significantly improved parallel non-local means by using granularity parameter in...
authorVadim Pisarevsky <vadim.pisarevsky@gmail.com>
Thu, 14 May 2015 12:39:42 +0000 (15:39 +0300)
committerVadim Pisarevsky <vadim.pisarevsky@gmail.com>
Thu, 14 May 2015 12:39:42 +0000 (15:39 +0300)
commitb37aaa8303129dfeb6f2e7b151493055a0ea9ee6
treebfc866caee4c393ce1d58f5777d0dedce25abd40
parentfeb5b6aa93a12859ba0fa9d99eff5d0b1cccb08f
significantly improved parallel non-local means by using granularity parameter in parallel_for_ loop. Because the algorithm deals with sliding sums, it's essential that each thread has enough work to do, otherwise the algorithm gets higher theoretical complexity and thus there is no speedup comparing to 1-thread code (at best).
modules/photo/src/denoising.cpp
modules/photo/test/test_denoising.cpp