LBP - local binary pattern

LBP (histogram and image) rotational invariant

LBP (histogram and image) for image block. rotational invariant and have uniform option.
uniform - vector with 2 transitions top
example : 00000000 (0 transitions)
00000111 (1 transitions)
00001000 (2 transitions)

and we also say that (01000000 and 00001000) are the same and call them '00000001' or '1'
this assumption gives us 'rotate invariant'.

if there are more transitions we say the vector is in class '9' - not uniform

example : 00000000 - class 0
00000111 - class 3
00001000 - class 1
01010101 - class 9

so in DoUniform you can see that under this assumptions you get only 10 options and every bin will get more 'vectors' that are in that class

in NOTuniform we have 2^8 options and the vectors will split to all of them so we cn expect lower values in histogram.