This project aims at providing a minimalist, yet expandible, image processing framework. With modular blocks of image processing steps the goal is to reach pipeline flexibility and expose used algorithms for further adjustments and performance optimization. This is of particular interest for embedded applications and for use with other frameworks that heavily rely on computer vision, such as ROS.
| Input | Output |
|---|---|
 |
 |
 |
 |
 |
 |
- Build essentials
- Git (to pull this source)
- CMake >= 2.8.3
- OpenCV >= 3.2.0
# apt install build-essential git cmake libopencv-dev
Get the source code:
$ git clone https://github.com/em-er-es/imb
For building packages use the overlay repository and follow the instructions:
$ git clone https://github.com/em-er-es/devops
# dpkg -i imb-*.deb
Set the prefix for local installation in accordance with FHS:
$ cd build
$ cmake .. -DCMAKE_INSTALL_PREFIX="$HOME/.local"
$ make install
Use the included script from this (base) directory:
$ bash scripts/build.sh
$ mkdir -p build && cd build
$ cmake .. -DCMAKE_BUILD_TYPE=Debug -DCMAKE_C_COMPILER=/usr/bin/gcc -DCMAKE_CXX_COMPILER=/usr/bin/g++ -DCMAKE_INSTALL_PREFIX="$HOME/.local" -DDEBUG=0 -DCOVERAGE=0 -DPROFILE=0 -DVERBOSE=0
$ make -j $(nproc)
$ make install
LLVM can be used as well, as per build.sh, however certain builds result in segmentation fault. Thorough testing is recommended.
Check the devops repository: https://github.com/em-er-es/devops
$ build/bin/imb -i input/input.png -o output/output.png -d
$ imb -i input/input.png -o output/output.png -d
Usage: build/bin/imb [-acCdDhmrRv] -i input -o output
-a Scan lines
-c Color mixing
-C Continuous image check
-d/D Enable/Disable input and output comparison display
-h This help
-i Input image (default: input.png)
-m VIC II model selector (default: latest)
-o Output image (default: output.png)
-r/R Enable/Disable RGB comparison (default: enabled)
-v Verbose