+add Base implementation of function BgrToYuv444pV2.

docs/2023.html

@@ -39,6 +39,7 @@ <h5>New features</h5>
 <ul>
  <li>Base implementation, SSE4.1, AVX2, AVX-512BW optimizations of function BgrToYuv420pV2.</li>
  <li>Base implementation, SSE4.1, AVX2 optimizations of function BgrToYuv422pV2.</li>
+ <li>Base implementation of function BgrToYuv444pV2.</li>
 </ul>
 <ul>
  <li>Error in AVX-512BW optimizations of function SynetSoftmaxLayerForward.</li>
@@ -49,6 +50,7 @@ <h5>New features</h5>
 <ul>
  <li>Tests for verifying functionality of function BgrToYuv420pV2.</li>
  <li>Tests for verifying functionality of function BgrToYuv422pV2.</li>
+ <li>Tests for verifying functionality of function BgrToYuv444pV2.</li>
 </ul>
 
 <a href="#HOME">Home</a>

src/Simd/SimdBase.h

@@ -178,6 +178,9 @@ namespace Simd
 
         void BgrToYuv444p(const uint8_t * bgr, size_t width, size_t height, size_t bgrStride, uint8_t * y, size_t yStride, uint8_t * u, size_t uStride, uint8_t * v, size_t vStride);
 
+        void BgrToYuv444pV2(const uint8_t* bgr, size_t bgrStride, size_t width, size_t height,
+            uint8_t* y, size_t yStride, uint8_t* u, size_t uStride, uint8_t* v, size_t vStride, SimdYuvType yuvType);
+
         void Binarization(const uint8_t * src, size_t srcStride, size_t width, size_t height,
             uint8_t value, uint8_t positive, uint8_t negative, uint8_t * dst, size_t dstStride, SimdCompareType compareType);
 

src/Simd/SimdBaseBgrToYuv.cpp

@@ -210,5 +210,43 @@ namespace Simd
                 assert(0);
             }
         }
+
+        //-------------------------------------------------------------------------------------------------
+
+        template <class YuvType> SIMD_INLINE void BgrToYuv444pV2(const uint8_t* bgr, uint8_t* y, uint8_t* u, uint8_t* v)
+        {
+            const int blue = bgr[0], green = bgr[1], red = bgr[2];
+            y[0] = BgrToY<YuvType>(blue, green, red);
+            u[0] = BgrToU<YuvType>(blue, green, red);
+            v[0] = BgrToV<YuvType>(blue, green, red);
+        }
+
+        template <class YuvType> void BgrToYuv444pV2(const uint8_t* bgr, size_t bgrStride, size_t width, size_t height,
+            uint8_t* y, size_t yStride, uint8_t* u, size_t uStride, uint8_t* v, size_t vStride)
+        {
+            for (size_t row = 0; row < height; row += 1)
+            {
+                for (size_t col = 0; col < width; col++)
+                    BgrToYuv444pV2<YuvType>(bgr + col * 3, y + col, u + col, v + col);
+                y += yStride;
+                u += uStride;
+                v += vStride;
+                bgr += bgrStride;
+            }
+        }
+
+        void BgrToYuv444pV2(const uint8_t* bgr, size_t bgrStride, size_t width, size_t height,
+            uint8_t* y, size_t yStride, uint8_t* u, size_t uStride, uint8_t* v, size_t vStride, SimdYuvType yuvType)
+        {
+            switch (yuvType)
+            {
+            case SimdYuvBt601: BgrToYuv444pV2<Bt601>(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride); break;
+            case SimdYuvBt709: BgrToYuv444pV2<Bt709>(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride); break;
+            case SimdYuvBt2020: BgrToYuv444pV2<Bt2020>(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride); break;
+            case SimdYuvTrect871: BgrToYuv444pV2<Trect871>(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride); break;
+            default:
+                assert(0);
+            }
+        }
     }
 }

src/Simd/SimdLib.cpp

@@ -1671,7 +1671,7 @@ SIMD_API void SimdBgrToYuv422pV2(const uint8_t* bgr, size_t bgrStride, size_t wi
 #endif
 //#ifdef SIMD_NEON_ENABLE
 //    if (Neon::Enable && width >= Neon::DA)
-//        Neon::BgrToYuv420pV2(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride, yuvType);
+//        Neon::BgrToYuv422pV2(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride, yuvType);
 //    else
 //#endif
         Base::BgrToYuv422pV2(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride, yuvType);
@@ -1708,6 +1708,33 @@ SIMD_API void SimdBgrToYuv444p(const uint8_t * bgr, size_t width, size_t height,
         Base::BgrToYuv444p(bgr, width, height, bgrStride, y, yStride, u, uStride, v, vStride);
 }
 
+SIMD_API void SimdBgrToYuv444pV2(const uint8_t* bgr, size_t bgrStride, size_t width, size_t height,
+    uint8_t* y, size_t yStride, uint8_t* u, size_t uStride, uint8_t* v, size_t vStride, SimdYuvType yuvType)
+{
+    SIMD_EMPTY();
+//#ifdef SIMD_AVX512BW_ENABLE
+//    if (Avx512bw::Enable)
+//        Avx512bw::BgrToYuv444pV2(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride, yuvType);
+//    else
+//#endif
+//#ifdef SIMD_AVX2_ENABLE
+//    if (Avx2::Enable && width >= Avx2::A)
+//        Avx2::BgrToYuv444pV2(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride, yuvType);
+//    else
+//#endif
+//#ifdef SIMD_SSE41_ENABLE
+//    if (Sse41::Enable && width >= Sse41::A)
+//        Sse41::BgrToYuv444pV2(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride, yuvType);
+//    else
+//#endif
+//#ifdef SIMD_NEON_ENABLE
+//    if (Neon::Enable && width >= Neon::A)
+//        Neon::BgrToYuv444pV2(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride, yuvType);
+//    else
+//#endif
+        Base::BgrToYuv444pV2(bgr, bgrStride, width, height, y, yStride, u, uStride, v, vStride, yuvType);
+}
+
 SIMD_API void SimdBinarization(const uint8_t * src, size_t srcStride, size_t width, size_t height,
                   uint8_t value, uint8_t positive, uint8_t negative, uint8_t * dst, size_t dstStride, SimdCompareType compareType)
 {

src/Simd/SimdLib.h

@@ -2241,6 +2241,28 @@ extern "C"
     */
     SIMD_API void SimdBgrToYuv444p(const uint8_t * bgr, size_t width, size_t height, size_t bgrStride, uint8_t * y, size_t yStride, uint8_t * u, size_t uStride, uint8_t * v, size_t vStride);
 
+    /*! @ingroup bgr_conversion
+
+        \fn void SimdBgrToYuv444pV2(const uint8_t * bgr, size_t bgrStride, size_t width, size_t height, uint8_t * y, size_t yStride, uint8_t * u, size_t uStride, uint8_t * v, size_t vStride, SimdYuvType yuvType);
+
+        \short Converts 24-bit BGR image to YUV444P.
+
+        The input BGR and output Y, U and V images must have the same width and height.
+
+        \param [in] bgr - a pointer to pixels data of input 24-bit BGR image.
+        \param [in] bgrStride - a row size of the BGR image.
+        \param [in] width - an image width.
+        \param [in] height - an image height.
+        \param [out] y - a pointer to pixels data of output 8-bit image with Y color plane.
+        \param [in] yStride - a row size of the y image.
+        \param [out] u - a pointer to pixels data of output 8-bit image with U color plane.
+        \param [in] uStride - a row size of the u image.
+        \param [out] v - a pointer to pixels data of output 8-bit image with V color plane.
+        \param [in] vStride - a row size of the v image.
+        \param [in] yuvType - a type of output YUV image (see descriprion of ::SimdYuvType).
+    */
+    SIMD_API void SimdBgrToYuv444pV2(const uint8_t* bgr, size_t bgrStride, size_t width, size_t height, uint8_t* y, size_t yStride, uint8_t* u, size_t uStride, uint8_t* v, size_t vStride, SimdYuvType yuvType);
+
     /*! @ingroup binarization
 
         \fn void SimdBinarization(const uint8_t * src, size_t srcStride, size_t width, size_t height, uint8_t value, uint8_t positive, uint8_t negative, uint8_t * dst, size_t dstStride, SimdCompareType compareType);

src/Test/Test.cpp

@@ -119,6 +119,7 @@ namespace Test
     TEST_ADD_GROUP_A0(BgrToYuv422p);
     TEST_ADD_GROUP_A0(BgrToYuv422pV2);
     TEST_ADD_GROUP_A0(BgrToYuv444p);
+    TEST_ADD_GROUP_A0(BgrToYuv444pV2);
     TEST_ADD_GROUP_A0(Uyvy422ToYuv420p);
     TEST_ADD_GROUP_A0(BgraToYuva420p);
     TEST_ADD_GROUP_A0(BgraToYuva420pV2);

src/Test/TestAnyToYuv.cpp

@@ -549,10 +549,39 @@ namespace Test
 //            result = result && AnyToYuvV2AutoTest(View::Bgr24, 2, 1, FUNC_YUV2(Simd::Avx512bw::BgrToYuv422pV2), FUNC_YUV2(SimdBgrToYuv422pV2));
 //#endif 
 
-    //#ifdef SIMD_NEON_ENABLE
-    //        if (Simd::Neon::Enable)
-    //            result = result && AnyToYuvV2AutoTest(View::Bgr24, 2, 1, FUNC_YUV2(Simd::Neon::BgrToYuv422pV2), FUNC_YUV2(SimdBgrToYuv422pV2));
-    //#endif
+//#ifdef SIMD_NEON_ENABLE
+//        if (Simd::Neon::Enable)
+//            result = result && AnyToYuvV2AutoTest(View::Bgr24, 2, 1, FUNC_YUV2(Simd::Neon::BgrToYuv422pV2), FUNC_YUV2(SimdBgrToYuv422pV2));
+//#endif
+
+        return result;
+    }
+
+    bool BgrToYuv444pV2AutoTest()
+    {
+        bool result = true;
+
+        result = result && AnyToYuvV2AutoTest(View::Bgr24, 1, 1, FUNC_YUV2(Simd::Base::BgrToYuv444pV2), FUNC_YUV2(SimdBgrToYuv444pV2));
+
+//#ifdef SIMD_SSE41_ENABLE
+//        if (Simd::Sse41::Enable)
+//            result = result && AnyToYuvV2AutoTest(View::Bgr24, 1, 1, FUNC_YUV2(Simd::Sse41::BgrToYuv444pV2), FUNC_YUV2(SimdBgrToYuv444pV2));
+//#endif 
+//
+//#ifdef SIMD_AVX2_ENABLE
+//        if (Simd::Avx2::Enable)
+//            result = result && AnyToYuvV2AutoTest(View::Bgr24, 1, 1, FUNC_YUV2(Simd::Avx2::BgrToYuv444pV2), FUNC_YUV2(SimdBgrToYuv444pV2));
+//#endif 
+
+//#ifdef SIMD_AVX512BW_ENABLE
+//        if (Simd::Avx512bw::Enable)
+//            result = result && AnyToYuvV2AutoTest(View::Bgr24, 1, 1, FUNC_YUV2(Simd::Avx512bw::BgrToYuv444pV2), FUNC_YUV2(SimdBgrToYuv444pV2));
+//#endif 
+
+//#ifdef SIMD_NEON_ENABLE
+//        if (Simd::Neon::Enable)
+//            result = result && AnyToYuvV2AutoTest(View::Bgr24, 1, 1, FUNC_YUV2(Simd::Neon::BgrToYuv444pV2), FUNC_YUV2(SimdBgrToYuv444pV2));
+//#endif
 
         return result;
     }