xiaohulugo / pointcloudsegmentation

A point cloud segmentation algorithm based on clustering analysis

PointCloudSegmentation-V2

Modification: (1) use nanoflann.cpp to replace the ann library; (2) remove some cpps

Three algorithms on point cloud segmentation used in the following paper:

Pairwise Linkage for Point Cloud Segmentation, Xiaohu Lu, etc. ISPRS2016. https://github.com/xiaohulugo/xiaohulugo.github.com/blob/master/papers/PLinkage_Point_Segmentation_ISPRS2016.pdf

*The algorithm used in the ISPRS2016 paper is ClusterGrowPLinkage.cpp

Prerequisites:

1. OpenCV > 2.4.x
2. OpenMP

Usage:

1. build the project with Cmake
2. run the code
3. see main.cpp for interfaces/demos

Docker: (For linux)

1. Build the docker image by running docker build -t opencv:cpp . to build the docker image with name opencv and tag cpp
2. Modify the code as you wish. Also the default data folder for test is ./data in the root path.
3. Modify the test_segment.sh. This is the script the container will run after it is set up.
4. Run the following command in the root path of the project.

docker run -it --rm \
    -v $(pwd):/pointSegment \
    opencv:cpp \
    /pointSegment/test_segment.sh

Performance:

Please cite these two papers if you feel this code useful:

@ARTICLE{Lu2016Pairwise,
author = {Lu, Xiaohu and Yao, Jian and Tu, Jinge and Li, Kai and Li, Li and Liu, Yahui},
title = {PAIRWISE LINKAGE FOR POINT CLOUD SEGMENTATION},
journal = {ISPRS Annals of Photogrammetry, Remote Sensing \& Spatial Information Sciences},
year = {2016},
}

}

Feel free to correct my code, if you spotted the mistakes. You are also welcomed to Email me: [email protected]

pointcloudsegmentation's Issues

ANN submodules does not exist

the HU005.txt and result.txt is not avaliable

作者您好，有个疑问是：如果对于当前点来说，有两个近邻点的密度相同，并且距离相同，那么当前点应该与哪个点进行关联？

如果对于当前点来说，有两个近邻点的密度相同，并且距离相同，那么当前点应该与哪个点进行关联？

how can i get a result like in the paper?

i run P-linkage-based segment to seg my data,but it just get a line with some color,could you help me?if you have time, i can send e-mail to you .thanks

Visualization tool

Hello, @xiaohulugo ,I want to know do you have any tool to visualize the data in test_data.txt?
I find it a little complicated to use pcl_viewer visualize the data in test_data.txt.
Any help will be appreciated! Thank you!

How to extract the outline of the point cloud

Hello @xiaohulugo, I'm trying to obtain the outline of the point cloud after segmentation. However, the result is terrible as you can see in the picture, especially those in the red circle.

I'm using normal-based region growing to segmentation and I would like to use your ouP-Linkage algorithm next time.

I wonder how do you extract the outline of the point cloud as you shown in the paper,

Any help will be appreciated! Thank you!

你好打扰下，请问这个代码跟平台有关吗？

你好打扰下，请问这个代码跟平台有关吗？我用同样的数据在windows平台和Linux平台下运行的效果不一样，并且windows平台的效果比较好，比如聚类的数量，win10下测试的为1170个，Linux下为1668，最后得到的面片数，win10下为33，linux下为89，请问这有可能是哪方面的原因？

How do I select best parameters?

Hi, Thanks for your work.

I had a Areal dataset, Which algorithm shall I use & what parameter combinations should I try?

Any pointers would help.

Python version of Point Cloud Segmentation

Hi.
This is really an awesome project. Do you have any reference of its implementation in Python? It will be great if you can provide that.