libviso-with-pointcloud

This is a project modified on the basis of libviso2 to show the semi-dense pointcloud using Pangolin. If you want to konow more about libviso2, please cite (http://www.cvlibs.net/software/libviso/)

The whole project is tested in Ubuntu Platorforms

Prerequisites needed for compiling and running the raw libviso2 :

• libpng (available at: http://www.libpng.org/pub/png/libpng.html)
• libpng++ (available at: http://www.nongnu.org/pngpp/)

libpng and png++ are needed for reading the png input images. On a ubuntu box you can get them via apt:

- sudo apt-get install libpng12-dev
- sudo apt-get install libpng++-dev

Additional Prerequisites for this project

Besides,to build this project, you need the followings:

Pangolin

We use Pangolin for visualization and interface. Dowload and install instructions can be found at: https://github.com/stevenlovegrove/Pangolin.

OpenCV

We use [OpenCV] to simply read and display image.

Eigen3

Download and install instructions can be found at: http://eigen.tuxfamily.org.

C++11 or C++0x Compiler

We use the some functionalities of C++11.

Build

Finaly cdto the project folder ,and

cmake .
make 

Run

This project is mainly for KITTI ODemotry dataset.You need to download a sequence. The dataset is available through http://projects.asl.ethz.ch/datasets/doku.php?id=kmavvisualinertialdatasets
To Run the project, you need to modify the camera instrinsic parameters in line 119~122 in the demo.cpp :

param.calib.f = fx or fy
param.calib.cu = cx
param.calib.cv = cy
param.base = baseline in meters

And

./viso2  path_to_KITTI_dataset

where "the path_to_KITTI_dataset" is dataset sequence root directory which contains "image_0" , "image_1" and so on

Result

Pangolin GUI: .