Convolutional Hough Matching

What is Convolutional Hough Matching (CHM)? Convolutional Hough Matching (CHM) is a geometric matching algorithm that uses a trainable neural layer for non-rigid matching. This powerful algorithm distributes similarities of candidate matches over a geometric transformation space and evaluates them in a convolutational manner. The semi-isotropic high-dimensional kernel featuring a small number of interpretable parameters learns non-rigid matching with a minimal number of training examples, makin