A set of utilities written in C++ for simplifying image alignment tasks. Supports image-to-image and image-to-mesh registration across a range of image depths and types. Fully supports gray, gray+alpha, RGB, and RGBA images in 8, 16, and 32 bits-per-channel.
std::filesystem. If it is not found, then Boost.Filesystem is required. This behavior can be controlled with the RT_USE_BOOSTFS CMake flag.Install using apt:
sudo apt install cmake libopencv-dev libvtk9-dev libvtk9-qt-dev libboost-program-options-dev libinsighttoolkit5-dev
Install using Homebrew and the provided Brewfile:
cd registration-toolkit/ brew bundle
Our research group maintains a CMake project called vc-deps for building dependencies that are common across many of our C++ projects. Once it is built, point this project's CMake configuration at the vc-deps libraries using the CMAKE_PREFIX_PATH flag:
cmake -S . -B build/ -DCMAKE_PREFIX_PATH=/path/to/vc-deps/deps/
This project uses a CMake build system and can be built using the default CMake workflow:
# Get the source code git clone https://gitlab.com/educelab/registration-toolkit.git cd registration-toolkit/ # Configure and build cmake -S . -B build/ cmake --build build/ cmake --install build/ # optional
The CMake project provides a number of flags for configuring the build:
RT_BUILD_APPS: Compile the utility applications. (Default: ON)RT_BUILD_DOCS: Build documentation. Dependencies: Doxygen, Graphviz (optional). (Default: ON if Doxygen is found)RT_INSTALL_DOCS: Install HTML documentation to the system. (Default: OFF)RT_BUILD_TESTS: Build project unit tests. This will download and build the Google Test framework. (Default: OFF)RT_USE_BOOSTFS: Use the Boost::filesystem library instead of std::filesystem. (Default: ON if std::filesystem is not found)RT_USE_VOLCART: Build with optional Volume Cartographer components (Default: OFF)Flags can be set using ccmake or by providing them at configuration time. Example:
cmake -DRT_BUILD_TESTS=ON ..Image-to-Image Registration
To align a moving image close-up.jpg to a fixed image wide-angle.jpg:
rt_register -f wide-angle.jpg -m close-up.jpg -o result.jpg
See the image registration guide for a more thorough walkthrough of image-to-image alignment.
Note: By default, this application will attempt to automatically detect and match features between the two images in order to perform registration. To provide pre-computed landmarks, please provide a Landmarks file using the --input-landmarks flag.
To align a moving image color-photo.jpg to a textured 3D mesh grayscale-mesh.obj:
rt_register -f grayscale-mesh.obj -m color-photo.jpg -o color-mesh.obj
Note: This process uses a 2D-to-2D registration process between the moving image and the texture image provided by the mesh file. This assumes that the provided mesh is roughly planar and has a "coherent" UV map (i.e. a single, continuous chart). Use rt_reorder_texture to convert a multi-chart mesh to a single chart one.
rt_apply_transform: Apply a Transform produced by rt_register to an image. Useful for duplicating exact registration results.rt_generate_landmarks: Automatically detect and generate landmarks between two images and save as a Landmarks file.rt_swap_landmarks: Swap the order (i.e. fixed <-> moving) of an existing Landmarks file.rt_plot_landmarks: Plot a Landmarks file on the fixed and moving images.rt_retexture_mesh: Re-save a mesh with a new texture image. Does not modify the UV map, so assumes that the replacement image has already been aligned to the current texture image.rt_segment_disegni: Separate a composite disegni image into individual pieces. More information coming soon.A Landmarks file is a space-separated plain-text document where each line represents a pair of matching pixel positions in the fixed and moving images. Pairs are in the format fX fY mX mY. Values are interpreted as floating point value strings using std::stod. The # character begins a comment which terminates at the end of the line:
# My Image Landmarks 500 100 2500 1000 # Integer 101.56 234.56 1010.911 2345.67 # Decimal float 1.0156e02 2.3456e02 1.010911e03 2.34567e03 # Scientific
Parsons, Stephen, C. Seth Parker, and W. Brent Seales. "The St. Chad Gospels: Diachronic Manuscript Registration and Visualization". In: Manuscript Studies: A Journal of the Schoenberg Institute for Manuscript Studies 2.2 (2017), pp. 483-498. DOI: 10.1353/mns.2017.0022
Christy Y. Chapman et al. “The Digital Compilation and Restoration of Herculaneum Fragment P.Herc.118”. In: Manuscript Studies: A Journal of the Schoenberg Institute for Manuscript Studies 6.1 (2021), pp. 1–32. DOI: 10.1353/mns.2021.0000
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4