Naive nearest neighbor finder

[dst,nidx] = cv.batchDistance(src1, src2)
[...] = cv.batchDistance(..., 'OptionName', optionValue, ...)

Input

• src1 samples matrix of size N1-by-M, type either 'single' or 'uint8'
• src2 samples matrix of size N2-by-M, type either 'single' or 'uint8'

Output

• dst distance matrix (see description below).
• nidx zero-based indices of nearest neighbors (matrix of size N1-by-K). Only computed if K>0, otherwise an empty matrix is returned.

Options

• DType default -1
• NormType Distance metric used. Default 'L2'
• K Number of nearest neighbors in to find. If K=0 (the default), the full pairwaise distance matrix is computed (of size N1-by-N2), otherwise only distances to the K-nearest neighbors is returned (matrix of size N1-by-K).
• Mask Not set by default.
• Update default 0.
• CrossCheck default false.

See http://en.wikipedia.org/wiki/Nearest_neighbor_search