PhotoMatch: An Open-Source Tool for Multi-View and Multi-Modal Feature-Based Image Matching

The accurate and reliable extraction matching of distinctive features (keypoints) in multi-view multi-modal datasets is still an open research topic the photogrammetric computer vision communities. However, one main milestones selecting which method a suitable choice for specific applications. This encourages us to develop educational tool that encloses different hand-crafted learning-based feature-extraction methods. article presents PhotoMatch, didactical, open-source feature-based image matching. software includes wide range state-of-the-art methodologies preprocessing, feature matching, including deep learning detectors descriptors. It also provides tools detailed assessment comparison approaches, allowing user select best combination methods each dataset. first version was awarded by ISPRS (ISPRS Scientific Initiatives, 2019). A set thirteen case studies, six datasets, processed following methodologies, results provided are analysed show capabilities tool. PhotoMatch Installer source code freely available.