Representation and comparison methods for semantically different images
Georgy Kukharev, Nadezhda Shchegoleva, E. Kamenskaya · Pattern Recognition and Image Analysis · 2014
This paper discusses the methods of presentation and comparison of semantically unrelated images with an assessment of their similarity in original feature space, and in the Space of Canonical Variables (SCV). The projection of source images in SCV is implemented using a two-dimensional canonical correlation analysis algorithm (2D CCA/2D KLT).