This module contains various image processing utilities and algorithms. The main header file, imgproc.hpp, contains implementations for image thinning, non-max suppression, affinity map computation and Gaussian mixture model learning. The module itself contains wrapper interfaces for edge detection and cosegmentation algorithms.
For edge detection, two versions of Canny's method are included: one based on OpenCV's implementation, and one using binary feature convolutions via LBSPs (see our CVPRW2016 paper for more information). If CUDA is found and enabled via CMake, a GPU version of Achanta et al.'s SLIC superpixels will also be available included (taken from fderue/SLIC_CUDA). The included shape cosegmentation algorithm is still a work-in-progress.