/rs12030390
by 1,2, 2,3,* and 1
Reviewer 1: Anonymous
Reviewer 2: Aliihsan ŞEKERTEKİN
Reviewer 3: Anonymous
Remote Sens. 2020, 12(3), 390; https://doi.org/10.3390/rs12030390
Received: 17 December 2019 / Revised: 23 January 2020 / Accepted: 23 January 2020 / Published: 26 January 2020
(This article belongs to the Special Issue Feature-Based Methods for Remote Sensing Image Classification)

Round 1

Reviewer 1 Report

This paper focuses on urban land-cover classification of remote sensing images. In order to address the limitation of the information of top-view images, this paper leveraged oblique images and proposed a feature extraction algorithm. Finally, the extracted side-view features incorporated with top-view information are fed into the random forest classifier to accomplish better classification results. My comments are as follows. 1. The boundaries of above-ground segments are used to describe the edges of related side-view images. So, does it have more serious impact on the extraction of side-view textures when the boundaries of segments are not accurate enough? Should you pay more attention to this problem? 2. In this paper, to get better side-views for the later feature extractions, a homographic transform is adopted to rectify the textures to the front view. I would like to know more details about transformation method. 3. I don’t quite grasp the means for the sentence in the 260 and 261 lines and it seems like to contradict the contribution of this research. 4. Compared with two results of classification with evenly distributed and non-evenly distributed samples, the accuracy improved more obviously after combining the side-view features when samples are non-evenly distributed. I suggest the authors provide the reasons for it. 5. I also would like to know the meaning of “the user accuracy” in the evaluation part. 6. There are also several format errors: 1) section 5.3 and section 5.3.1 are written wrongly according to my conjecture; 2) Figure 6 should be refined (e.g. the overlapping line and the distribution of arrows); 3) as for Figure 7, why are the symbols of six sites not marked on the original image.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Dear Author,

Please consider the attached file for my comments.

All the best.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The manuscript is good and need some minor checks like the following:

the English language can be a little improved with the help of a professional (if possible).

the structure of the presentation can be simplified in a more simple, accessible formula where methodology and results (experiments results) can be better emphasized with distinct and bold titles.

a description of land cover classes might be better presented with the help of a table, although there is information within the text. an explanation regarding the selection of the test imagery can be useful.

scale factor for the analysis is important. Image samples need scale bar sometimes in order to understand the scale of the analysis.

it is better to bring a little more information about the imagery features like sensor, resolution etc. within a synthetic table. The complementarity of the selected imagery is important and need to be a little explained (partly done).

the scheme in figure 6 is important and very useful but this can be more visible from the beginning of methodological section. It is useful to put the emphasis a little bit more on segmentation rules.

reference list need to be made according to a style from special editing software, in alphabetical order. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Acceptable

Author Response

Thank you very much for your careful work and thoughtful suggestions

Back to TopTop