📝 Hough Transform based 3D Line Detection

---

• This is the reimplementation of Iterative Hough Transform for Line Detection in 3D Point Clouds. This implementation organizes the algorithm into a library that can be used in plug-and-play style easily.

• Image of semihemisphere whose vertices show directions for the method's parameter space discretization

🎉 TODO

---

• Implementation of Iterative Hough Transform for Line Detection in 3D Point Clouds
• Test on pointcloud data

🎛 Dependencies

---

sudo apt-get install \
    libpcl-dev \

🔨 How to Build

---

make default # to build the library
make apps # to build the examples

How to Run with Docker

# build
docker build -f ./dockerfiles/ubuntu2004.dockerfile -t 3d_line_detection .

# run
docker run -it --rm --net host -e DISPLAY=$DISPLAY -v $HOME/.Xauthority:/root/.Xauthority -v `pwd`:/workspace 3d_line_detection

🏃 How to Run

---

• draw semihemisphere

# ./build/examples/draw_sphere_app [number/of/subdivision]
# eg: ./build/examples/draw_sphere_app 0
# number of subdivision: 0 -> 5

• test 3d line detection example

# ./build/examples/3d_line_detection_app [path/to/pcd/file] [use/default/setting/yes:1/no:0] [num/range/bin] [sphere/granularity] [min/num/vote] [distance/to/line/thresh]
# eg: ./build/examples/3d_line_detection_app ./data/test_cloud.pcd 0 64 4 10 0.2
# note that if you increase number of accumulator cells, you need to reduce the threshold of minimum vote number

original pointcloud	detected 3d lines

💎 References

---

• Iterative Hough Transform for Line Detection in 3D Point Clouds. BSD LICENSE