Add special optimization for I420ToI444 and I422ToI444

These functions use (bi)linear filter, to scale U and V planes to the size of Y plane.
This will help enhance the quality of YUV to RGB conversion.

Also added 10bit and 12bit version:
I010ToI410
I210ToI410
I012ToI412
I212ToI412

libyuv_unittest --gtest_filter=LibYUVConvertTest.I42*ToI444*:LibYUVConvertTest.I*1*ToI41*

R=fbarchard@chromium.org

Change-Id: Ie4a711a5ba28f2ff1f44c021f7a5c149022264c5
Bug: libyuv:872
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/2658097
Reviewed-by: Frank Barchard <fbarchard@chromium.org>

1 files changed, 163 insertions, 0 deletions

diff --git a/include/libyuv/scale_row.h b/include/libyuv/scale_row.h
index a386d499..ee77d228 100644
--- a/include/libyuv/scale_row.h
+++ b/include/libyuv/scale_row.h
@@ -77,6 +77,12 @@ extern "C" {
 #if !defined(LIBYUV_DISABLE_X86) && \
     (defined(__x86_64__) || (defined(__i386__) && !defined(_MSC_VER)))
 #define HAS_SCALEUVROWDOWN2BOX_SSSE3
+#define HAS_SCALECOLUP2LINEAR_SSE2
+#define HAS_SCALECOLUP2LINEAR_SSSE3
+#define HAS_SCALEROWUP2LINEAR_SSE2
+#define HAS_SCALEROWUP2LINEAR_SSSE3
+#define HAS_SCALECOLUP2LINEAR_16_SSE2
+#define HAS_SCALEROWUP2LINEAR_16_SSE2
 #endif
 
 // The following are available for gcc/clang x86 platforms, but
@@ -86,6 +92,10 @@ extern "C" {
     (defined(__x86_64__) || defined(__i386__)) && !defined(_MSC_VER) && \
     (defined(CLANG_HAS_AVX2) || defined(GCC_HAS_AVX2))
 #define HAS_SCALEUVROWDOWN2BOX_AVX2
+#define HAS_SCALECOLUP2LINEAR_AVX2
+#define HAS_SCALEROWUP2LINEAR_AVX2
+#define HAS_SCALECOLUP2LINEAR_16_AVX2
+#define HAS_SCALEROWUP2LINEAR_16_AVX2
 #endif
 
 // The following are available on all x86 platforms, but
@@ -114,6 +124,10 @@ extern "C" {
 #define HAS_SCALEROWDOWN4_NEON
 #define HAS_SCALEUVROWDOWN2BOX_NEON
 #define HAS_SCALEUVROWDOWNEVEN_NEON
+#define HAS_SCALECOLUP2LINEAR_NEON
+#define HAS_SCALEROWUP2LINEAR_NEON
+#define HAS_SCALECOLUP2LINEAR_16_NEON
+#define HAS_SCALEROWUP2LINEAR_16_NEON
 #endif
 
 #if !defined(LIBYUV_DISABLE_MSA) && defined(__mips_msa)
@@ -279,6 +293,40 @@ void ScaleRowDown34_1_Box_16_C(const uint16_t* src_ptr,
                                ptrdiff_t src_stride,
                                uint16_t* d,
                                int dst_width);
+
+void ScaleRowUp2_Linear_C(const uint8_t* src_ptr,
+                          uint8_t* dst_ptr,
+                          int dst_width);
+void ScaleRowUp2_Bilinear_C(const uint8_t* src_ptr,
+                            ptrdiff_t src_stride,
+                            uint8_t* dst_ptr,
+                            ptrdiff_t dst_stride,
+                            int dst_width);
+void ScaleRowUp2_Linear_16_C(const uint16_t* src_ptr,
+                             uint16_t* dst_ptr,
+                             int dst_width);
+void ScaleRowUp2_Bilinear_16_C(const uint16_t* src_ptr,
+                               ptrdiff_t src_stride,
+                               uint16_t* dst_ptr,
+                               ptrdiff_t dst_stride,
+                               int dst_width);
+void ScaleRowUp2_Linear_Any_C(const uint8_t* src_ptr,
+                              uint8_t* dst_ptr,
+                              int dst_width);
+void ScaleRowUp2_Bilinear_Any_C(const uint8_t* src_ptr,
+                                ptrdiff_t src_stride,
+                                uint8_t* dst_ptr,
+                                ptrdiff_t dst_stride,
+                                int dst_width);
+void ScaleRowUp2_Linear_16_Any_C(const uint16_t* src_ptr,
+                                 uint16_t* dst_ptr,
+                                 int dst_width);
+void ScaleRowUp2_Bilinear_16_Any_C(const uint16_t* src_ptr,
+                                   ptrdiff_t src_stride,
+                                   uint16_t* dst_ptr,
+                                   ptrdiff_t dst_stride,
+                                   int dst_width);
+
 void ScaleCols_C(uint8_t* dst_ptr,
                  const uint8_t* src_ptr,
                  int dst_width,
@@ -508,6 +556,88 @@ void ScaleRowDown38_2_Box_SSSE3(const uint8_t* src_ptr,
                                 ptrdiff_t src_stride,
                                 uint8_t* dst_ptr,
                                 int dst_width);
+
+void ScaleRowUp2_Linear_SSE2(const uint8_t* src_ptr,
+                             uint8_t* dst_ptr,
+                             int dst_width);
+void ScaleRowUp2_Bilinear_SSE2(const uint8_t* src_ptr,
+                               ptrdiff_t src_stride,
+                               uint8_t* dst_ptr,
+                               ptrdiff_t dst_stride,
+                               int dst_width);
+void ScaleRowUp2_Linear_16_SSE2(const uint16_t* src_ptr,
+                                uint16_t* dst_ptr,
+                                int dst_width);
+void ScaleRowUp2_Bilinear_16_SSE2(const uint16_t* src_ptr,
+                                  ptrdiff_t src_stride,
+                                  uint16_t* dst_ptr,
+                                  ptrdiff_t dst_stride,
+                                  int dst_width);
+void ScaleRowUp2_Linear_SSSE3(const uint8_t* src_ptr,
+                              uint8_t* dst_ptr,
+                              int dst_width);
+void ScaleRowUp2_Bilinear_SSSE3(const uint8_t* src_ptr,
+                                ptrdiff_t src_stride,
+                                uint8_t* dst_ptr,
+                                ptrdiff_t dst_stride,
+                                int dst_width);
+void ScaleRowUp2_Linear_AVX2(const uint8_t* src_ptr,
+                             uint8_t* dst_ptr,
+                             int dst_width);
+void ScaleRowUp2_Bilinear_AVX2(const uint8_t* src_ptr,
+                               ptrdiff_t src_stride,
+                               uint8_t* dst_ptr,
+                               ptrdiff_t dst_stride,
+                               int dst_width);
+void ScaleRowUp2_Linear_16_AVX2(const uint16_t* src_ptr,
+                                uint16_t* dst_ptr,
+                                int dst_width);
+void ScaleRowUp2_Bilinear_16_AVX2(const uint16_t* src_ptr,
+                                  ptrdiff_t src_stride,
+                                  uint16_t* dst_ptr,
+                                  ptrdiff_t dst_stride,
+                                  int dst_width);
+void ScaleRowUp2_Linear_Any_SSE2(const uint8_t* src_ptr,
+                                 uint8_t* dst_ptr,
+                                 int dst_width);
+void ScaleRowUp2_Bilinear_Any_SSE2(const uint8_t* src_ptr,
+                                   ptrdiff_t src_stride,
+                                   uint8_t* dst_ptr,
+                                   ptrdiff_t dst_stride,
+                                   int dst_width);
+void ScaleRowUp2_Linear_16_Any_SSE2(const uint16_t* src_ptr,
+                                    uint16_t* dst_ptr,
+                                    int dst_width);
+void ScaleRowUp2_Bilinear_16_Any_SSE2(const uint16_t* src_ptr,
+                                      ptrdiff_t src_stride,
+                                      uint16_t* dst_ptr,
+                                      ptrdiff_t dst_stride,
+                                      int dst_width);
+void ScaleRowUp2_Linear_Any_SSSE3(const uint8_t* src_ptr,
+                                  uint8_t* dst_ptr,
+                                  int dst_width);
+void ScaleRowUp2_Bilinear_Any_SSSE3(const uint8_t* src_ptr,
+                                    ptrdiff_t src_stride,
+                                    uint8_t* dst_ptr,
+                                    ptrdiff_t dst_stride,
+                                    int dst_width);
+void ScaleRowUp2_Linear_Any_AVX2(const uint8_t* src_ptr,
+                                 uint8_t* dst_ptr,
+                                 int dst_width);
+void ScaleRowUp2_Bilinear_Any_AVX2(const uint8_t* src_ptr,
+                                   ptrdiff_t src_stride,
+                                   uint8_t* dst_ptr,
+                                   ptrdiff_t dst_stride,
+                                   int dst_width);
+void ScaleRowUp2_Linear_16_Any_AVX2(const uint16_t* src_ptr,
+                                    uint16_t* dst_ptr,
+                                    int dst_width);
+void ScaleRowUp2_Bilinear_16_Any_AVX2(const uint16_t* src_ptr,
+                                      ptrdiff_t src_stride,
+                                      uint16_t* dst_ptr,
+                                      ptrdiff_t dst_stride,
+                                      int dst_width);
+
 void ScaleRowDown2_Any_SSSE3(const uint8_t* src_ptr,
                              ptrdiff_t src_stride,
                              uint8_t* dst_ptr,
@@ -1143,6 +1273,39 @@ void ScaleRowDown38_2_Box_Any_NEON(const uint8_t* src_ptr,
                                    uint8_t* dst_ptr,
                                    int dst_width);
 
+void ScaleRowUp2_Linear_NEON(const uint8_t* src_ptr,
+                             uint8_t* dst_ptr,
+                             int dst_width);
+void ScaleRowUp2_Bilinear_NEON(const uint8_t* src_ptr,
+                               ptrdiff_t src_stride,
+                               uint8_t* dst_ptr,
+                               ptrdiff_t dst_stride,
+                               int dst_width);
+void ScaleRowUp2_Linear_16_NEON(const uint16_t* src_ptr,
+                                uint16_t* dst_ptr,
+                                int dst_width);
+void ScaleRowUp2_Bilinear_16_NEON(const uint16_t* src_ptr,
+                                  ptrdiff_t src_stride,
+                                  uint16_t* dst_ptr,
+                                  ptrdiff_t dst_stride,
+                                  int dst_width);
+void ScaleRowUp2_Linear_Any_NEON(const uint8_t* src_ptr,
+                                 uint8_t* dst_ptr,
+                                 int dst_width);
+void ScaleRowUp2_Bilinear_Any_NEON(const uint8_t* src_ptr,
+                                   ptrdiff_t src_stride,
+                                   uint8_t* dst_ptr,
+                                   ptrdiff_t dst_stride,
+                                   int dst_width);
+void ScaleRowUp2_Linear_16_Any_NEON(const uint16_t* src_ptr,
+                                    uint16_t* dst_ptr,
+                                    int dst_width);
+void ScaleRowUp2_Bilinear_16_Any_NEON(const uint16_t* src_ptr,
+                                      ptrdiff_t src_stride,
+                                      uint16_t* dst_ptr,
+                                      ptrdiff_t dst_stride,
+                                      int dst_width);
+
 void ScaleAddRow_NEON(const uint8_t* src_ptr, uint16_t* dst_ptr, int src_width);
 void ScaleAddRow_Any_NEON(const uint8_t* src_ptr,
                           uint16_t* dst_ptr,