Is there any algorithm that could be used to calculate the visual similarity between visual symbols (letters)? I believe it would take into account rotational symmetry, topological identities and the spatial extent of the topological features. The similarity metric would say something like this:
A H are similar to each other. D C O Q are similar to each other. p b q d are similar to each other. I J are similar to each other. On the other hand, these sets of letters are not similar to each other, and the similarity can be measured quantitatively.
Is there an algorithm to measure this spatial visual similarity of symbols?
If these symbols can be mapped as raster pixels, the answer is Yes. Co-location, variation and distribution of spatial data (including spatial structure of the symbols) are checked in one very discriminative index below.
Spatial Efficiency metric (SPAEF) is proven to be robust when comparing two raster maps. Python and Matlab codes are available at: http://space.geus.dk/tools_products/index.html
Recent paper on the metric: https://www.hydrol-earth-syst-sci.net/22/1299/2018/hess-22-1299-2018-assets.html
Observed vs Simulated raster maps
