Merge pull request #15358 from ChipKerchner:imgwarpToHal
authorChip Kerchner <49959681+ChipKerchner@users.noreply.github.com>
Sat, 31 Aug 2019 10:47:58 +0000 (06:47 -0400)
committerAlexander Alekhin <alexander.a.alekhin@gmail.com>
Sat, 31 Aug 2019 10:47:58 +0000 (13:47 +0300)
commit26228e6b4de5f897973557469a608ddf44bb20a6
tree5f3fae9a2fd515ae8f38427110ab49f86fd64c1d
parentf224d740a32f3cff3bc490322c3e875443b79779
Merge pull request #15358 from ChipKerchner:imgwarpToHal

* Convert ImgWarp from SSE SIMD to HAL - 2.8x faster on Power (VSX) and 15% speedup on x86

* Change compile flag from CV_SIMD128 to CV_SIMD128_64F for use of v_float64x2 type

* Changing WarpPerspectiveLine from class functions and dispatching to static functions.

* Re-add dynamic runtime and dispatch execution.

* RRestore SSE4_1 optimizations inside opt_SSE4_1 namespace
modules/imgproc/src/imgwarp.cpp
modules/imgproc/src/imgwarp.hpp