mybm

A stereo blockmatching algorithm I programmed for my thesis.

It uses dynamic programming cost minimization and has the ability to aggregate cost functions.

• Input:

  

  rectified stereo image (for example left.png right.png)

• Output:

  

  Disparity image/ inverse depth map.

There is no minimal or maximal disparity so calculating the image might take a while.

Usage:

mybm -s left.png right.png

will calculate a disparity map for the stereoset left.png right.png.

mybm -s left.png right.png -c -o disparity.png

will save the disparity map in an image file. "-c" will make sure that the disparity map is colored in the opencv "jet" colormap for better reckognition of height. disparity.png will be the output image

Building and execution on a linux based system

Based on the following libraries:

• OpenCV

Install dependencies on Ubuntu (or Debian based distributions):

sudo apt-get install build-essential cmake libopencv-dev

Build the programm:

1. cmake .
2. make

Release version:

1. mkdir release
2. cd release
3. cmake -DCMAKE_BUILD_TYPE=Release ..
4. make

If you build the programm succesfully you should be able to see the "mybm" executable