Could this technique be used to index images for search? e.g. intuitively, the triangles and their layout seem to preseve more unique information about the image than a grid would, kind of like Vornoi partitions, or an entropy preserving sample.
This isn't my area of expertise at all, but intuitively it seems like if you treated the triangles as a graph, you could sample a minimal subgraph from an image and then search for other files that have a similar subgraph. It's conceptually like "geometric hash."
Is that what this technique is for, as it seems to have more applicaitons than just an image filter.
This isn't my area of expertise at all, but intuitively it seems like if you treated the triangles as a graph, you could sample a minimal subgraph from an image and then search for other files that have a similar subgraph. It's conceptually like "geometric hash."
Is that what this technique is for, as it seems to have more applicaitons than just an image filter.