Skip to content

README.md

Latest commit

 

History

History
123 lines (94 loc) · 5.26 KB

README.md

File metadata and controls

123 lines (94 loc) · 5.26 KB


Image Component Library

Version License

Table of Contents

About ICL

ICL is a C++ computer vision framework developed at the Neuroinformatics Group, University of Bielefeld. It provides a large set of classes and functions for image processing, computer vision, 3D geometry, GUI creation, and physics simulation — all designed to work together seamlessly.

Key Design Principles

  • Optimal Performance — optional Intel IPP acceleration, SIMD, OpenMP, and OpenCL support. Images use planar channel storage with zero-copy shallow wrapping of existing data.
  • Simple C++ Interface — readable, low-boilerplate code. A complete image acquisition and visualization application fits in ~15 lines of C++. PIMPL pattern hides complexity.
  • Self-Contained Framework — write applications in ICL rather than gluing libraries together. Includes GUI creation, program argument parsing, multi-threaded application lifecycle, and build tools.
  • No Compulsory Dependencies — all external dependencies are optional. Build a slim ICL for algorithm development, then link against a full build for deployment.

Installation

Building from Source

Required tools:

  • CMake 3.16+
  • C++17 compiler (GCC 7+, Clang 5+, MSVC 2017+)

Required libraries:

  • libjpeg (or libjpeg-turbo)
  • libpng

Minimal build

git clone https://github.com/iclcv/icl.git
mkdir icl/build && cd icl/build
cmake ..
make -j$(nproc)

Recommended build (with Qt6, OpenCV, and common features)

cmake .. \
  -DBUILD_WITH_QT=ON \
  -DBUILD_WITH_OPENCV=ON \
  -DBUILD_WITH_EIGEN3=ON \
  -DBUILD_WITH_IMAGEMAGICK=ON \
  -DBUILD_WITH_LIBAV=ON \
  -DBUILD_WITH_OPENCL=ON \
  -DBUILD_APPS=ON \
  -DBUILD_DEMOS=ON \
  -DBUILD_EXAMPLES=ON
make -j$(nproc)

Dependencies

All dependencies except libjpeg and libpng are optional. Enable them with -DBUILD_WITH_<NAME>=ON.

Core (recommended)

Feature CMake Flag Description Install (apt) Install (brew)
Qt6 BUILD_WITH_QT GUI framework, visualization, image display qt6-base-dev qt6-multimedia-dev libqt6opengl6-dev libglew-dev qt glew
OpenCV BUILD_WITH_OPENCV Image type conversion, ORB features libopencv-dev opencv
Eigen3 BUILD_WITH_EIGEN3 Faster matrix operations libeigen3-dev eigen
OpenCL BUILD_WITH_OPENCL GPU-accelerated processing ocl-icd-opencl-dev opencl-headers (built-in on macOS)
ImageMagick BUILD_WITH_IMAGEMAGICK Extended image format I/O libmagick++-dev imagemagick
FFmpeg BUILD_WITH_LIBAV Video file I/O libavcodec-dev libavformat-dev libswscale-dev ffmpeg

Camera backends

Feature CMake Flag Description Install (apt) Install (brew)
V4L2 BUILD_WITH_V4L USB cameras/webcams (Linux) (usually in kernel headers) n/a
libdc1394 BUILD_WITH_LIBDC FireWire cameras libdc1394-dev libdc1394
libfreenect BUILD_WITH_LIBFREENECT Kinect v1 libfreenect-dev libfreenect
libfreenect2 BUILD_WITH_LIBFREENECT2 Kinect v2 build from source build from source
Pylon BUILD_WITH_PYLON Basler GigE cameras Basler SDK Basler SDK

Advanced

Feature CMake Flag Description Install (apt) Install (brew)
PCL BUILD_WITH_PCL Point Cloud Library compatibility libpcl-dev pcl
Bullet3 BUILD_WITH_BULLET Physics simulation (ICLPhysics) libbullet-dev bullet
Qt6 WebSockets (auto when Qt found) WSGrabber/WSImageOutput for cross-process / cross-host image transfer with auto-reconnect qt6-websockets-dev qtwebsockets
zstd (auto-detect) zstd codec plugin for ImageCompressor (lossless, modern) libzstd-dev zstd
libusb BUILD_WITH_LIBUSB USB device support libusb-dev libusb
Intel IPP BUILD_WITH_IPP Intel Performance Primitives (Linux x86 only) Intel oneAPI n/a
Intel MKL BUILD_WITH_MKL Intel Math Kernel Library (Linux x86 only) Intel oneAPI n/a

Build options

Flag Default Description
BUILD_APPS OFF Build end-user applications (installed as icl-*)
BUILD_DEMOS OFF Build demo programs
BUILD_EXAMPLES OFF Build documentation examples
BUILD_TESTS OFF Build test suite (downloads Google Test)

Running Tests

cmake .. -DBUILD_TESTS=ON
make tests    # build test executables
ctest         # run all tests

# or run individually:
./ICLMath/tests_iclmath

Questions, Feedback, or Issues?

Please open an issue on GitHub.