potok/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2026-07-31 00:03:03 +04:00
Code Packages Releases Wiki Activity

Merge remote-tracking branch 'upstream/3.4' into merge-3.4

Browse Source
This commit is contained in:
Alexander Alekhin
2019-06-05 19:11:22 +03:00
parent e329c84d5e 51631b90af
commit f3de2b4be7
155 changed files with 5856 additions and 1800 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/imgproc/src/contours.cpp
+26 -8
View File
@@ -1061,10 +1061,16 @@ cvFindNextContour( CvContourScanner scanner )
}
else
{
v_uint8 v_prev = vx_setall_u8((uchar)prev);
for (; x <= width - v_uint8::nlanes; x += v_uint8::nlanes)
#if CV_SIMD_WIDTH > 16
v_uint8 vx_prev = vx_setall_u8((uchar)prev);
while (x <= width - v_uint8::nlanes &&
v_check_all(vx_load((uchar*)(img + x)) == vx_prev))
x += v_uint8::nlanes;
#endif
v_uint8x16 v_prev = v_setall_u8((uchar)prev);
for (; x <= width - v_uint8x16::nlanes; x += v_uint8x16::nlanes)
{
unsigned int mask = (unsigned int)v_signmask(vx_load((uchar*)(img + x)) != v_prev);
unsigned int mask = (unsigned int)v_signmask(v_load((uchar*)(img + x)) != v_prev);
if (mask)
{
p = img[(x += cv::trailingZeros32(mask))];
@@ -1328,10 +1334,16 @@ CvLinkedRunPoint;
inline int findStartContourPoint(uchar *src_data, CvSize img_size, int j)
{
#if CV_SIMD
v_uint8 v_zero = vx_setzero_u8();
for (; j <= img_size.width - v_uint8::nlanes; j += v_uint8::nlanes)
#if CV_SIMD_WIDTH > 16
v_uint8 vx_zero = vx_setzero_u8();
while (j <= img_size.width - v_uint8::nlanes &&
v_check_all(vx_load((uchar*)(src_data + j)) == vx_zero))
j += v_uint8::nlanes;
#endif
v_uint8x16 v_zero = v_setzero_u8();
for (; j <= img_size.width - v_uint8x16::nlanes; j += v_uint8x16::nlanes)
{
unsigned int mask = (unsigned int)v_signmask(vx_load((uchar*)(src_data + j)) != v_zero);
unsigned int mask = (unsigned int)v_signmask(v_load((uchar*)(src_data + j)) != v_zero);
if (mask)
{
j += cv::trailingZeros32(mask);
@@ -1353,10 +1365,16 @@ inline int findEndContourPoint(uchar *src_data, CvSize img_size, int j)
}
else
{
v_uint8 v_zero = vx_setzero_u8();
#if CV_SIMD_WIDTH > 16
v_uint8 vx_zero = vx_setzero_u8();
while (j <= img_size.width - v_uint8::nlanes &&
v_check_all(vx_load((uchar*)(src_data + j)) != vx_zero))
j += v_uint8::nlanes;
#endif
v_uint8x16 v_zero = v_setzero_u8();
for (; j <= img_size.width - v_uint8::nlanes; j += v_uint8::nlanes)
{
unsigned int mask = (unsigned int)v_signmask(vx_load((uchar*)(src_data + j)) == v_zero);
unsigned int mask = (unsigned int)v_signmask(v_load((uchar*)(src_data + j)) == v_zero);
if (mask)
{
j += cv::trailingZeros32(mask);
modules/imgproc/src/imgwarp.cpp
+9 -6
View File
@@ -436,6 +436,9 @@ struct RemapNoVec
#if CV_SIMD128
typedef unsigned short CV_DECL_ALIGNED(1) unaligned_ushort;
typedef int CV_DECL_ALIGNED(1) unaligned_int;
struct RemapVec_8u
{
int operator()( const Mat& _src, void* _dst, const short* XY,
@@ -461,8 +464,8 @@ struct RemapVec_8u
{ \
v_uint8x16 rrggbb, dummy; \
v_uint16x8 rrggbb8, dummy8; \
v_uint8x16 rgb0 = v_reinterpret_as_u8(v_int32x4(*(int*)(p), 0, 0, 0)); \
v_uint8x16 rgb1 = v_reinterpret_as_u8(v_int32x4(*(int*)(p + 3), 0, 0, 0)); \
v_uint8x16 rgb0 = v_reinterpret_as_u8(v_int32x4(*(unaligned_int*)(p), 0, 0, 0)); \
v_uint8x16 rgb1 = v_reinterpret_as_u8(v_int32x4(*(unaligned_int*)(p + 3), 0, 0, 0)); \
v_zip(rgb0, rgb1, rrggbb, dummy); \
v_expand(rrggbb, rrggbb8, dummy8); \
result = v_reinterpret_as_s16(rrggbb8); \
@@ -480,15 +483,15 @@ struct RemapVec_8u
CV_DbgAssert(p <= src_limit_8bytes); \
v_uint8x16 rrggbbaa, dummy; \
v_uint16x8 rrggbbaa8, dummy8; \
v_uint8x16 rgba0 = v_reinterpret_as_u8(v_int32x4(*(int*)(p), 0, 0, 0)); \
v_uint8x16 rgba1 = v_reinterpret_as_u8(v_int32x4(*(int*)(p + v_int32x4::nlanes), 0, 0, 0)); \
v_uint8x16 rgba0 = v_reinterpret_as_u8(v_int32x4(*(unaligned_int*)(p), 0, 0, 0)); \
v_uint8x16 rgba1 = v_reinterpret_as_u8(v_int32x4(*(unaligned_int*)(p + v_int32x4::nlanes), 0, 0, 0)); \
v_zip(rgba0, rgba1, rrggbbaa, dummy); \
v_expand(rrggbbaa, rrggbbaa8, dummy8); \
result = v_reinterpret_as_s16(rrggbbaa8); \
}
#define CV_PICK_AND_PACK4(base,offset) \
v_uint16x8(*(ushort*)(base + offset[0]), *(ushort*)(base + offset[1]), \
*(ushort*)(base + offset[2]), *(ushort*)(base + offset[3]), \
v_uint16x8(*(unaligned_ushort*)(base + offset[0]), *(unaligned_ushort*)(base + offset[1]), \
*(unaligned_ushort*)(base + offset[2]), *(unaligned_ushort*)(base + offset[3]), \
0, 0, 0, 0)
if( cn == 1 )
modules/imgproc/src/resize.cpp
+2
View File
@@ -2148,6 +2148,7 @@ public:
v_zip(t0, t3, s0, s1); v_zip(t1, t4, s2, s3); v_zip(t2, t5, s4, s5);
bl = s0 + s3; gl = s1 + s4; rl = s2 + s5;
#elif CV_SIMD_WIDTH == 64
v_zip(t0, t3, s0, s1); v_zip(t1, t4, s2, s3); v_zip(t2, t5, s4, s5);
v_zip(s0, s3, t0, t1); v_zip(s1, s4, t2, t3); v_zip(s2, s5, t4, t5);
bl = t0 + t3; gl = t1 + t4; rl = t2 + t5;
#endif
@@ -2167,6 +2168,7 @@ public:
v_zip(t0, t3, s0, s1); v_zip(t1, t4, s2, s3); v_zip(t2, t5, s4, s5);
bh = s0 + s3; gh = s1 + s4; rh = s2 + s5;
#elif CV_SIMD_WIDTH == 64
v_zip(t0, t3, s0, s1); v_zip(t1, t4, s2, s3); v_zip(t2, t5, s4, s5);
v_zip(s0, s3, t0, t1); v_zip(s1, s4, t2, t3); v_zip(s2, s5, t4, t5);
bh = t0 + t3; gh = t1 + t4; rh = t2 + t5;
#endif
modules/imgproc/src/sumpixels.cpp
+4 -4
View File
@@ -127,7 +127,7 @@ struct Integral_SIMD<uchar, int, double>
{
v_int16 el8 = v_reinterpret_as_s16(vx_load_expand(src_row + j));
v_int32 el4l, el4h;
#if CV_AVX2
#if CV_AVX2 && CV_SIMD_WIDTH == 32
__m256i vsum = _mm256_add_epi16(el8.val, _mm256_slli_si256(el8.val, 2));
vsum = _mm256_add_epi16(vsum, _mm256_slli_si256(vsum, 4));
vsum = _mm256_add_epi16(vsum, _mm256_slli_si256(vsum, 8));
@@ -138,7 +138,7 @@ struct Integral_SIMD<uchar, int, double>
#else
el8 += v_rotate_left<1>(el8);
el8 += v_rotate_left<2>(el8);
#if CV_SIMD_WIDTH == 32
#if CV_SIMD_WIDTH >= 32
el8 += v_rotate_left<4>(el8);
#if CV_SIMD_WIDTH == 64
el8 += v_rotate_left<8>(el8);
@@ -194,7 +194,7 @@ struct Integral_SIMD<uchar, float, double>
{
v_int16 el8 = v_reinterpret_as_s16(vx_load_expand(src_row + j));
v_float32 el4l, el4h;
#if CV_AVX2
#if CV_AVX2 && CV_SIMD_WIDTH == 32
__m256i vsum = _mm256_add_epi16(el8.val, _mm256_slli_si256(el8.val, 2));
vsum = _mm256_add_epi16(vsum, _mm256_slli_si256(vsum, 4));
vsum = _mm256_add_epi16(vsum, _mm256_slli_si256(vsum, 8));
@@ -205,7 +205,7 @@ struct Integral_SIMD<uchar, float, double>
#else
el8 += v_rotate_left<1>(el8);
el8 += v_rotate_left<2>(el8);
#if CV_SIMD_WIDTH == 32
#if CV_SIMD_WIDTH >= 32
el8 += v_rotate_left<4>(el8);
#if CV_SIMD_WIDTH == 64
el8 += v_rotate_left<8>(el8);