“Hungarian” Image (Differencing) Descriptor

Cultural or natural heritage digitization has become an essential part of the knowledge economy in the EU. We propose a tool to improve the accessibility of hi-tech devices and the sustainability of cooperation in multidisciplinary teams in the first scanning session. The CRUSE scanning beginners may take into account a novel image descriptor of moderate size for practical applications, e.g. differencing similar CRUSE scans. Our approach combines a set of Harris corners with the Hungarian algorithm to achieve an informative visual representation in the form of a planar subgraph. We compare, on selected use-cases, the solution quality and/or disadvantages. The key practical contribution of our research is an original approach for an alternative way of image structure understanding, named "polygon shape descriptor". Our innovation is about transforming the image differencing using Harris corners and Hungarian edges for support and/or speed-up of visual comparison of very similar scans. Methodologically, we interrelate computer vision and computational geometry to enhance hi-tech accessibility in virtual reality classes and student projects.