#include "precomp.hpp"
#include "opencl_kernels_imgproc.hpp"
+#include "opencv2/core/hal/intrin.hpp"
-#if CV_NEON && defined(__aarch64__)
-#include <arm_neon.h>
-namespace cv {
-// Workaround with missing definitions of vreinterpretq_u64_f64/vreinterpretq_f64_u64
-template <typename T> static inline
-uint64x2_t vreinterpretq_u64_f64(T a)
-{
- return (uint64x2_t) a;
-}
-template <typename T> static inline
-float64x2_t vreinterpretq_f64_u64(T a)
-{
- return (float64x2_t) a;
-}
-} // namespace cv
+#ifdef HAVE_OPENVX
+#define IVX_HIDE_INFO_WARNINGS
+#define IVX_USE_OPENCV
+#include "ivx.hpp"
#endif
namespace cv
static void
thresh_8u( const Mat& _src, Mat& _dst, uchar thresh, uchar maxval, int type )
{
- int i, j, j_scalar = 0;
- uchar tab[256];
Size roi = _src.size();
roi.width *= _src.channels();
size_t src_step = _src.step;
switch( type )
{
case THRESH_TRUNC:
-#ifndef HAVE_IPP_ICV_ONLY
- if (_src.data == _dst.data && ippiThreshold_GT_8u_C1IR(_dst.ptr(), (int)dst_step, sz, thresh) >= 0)
+ if (_src.data == _dst.data && CV_INSTRUMENT_FUN_IPP(ippiThreshold_GT_8u_C1IR, _dst.ptr(), (int)dst_step, sz, thresh) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
}
-#endif
- if (ippiThreshold_GT_8u_C1R(_src.ptr(), (int)src_step, _dst.ptr(), (int)dst_step, sz, thresh) >= 0)
+ if (CV_INSTRUMENT_FUN_IPP(ippiThreshold_GT_8u_C1R, _src.ptr(), (int)src_step, _dst.ptr(), (int)dst_step, sz, thresh) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
setIppErrorStatus();
break;
case THRESH_TOZERO:
-#ifndef HAVE_IPP_ICV_ONLY
- if (_src.data == _dst.data && ippiThreshold_LTVal_8u_C1IR(_dst.ptr(), (int)dst_step, sz, thresh+1, 0) >= 0)
+ if (_src.data == _dst.data && CV_INSTRUMENT_FUN_IPP(ippiThreshold_LTVal_8u_C1IR, _dst.ptr(), (int)dst_step, sz, thresh+1, 0) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
}
-#endif
- if (ippiThreshold_LTVal_8u_C1R(_src.ptr(), (int)src_step, _dst.ptr(), (int)dst_step, sz, thresh+1, 0) >= 0)
+ if (CV_INSTRUMENT_FUN_IPP(ippiThreshold_LTVal_8u_C1R, _src.ptr(), (int)src_step, _dst.ptr(), (int)dst_step, sz, thresh + 1, 0) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
setIppErrorStatus();
break;
case THRESH_TOZERO_INV:
-#ifndef HAVE_IPP_ICV_ONLY
- if (_src.data == _dst.data && ippiThreshold_GTVal_8u_C1IR(_dst.ptr(), (int)dst_step, sz, thresh, 0) >= 0)
+ if (_src.data == _dst.data && CV_INSTRUMENT_FUN_IPP(ippiThreshold_GTVal_8u_C1IR, _dst.ptr(), (int)dst_step, sz, thresh, 0) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
}
-#endif
- if (ippiThreshold_GTVal_8u_C1R(_src.ptr(), (int)src_step, _dst.ptr(), (int)dst_step, sz, thresh, 0) >= 0)
+ if (CV_INSTRUMENT_FUN_IPP(ippiThreshold_GTVal_8u_C1R, _src.ptr(), (int)src_step, _dst.ptr(), (int)dst_step, sz, thresh, 0) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
}
#endif
- switch( type )
- {
- case THRESH_BINARY:
- for( i = 0; i <= thresh; i++ )
- tab[i] = 0;
- for( ; i < 256; i++ )
- tab[i] = maxval;
- break;
- case THRESH_BINARY_INV:
- for( i = 0; i <= thresh; i++ )
- tab[i] = maxval;
- for( ; i < 256; i++ )
- tab[i] = 0;
- break;
- case THRESH_TRUNC:
- for( i = 0; i <= thresh; i++ )
- tab[i] = (uchar)i;
- for( ; i < 256; i++ )
- tab[i] = thresh;
- break;
- case THRESH_TOZERO:
- for( i = 0; i <= thresh; i++ )
- tab[i] = 0;
- for( ; i < 256; i++ )
- tab[i] = (uchar)i;
- break;
- case THRESH_TOZERO_INV:
- for( i = 0; i <= thresh; i++ )
- tab[i] = (uchar)i;
- for( ; i < 256; i++ )
- tab[i] = 0;
- break;
- default:
- CV_Error( CV_StsBadArg, "Unknown threshold type" );
- }
-
-#if CV_SSE2
- if( checkHardwareSupport(CV_CPU_SSE2) )
+ int j = 0;
+ const uchar* src = _src.ptr();
+ uchar* dst = _dst.ptr();
+#if CV_SIMD128
+ bool useSIMD = checkHardwareSupport( CV_CPU_SSE2 ) || checkHardwareSupport( CV_CPU_NEON );
+ if( useSIMD )
{
- __m128i _x80 = _mm_set1_epi8('\x80');
- __m128i thresh_u = _mm_set1_epi8(thresh);
- __m128i thresh_s = _mm_set1_epi8(thresh ^ 0x80);
- __m128i maxval_ = _mm_set1_epi8(maxval);
- j_scalar = roi.width & -8;
+ v_uint8x16 thresh_u = v_setall_u8( thresh );
+ v_uint8x16 maxval16 = v_setall_u8( maxval );
- for( i = 0; i < roi.height; i++ )
+ switch( type )
{
- const uchar* src = _src.ptr() + src_step*i;
- uchar* dst = _dst.ptr() + dst_step*i;
-
- switch( type )
+ case THRESH_BINARY:
+ for( int i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- case THRESH_BINARY:
- for( j = 0; j <= roi.width - 32; j += 32 )
- {
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 16) );
- v0 = _mm_cmpgt_epi8( _mm_xor_si128(v0, _x80), thresh_s );
- v1 = _mm_cmpgt_epi8( _mm_xor_si128(v1, _x80), thresh_s );
- v0 = _mm_and_si128( v0, maxval_ );
- v1 = _mm_and_si128( v1, maxval_ );
- _mm_storeu_si128( (__m128i*)(dst + j), v0 );
- _mm_storeu_si128( (__m128i*)(dst + j + 16), v1 );
- }
-
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128i v0 = _mm_loadl_epi64( (const __m128i*)(src + j) );
- v0 = _mm_cmpgt_epi8( _mm_xor_si128(v0, _x80), thresh_s );
- v0 = _mm_and_si128( v0, maxval_ );
- _mm_storel_epi64( (__m128i*)(dst + j), v0 );
- }
- break;
-
- case THRESH_BINARY_INV:
- for( j = 0; j <= roi.width - 32; j += 32 )
- {
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 16) );
- v0 = _mm_cmpgt_epi8( _mm_xor_si128(v0, _x80), thresh_s );
- v1 = _mm_cmpgt_epi8( _mm_xor_si128(v1, _x80), thresh_s );
- v0 = _mm_andnot_si128( v0, maxval_ );
- v1 = _mm_andnot_si128( v1, maxval_ );
- _mm_storeu_si128( (__m128i*)(dst + j), v0 );
- _mm_storeu_si128( (__m128i*)(dst + j + 16), v1 );
- }
-
- for( ; j <= roi.width - 8; j += 8 )
+ for( j = 0; j <= roi.width - 16; j += 16 )
{
- __m128i v0 = _mm_loadl_epi64( (const __m128i*)(src + j) );
- v0 = _mm_cmpgt_epi8( _mm_xor_si128(v0, _x80), thresh_s );
- v0 = _mm_andnot_si128( v0, maxval_ );
- _mm_storel_epi64( (__m128i*)(dst + j), v0 );
+ v_uint8x16 v0;
+ v0 = v_load( src + j );
+ v0 = thresh_u < v0;
+ v0 = v0 & maxval16;
+ v_store( dst + j, v0 );
}
- break;
-
- case THRESH_TRUNC:
- for( j = 0; j <= roi.width - 32; j += 32 )
- {
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 16) );
- v0 = _mm_subs_epu8( v0, _mm_subs_epu8( v0, thresh_u ));
- v1 = _mm_subs_epu8( v1, _mm_subs_epu8( v1, thresh_u ));
- _mm_storeu_si128( (__m128i*)(dst + j), v0 );
- _mm_storeu_si128( (__m128i*)(dst + j + 16), v1 );
- }
-
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128i v0 = _mm_loadl_epi64( (const __m128i*)(src + j) );
- v0 = _mm_subs_epu8( v0, _mm_subs_epu8( v0, thresh_u ));
- _mm_storel_epi64( (__m128i*)(dst + j), v0 );
- }
- break;
+ }
+ break;
- case THRESH_TOZERO:
- for( j = 0; j <= roi.width - 32; j += 32 )
+ case THRESH_BINARY_INV:
+ for( int i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j <= roi.width - 16; j += 16 )
{
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 16) );
- v0 = _mm_and_si128( v0, _mm_cmpgt_epi8(_mm_xor_si128(v0, _x80), thresh_s ));
- v1 = _mm_and_si128( v1, _mm_cmpgt_epi8(_mm_xor_si128(v1, _x80), thresh_s ));
- _mm_storeu_si128( (__m128i*)(dst + j), v0 );
- _mm_storeu_si128( (__m128i*)(dst + j + 16), v1 );
+ v_uint8x16 v0;
+ v0 = v_load( src + j );
+ v0 = v0 <= thresh_u;
+ v0 = v0 & maxval16;
+ v_store( dst + j, v0 );
}
+ }
+ break;
- for( ; j <= roi.width - 8; j += 8 )
+ case THRESH_TRUNC:
+ for( int i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j <= roi.width - 16; j += 16 )
{
- __m128i v0 = _mm_loadl_epi64( (const __m128i*)(src + j) );
- v0 = _mm_and_si128( v0, _mm_cmpgt_epi8(_mm_xor_si128(v0, _x80), thresh_s ));
- _mm_storel_epi64( (__m128i*)(dst + j), v0 );
+ v_uint8x16 v0;
+ v0 = v_load( src + j );
+ v0 = v0 - ( v0 - thresh_u );
+ v_store( dst + j, v0 );
}
- break;
+ }
+ break;
- case THRESH_TOZERO_INV:
- for( j = 0; j <= roi.width - 32; j += 32 )
+ case THRESH_TOZERO:
+ for( int i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j <= roi.width - 16; j += 16 )
{
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 16) );
- v0 = _mm_andnot_si128( _mm_cmpgt_epi8(_mm_xor_si128(v0, _x80), thresh_s ), v0 );
- v1 = _mm_andnot_si128( _mm_cmpgt_epi8(_mm_xor_si128(v1, _x80), thresh_s ), v1 );
- _mm_storeu_si128( (__m128i*)(dst + j), v0 );
- _mm_storeu_si128( (__m128i*)(dst + j + 16), v1 );
+ v_uint8x16 v0;
+ v0 = v_load( src + j );
+ v0 = ( thresh_u < v0 ) & v0;
+ v_store( dst + j, v0 );
}
+ }
+ break;
- for( ; j <= roi.width - 8; j += 8 )
+ case THRESH_TOZERO_INV:
+ for( int i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j <= roi.width - 16; j += 16 )
{
- __m128i v0 = _mm_loadl_epi64( (const __m128i*)(src + j) );
- v0 = _mm_andnot_si128( _mm_cmpgt_epi8(_mm_xor_si128(v0, _x80), thresh_s ), v0 );
- _mm_storel_epi64( (__m128i*)(dst + j), v0 );
+ v_uint8x16 v0;
+ v0 = v_load( src + j );
+ v0 = ( v0 <= thresh_u ) & v0;
+ v_store( dst + j, v0 );
}
- break;
}
+ break;
}
}
-#elif CV_NEON
- uint8x16_t v_thresh = vdupq_n_u8(thresh), v_maxval = vdupq_n_u8(maxval);
+#endif
- switch( type )
+ int j_scalar = j;
+ if( j_scalar < roi.width )
{
- case THRESH_BINARY:
- for( i = 0; i < roi.height; i++ )
- {
- const uchar* src = _src.ptr() + src_step*i;
- uchar* dst = _dst.ptr() + dst_step*i;
-
- for ( j_scalar = 0; j_scalar <= roi.width - 16; j_scalar += 16)
- vst1q_u8(dst + j_scalar, vandq_u8(vcgtq_u8(vld1q_u8(src + j_scalar), v_thresh), v_maxval));
- }
- break;
-
- case THRESH_BINARY_INV:
- for( i = 0; i < roi.height; i++ )
- {
- const uchar* src = _src.ptr() + src_step*i;
- uchar* dst = _dst.ptr() + dst_step*i;
-
- for ( j_scalar = 0; j_scalar <= roi.width - 16; j_scalar += 16)
- vst1q_u8(dst + j_scalar, vandq_u8(vcleq_u8(vld1q_u8(src + j_scalar), v_thresh), v_maxval));
- }
- break;
-
- case THRESH_TRUNC:
- for( i = 0; i < roi.height; i++ )
- {
- const uchar* src = _src.ptr() + src_step*i;
- uchar* dst = _dst.ptr() + dst_step*i;
-
- for ( j_scalar = 0; j_scalar <= roi.width - 16; j_scalar += 16)
- vst1q_u8(dst + j_scalar, vminq_u8(vld1q_u8(src + j_scalar), v_thresh));
- }
- break;
-
- case THRESH_TOZERO:
- for( i = 0; i < roi.height; i++ )
+ const int thresh_pivot = thresh + 1;
+ uchar tab[256];
+ switch( type )
{
- const uchar* src = _src.ptr() + src_step*i;
- uchar* dst = _dst.ptr() + dst_step*i;
-
- for ( j_scalar = 0; j_scalar <= roi.width - 16; j_scalar += 16)
- {
- uint8x16_t v_src = vld1q_u8(src + j_scalar), v_mask = vcgtq_u8(v_src, v_thresh);
- vst1q_u8(dst + j_scalar, vandq_u8(v_mask, v_src));
+ case THRESH_BINARY:
+ memset(tab, 0, thresh_pivot);
+ if (thresh_pivot < 256) {
+ memset(tab + thresh_pivot, maxval, 256 - thresh_pivot);
}
- }
- break;
-
- case THRESH_TOZERO_INV:
- for( i = 0; i < roi.height; i++ )
- {
- const uchar* src = _src.ptr() + src_step*i;
- uchar* dst = _dst.ptr() + dst_step*i;
-
- for ( j_scalar = 0; j_scalar <= roi.width - 16; j_scalar += 16)
- {
- uint8x16_t v_src = vld1q_u8(src + j_scalar), v_mask = vcleq_u8(v_src, v_thresh);
- vst1q_u8(dst + j_scalar, vandq_u8(v_mask, v_src));
+ break;
+ case THRESH_BINARY_INV:
+ memset(tab, maxval, thresh_pivot);
+ if (thresh_pivot < 256) {
+ memset(tab + thresh_pivot, 0, 256 - thresh_pivot);
+ }
+ break;
+ case THRESH_TRUNC:
+ for( int i = 0; i <= thresh; i++ )
+ tab[i] = (uchar)i;
+ if (thresh_pivot < 256) {
+ memset(tab + thresh_pivot, thresh, 256 - thresh_pivot);
+ }
+ break;
+ case THRESH_TOZERO:
+ memset(tab, 0, thresh_pivot);
+ for( int i = thresh_pivot; i < 256; i++ )
+ tab[i] = (uchar)i;
+ break;
+ case THRESH_TOZERO_INV:
+ for( int i = 0; i <= thresh; i++ )
+ tab[i] = (uchar)i;
+ if (thresh_pivot < 256) {
+ memset(tab + thresh_pivot, 0, 256 - thresh_pivot);
}
+ break;
}
- break;
- default:
- return CV_Error( CV_StsBadArg, "" );
- }
-#endif
- if( j_scalar < roi.width )
- {
- for( i = 0; i < roi.height; i++ )
+ src = _src.ptr();
+ dst = _dst.ptr();
+ for( int i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- const uchar* src = _src.ptr() + src_step*i;
- uchar* dst = _dst.ptr() + dst_step*i;
j = j_scalar;
#if CV_ENABLE_UNROLLED
for( ; j <= roi.width - 4; j += 4 )
size_t src_step = _src.step/sizeof(src[0]);
size_t dst_step = _dst.step/sizeof(dst[0]);
-#if CV_SSE2
- volatile bool useSIMD = checkHardwareSupport(CV_CPU_SSE2);
-#endif
-
if( _src.isContinuous() && _dst.isContinuous() )
{
roi.width *= roi.height;
switch( type )
{
case THRESH_TRUNC:
-#ifndef HAVE_IPP_ICV_ONLY
- if (_src.data == _dst.data && ippiThreshold_GT_16s_C1IR(dst, (int)dst_step*sizeof(dst[0]), sz, thresh) >= 0)
+ if (_src.data == _dst.data && CV_INSTRUMENT_FUN_IPP(ippiThreshold_GT_16s_C1IR, dst, (int)dst_step*sizeof(dst[0]), sz, thresh) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
}
-#endif
- if (ippiThreshold_GT_16s_C1R(src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh) >= 0)
+ if (CV_INSTRUMENT_FUN_IPP(ippiThreshold_GT_16s_C1R, src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
setIppErrorStatus();
break;
case THRESH_TOZERO:
-#ifndef HAVE_IPP_ICV_ONLY
- if (_src.data == _dst.data && ippiThreshold_LTVal_16s_C1IR(dst, (int)dst_step*sizeof(dst[0]), sz, thresh + 1, 0) >= 0)
+ if (_src.data == _dst.data && CV_INSTRUMENT_FUN_IPP(ippiThreshold_LTVal_16s_C1IR, dst, (int)dst_step*sizeof(dst[0]), sz, thresh + 1, 0) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
}
-#endif
- if (ippiThreshold_LTVal_16s_C1R(src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh+1, 0) >= 0)
+ if (CV_INSTRUMENT_FUN_IPP(ippiThreshold_LTVal_16s_C1R, src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh + 1, 0) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
setIppErrorStatus();
break;
case THRESH_TOZERO_INV:
-#ifndef HAVE_IPP_ICV_ONLY
- if (_src.data == _dst.data && ippiThreshold_GTVal_16s_C1IR(dst, (int)dst_step*sizeof(dst[0]), sz, thresh, 0) >= 0)
+ if (_src.data == _dst.data && CV_INSTRUMENT_FUN_IPP(ippiThreshold_GTVal_16s_C1IR, dst, (int)dst_step*sizeof(dst[0]), sz, thresh, 0) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
}
-#endif
- if (ippiThreshold_GTVal_16s_C1R(src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh, 0) >= 0)
+ if (CV_INSTRUMENT_FUN_IPP(ippiThreshold_GTVal_16s_C1R, src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh, 0) >= 0)
{
CV_IMPL_ADD(CV_IMPL_IPP);
return;
}
#endif
- switch( type )
+#if CV_SIMD128
+ bool useSIMD = checkHardwareSupport( CV_CPU_SSE2 ) || checkHardwareSupport( CV_CPU_NEON );
+ if( useSIMD )
{
- case THRESH_BINARY:
- for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ v_int16x8 thresh8 = v_setall_s16( thresh );
+ v_int16x8 maxval8 = v_setall_s16( maxval );
+
+ switch( type )
{
- j = 0;
- #if CV_SSE2
- if( useSIMD )
+ case THRESH_BINARY:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- __m128i thresh8 = _mm_set1_epi16(thresh), maxval8 = _mm_set1_epi16(maxval);
+ j = 0;
for( ; j <= roi.width - 16; j += 16 )
{
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 8) );
- v0 = _mm_cmpgt_epi16( v0, thresh8 );
- v1 = _mm_cmpgt_epi16( v1, thresh8 );
- v0 = _mm_and_si128( v0, maxval8 );
- v1 = _mm_and_si128( v1, maxval8 );
- _mm_storeu_si128((__m128i*)(dst + j), v0 );
- _mm_storeu_si128((__m128i*)(dst + j + 8), v1 );
+ v_int16x8 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 8 );
+ v0 = thresh8 < v0;
+ v1 = thresh8 < v1;
+ v0 = v0 & maxval8;
+ v1 = v1 & maxval8;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 8, v1 );
}
- }
- #elif CV_NEON
- int16x8_t v_thresh = vdupq_n_s16(thresh), v_maxval = vdupq_n_s16(maxval);
- for( ; j <= roi.width - 8; j += 8 )
- {
- uint16x8_t v_mask = vcgtq_s16(vld1q_s16(src + j), v_thresh);
- vst1q_s16(dst + j, vandq_s16(vreinterpretq_s16_u16(v_mask), v_maxval));
+ for( ; j < roi.width; j++ )
+ dst[j] = src[j] > thresh ? maxval : 0;
}
- #endif
-
- for( ; j < roi.width; j++ )
- dst[j] = src[j] > thresh ? maxval : 0;
- }
- break;
+ break;
- case THRESH_BINARY_INV:
- for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
- {
- j = 0;
- #if CV_SSE2
- if( useSIMD )
+ case THRESH_BINARY_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- __m128i thresh8 = _mm_set1_epi16(thresh), maxval8 = _mm_set1_epi16(maxval);
+ j = 0;
for( ; j <= roi.width - 16; j += 16 )
{
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 8) );
- v0 = _mm_cmpgt_epi16( v0, thresh8 );
- v1 = _mm_cmpgt_epi16( v1, thresh8 );
- v0 = _mm_andnot_si128( v0, maxval8 );
- v1 = _mm_andnot_si128( v1, maxval8 );
- _mm_storeu_si128((__m128i*)(dst + j), v0 );
- _mm_storeu_si128((__m128i*)(dst + j + 8), v1 );
+ v_int16x8 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 8 );
+ v0 = v0 <= thresh8;
+ v1 = v1 <= thresh8;
+ v0 = v0 & maxval8;
+ v1 = v1 & maxval8;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 8, v1 );
}
- }
- #elif CV_NEON
- int16x8_t v_thresh = vdupq_n_s16(thresh), v_maxval = vdupq_n_s16(maxval);
- for( ; j <= roi.width - 8; j += 8 )
- {
- uint16x8_t v_mask = vcleq_s16(vld1q_s16(src + j), v_thresh);
- vst1q_s16(dst + j, vandq_s16(vreinterpretq_s16_u16(v_mask), v_maxval));
+ for( ; j < roi.width; j++ )
+ dst[j] = src[j] <= thresh ? maxval : 0;
}
- #endif
-
- for( ; j < roi.width; j++ )
- dst[j] = src[j] <= thresh ? maxval : 0;
- }
- break;
+ break;
- case THRESH_TRUNC:
- for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
- {
- j = 0;
- #if CV_SSE2
- if( useSIMD )
+ case THRESH_TRUNC:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- __m128i thresh8 = _mm_set1_epi16(thresh);
+ j = 0;
for( ; j <= roi.width - 16; j += 16 )
{
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 8) );
- v0 = _mm_min_epi16( v0, thresh8 );
- v1 = _mm_min_epi16( v1, thresh8 );
- _mm_storeu_si128((__m128i*)(dst + j), v0 );
- _mm_storeu_si128((__m128i*)(dst + j + 8), v1 );
+ v_int16x8 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 8 );
+ v0 = v_min( v0, thresh8 );
+ v1 = v_min( v1, thresh8 );
+ v_store( dst + j, v0 );
+ v_store( dst + j + 8, v1 );
}
+
+ for( ; j < roi.width; j++ )
+ dst[j] = std::min( src[j], thresh );
}
- #elif CV_NEON
- int16x8_t v_thresh = vdupq_n_s16(thresh);
+ break;
- for( ; j <= roi.width - 8; j += 8 )
- vst1q_s16(dst + j, vminq_s16(vld1q_s16(src + j), v_thresh));
- #endif
+ case THRESH_TOZERO:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ j = 0;
+ for( ; j <= roi.width - 16; j += 16 )
+ {
+ v_int16x8 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 8 );
+ v0 = ( thresh8 < v0 ) & v0;
+ v1 = ( thresh8 < v1 ) & v1;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 8, v1 );
+ }
- for( ; j < roi.width; j++ )
- dst[j] = std::min(src[j], thresh);
- }
- break;
+ for( ; j < roi.width; j++ )
+ {
+ short v = src[j];
+ dst[j] = v > thresh ? v : 0;
+ }
+ }
+ break;
- case THRESH_TOZERO:
- for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
- {
- j = 0;
- #if CV_SSE2
- if( useSIMD )
+ case THRESH_TOZERO_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- __m128i thresh8 = _mm_set1_epi16(thresh);
+ j = 0;
for( ; j <= roi.width - 16; j += 16 )
{
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 8) );
- v0 = _mm_and_si128(v0, _mm_cmpgt_epi16(v0, thresh8));
- v1 = _mm_and_si128(v1, _mm_cmpgt_epi16(v1, thresh8));
- _mm_storeu_si128((__m128i*)(dst + j), v0 );
- _mm_storeu_si128((__m128i*)(dst + j + 8), v1 );
+ v_int16x8 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 8 );
+ v0 = ( v0 <= thresh8 ) & v0;
+ v1 = ( v1 <= thresh8 ) & v1;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 8, v1 );
+ }
+
+ for( ; j < roi.width; j++ )
+ {
+ short v = src[j];
+ dst[j] = v <= thresh ? v : 0;
}
}
- #elif CV_NEON
- int16x8_t v_thresh = vdupq_n_s16(thresh);
+ break;
+ default:
+ return CV_Error( CV_StsBadArg, "" );
+ }
+ }
+ else
+#endif
+ {
+ switch( type )
+ {
+ case THRESH_BINARY:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j < roi.width; j++ )
+ dst[j] = src[j] > thresh ? maxval : 0;
+ }
+ break;
- for( ; j <= roi.width - 8; j += 8 )
+ case THRESH_BINARY_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- int16x8_t v_src = vld1q_s16(src + j);
- uint16x8_t v_mask = vcgtq_s16(v_src, v_thresh);
- vst1q_s16(dst + j, vandq_s16(vreinterpretq_s16_u16(v_mask), v_src));
+ for( j = 0; j < roi.width; j++ )
+ dst[j] = src[j] <= thresh ? maxval : 0;
}
- #endif
+ break;
- for( ; j < roi.width; j++ )
+ case THRESH_TRUNC:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- short v = src[j];
- dst[j] = v > thresh ? v : 0;
+ for( j = 0; j < roi.width; j++ )
+ dst[j] = std::min( src[j], thresh );
}
- }
- break;
+ break;
- case THRESH_TOZERO_INV:
- for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
- {
- j = 0;
- #if CV_SSE2
- if( useSIMD )
+ case THRESH_TOZERO:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- __m128i thresh8 = _mm_set1_epi16(thresh);
- for( ; j <= roi.width - 16; j += 16 )
+ for( j = 0; j < roi.width; j++ )
{
- __m128i v0, v1;
- v0 = _mm_loadu_si128( (const __m128i*)(src + j) );
- v1 = _mm_loadu_si128( (const __m128i*)(src + j + 8) );
- v0 = _mm_andnot_si128(_mm_cmpgt_epi16(v0, thresh8), v0);
- v1 = _mm_andnot_si128(_mm_cmpgt_epi16(v1, thresh8), v1);
- _mm_storeu_si128((__m128i*)(dst + j), v0 );
- _mm_storeu_si128((__m128i*)(dst + j + 8), v1 );
+ short v = src[j];
+ dst[j] = v > thresh ? v : 0;
}
}
- #elif CV_NEON
- int16x8_t v_thresh = vdupq_n_s16(thresh);
+ break;
- for( ; j <= roi.width - 8; j += 8 )
+ case THRESH_TOZERO_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- int16x8_t v_src = vld1q_s16(src + j);
- uint16x8_t v_mask = vcleq_s16(v_src, v_thresh);
- vst1q_s16(dst + j, vandq_s16(vreinterpretq_s16_u16(v_mask), v_src));
+ for( j = 0; j < roi.width; j++ )
+ {
+ short v = src[j];
+ dst[j] = v <= thresh ? v : 0;
+ }
}
- #endif
- for( ; j < roi.width; j++ )
+ break;
+ default:
+ return CV_Error( CV_StsBadArg, "" );
+ }
+ }
+}
+
+
+static void
+thresh_32f( const Mat& _src, Mat& _dst, float thresh, float maxval, int type )
+{
+ int i, j;
+ Size roi = _src.size();
+ roi.width *= _src.channels();
+ const float* src = _src.ptr<float>();
+ float* dst = _dst.ptr<float>();
+ size_t src_step = _src.step/sizeof(src[0]);
+ size_t dst_step = _dst.step/sizeof(dst[0]);
+
+ if( _src.isContinuous() && _dst.isContinuous() )
+ {
+ roi.width *= roi.height;
+ roi.height = 1;
+ }
+
+#ifdef HAVE_TEGRA_OPTIMIZATION
+ if (tegra::useTegra() && tegra::thresh_32f(_src, _dst, roi.width, roi.height, thresh, maxval, type))
+ return;
+#endif
+
+#if defined(HAVE_IPP)
+ CV_IPP_CHECK()
+ {
+ IppiSize sz = { roi.width, roi.height };
+ switch( type )
+ {
+ case THRESH_TRUNC:
+ if (0 <= CV_INSTRUMENT_FUN_IPP(ippiThreshold_GT_32f_C1R, src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh))
{
- short v = src[j];
- dst[j] = v <= thresh ? v : 0;
+ CV_IMPL_ADD(CV_IMPL_IPP);
+ return;
}
+ setIppErrorStatus();
+ break;
+ case THRESH_TOZERO:
+ if (0 <= CV_INSTRUMENT_FUN_IPP(ippiThreshold_LTVal_32f_C1R, src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh + FLT_EPSILON, 0))
+ {
+ CV_IMPL_ADD(CV_IMPL_IPP);
+ return;
+ }
+ setIppErrorStatus();
+ break;
+ case THRESH_TOZERO_INV:
+ if (0 <= CV_INSTRUMENT_FUN_IPP(ippiThreshold_GTVal_32f_C1R, src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh, 0))
+ {
+ CV_IMPL_ADD(CV_IMPL_IPP);
+ return;
+ }
+ setIppErrorStatus();
+ break;
+ }
+ }
+#endif
+
+#if CV_SIMD128
+ bool useSIMD = checkHardwareSupport( CV_CPU_SSE2 ) || checkHardwareSupport( CV_CPU_NEON );
+ if( useSIMD )
+ {
+ v_float32x4 thresh4 = v_setall_f32( thresh );
+ v_float32x4 maxval4 = v_setall_f32( maxval );
+
+ switch( type )
+ {
+ case THRESH_BINARY:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ j = 0;
+ for( ; j <= roi.width - 8; j += 8 )
+ {
+ v_float32x4 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 4 );
+ v0 = thresh4 < v0;
+ v1 = thresh4 < v1;
+ v0 = v0 & maxval4;
+ v1 = v1 & maxval4;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 4, v1 );
+ }
+
+ for( ; j < roi.width; j++ )
+ dst[j] = src[j] > thresh ? maxval : 0;
+ }
+ break;
+
+ case THRESH_BINARY_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ j = 0;
+ for( ; j <= roi.width - 8; j += 8 )
+ {
+ v_float32x4 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 4 );
+ v0 = v0 <= thresh4;
+ v1 = v1 <= thresh4;
+ v0 = v0 & maxval4;
+ v1 = v1 & maxval4;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 4, v1 );
+ }
+
+ for( ; j < roi.width; j++ )
+ dst[j] = src[j] <= thresh ? maxval : 0;
+ }
+ break;
+
+ case THRESH_TRUNC:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ j = 0;
+ for( ; j <= roi.width - 8; j += 8 )
+ {
+ v_float32x4 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 4 );
+ v0 = v_min( v0, thresh4 );
+ v1 = v_min( v1, thresh4 );
+ v_store( dst + j, v0 );
+ v_store( dst + j + 4, v1 );
+ }
+
+ for( ; j < roi.width; j++ )
+ dst[j] = std::min( src[j], thresh );
+ }
+ break;
+
+ case THRESH_TOZERO:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ j = 0;
+ for( ; j <= roi.width - 8; j += 8 )
+ {
+ v_float32x4 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 4 );
+ v0 = ( thresh4 < v0 ) & v0;
+ v1 = ( thresh4 < v1 ) & v1;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 4, v1 );
+ }
+
+ for( ; j < roi.width; j++ )
+ {
+ float v = src[j];
+ dst[j] = v > thresh ? v : 0;
+ }
+ }
+ break;
+
+ case THRESH_TOZERO_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ j = 0;
+ for( ; j <= roi.width - 8; j += 8 )
+ {
+ v_float32x4 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 4 );
+ v0 = ( v0 <= thresh4 ) & v0;
+ v1 = ( v1 <= thresh4 ) & v1;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 4, v1 );
+ }
+
+ for( ; j < roi.width; j++ )
+ {
+ float v = src[j];
+ dst[j] = v <= thresh ? v : 0;
+ }
+ }
+ break;
+ default:
+ return CV_Error( CV_StsBadArg, "" );
+ }
+ }
+ else
+#endif
+ {
+ switch( type )
+ {
+ case THRESH_BINARY:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j < roi.width; j++ )
+ dst[j] = src[j] > thresh ? maxval : 0;
+ }
+ break;
+
+ case THRESH_BINARY_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j < roi.width; j++ )
+ dst[j] = src[j] <= thresh ? maxval : 0;
+ }
+ break;
+
+ case THRESH_TRUNC:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j < roi.width; j++ )
+ dst[j] = std::min( src[j], thresh );
+ }
+ break;
+
+ case THRESH_TOZERO:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j < roi.width; j++ )
+ {
+ float v = src[j];
+ dst[j] = v > thresh ? v : 0;
+ }
+ }
+ break;
+
+ case THRESH_TOZERO_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
+ {
+ for( j = 0; j < roi.width; j++ )
+ {
+ float v = src[j];
+ dst[j] = v <= thresh ? v : 0;
+ }
+ }
+ break;
+ default:
+ return CV_Error( CV_StsBadArg, "" );
}
- break;
- default:
- return CV_Error( CV_StsBadArg, "" );
}
}
-
static void
-thresh_32f( const Mat& _src, Mat& _dst, float thresh, float maxval, int type )
+thresh_64f(const Mat& _src, Mat& _dst, double thresh, double maxval, int type)
{
int i, j;
Size roi = _src.size();
roi.width *= _src.channels();
- const float* src = _src.ptr<float>();
- float* dst = _dst.ptr<float>();
- size_t src_step = _src.step/sizeof(src[0]);
- size_t dst_step = _dst.step/sizeof(dst[0]);
-
-#if CV_SSE
- volatile bool useSIMD = checkHardwareSupport(CV_CPU_SSE);
-#endif
+ const double* src = _src.ptr<double>();
+ double* dst = _dst.ptr<double>();
+ size_t src_step = _src.step / sizeof(src[0]);
+ size_t dst_step = _dst.step / sizeof(dst[0]);
- if( _src.isContinuous() && _dst.isContinuous() )
+ if (_src.isContinuous() && _dst.isContinuous())
{
roi.width *= roi.height;
roi.height = 1;
}
-#ifdef HAVE_TEGRA_OPTIMIZATION
- if (tegra::useTegra() && tegra::thresh_32f(_src, _dst, roi.width, roi.height, thresh, maxval, type))
- return;
-#endif
-
-#if defined(HAVE_IPP)
- CV_IPP_CHECK()
+#if CV_SIMD128_64F
+ bool useSIMD = checkHardwareSupport( CV_CPU_SSE2 ) || checkHardwareSupport( CV_CPU_NEON );
+ if( useSIMD )
{
- IppiSize sz = { roi.width, roi.height };
+ v_float64x2 thresh2 = v_setall_f64( thresh );
+ v_float64x2 maxval2 = v_setall_f64( maxval );
+
switch( type )
{
- case THRESH_TRUNC:
- if (0 <= ippiThreshold_GT_32f_C1R(src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh))
- {
- CV_IMPL_ADD(CV_IMPL_IPP);
- return;
- }
- setIppErrorStatus();
- break;
- case THRESH_TOZERO:
- if (0 <= ippiThreshold_LTVal_32f_C1R(src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh+FLT_EPSILON, 0))
- {
- CV_IMPL_ADD(CV_IMPL_IPP);
- return;
- }
- setIppErrorStatus();
- break;
- case THRESH_TOZERO_INV:
- if (0 <= ippiThreshold_GTVal_32f_C1R(src, (int)src_step*sizeof(src[0]), dst, (int)dst_step*sizeof(dst[0]), sz, thresh, 0))
- {
- CV_IMPL_ADD(CV_IMPL_IPP);
- return;
- }
- setIppErrorStatus();
- break;
- }
- }
-#endif
-
- switch( type )
- {
case THRESH_BINARY:
for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
j = 0;
-#if CV_SSE
- if( useSIMD )
- {
- __m128 thresh4 = _mm_set1_ps(thresh), maxval4 = _mm_set1_ps(maxval);
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128 v0, v1;
- v0 = _mm_loadu_ps( src + j );
- v1 = _mm_loadu_ps( src + j + 4 );
- v0 = _mm_cmpgt_ps( v0, thresh4 );
- v1 = _mm_cmpgt_ps( v1, thresh4 );
- v0 = _mm_and_ps( v0, maxval4 );
- v1 = _mm_and_ps( v1, maxval4 );
- _mm_storeu_ps( dst + j, v0 );
- _mm_storeu_ps( dst + j + 4, v1 );
- }
- }
-#elif CV_NEON
- float32x4_t v_thresh = vdupq_n_f32(thresh);
- uint32x4_t v_maxval = vreinterpretq_u32_f32(vdupq_n_f32(maxval));
-
for( ; j <= roi.width - 4; j += 4 )
{
- float32x4_t v_src = vld1q_f32(src + j);
- uint32x4_t v_dst = vandq_u32(vcgtq_f32(v_src, v_thresh), v_maxval);
- vst1q_f32(dst + j, vreinterpretq_f32_u32(v_dst));
+ v_float64x2 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 2 );
+ v0 = thresh2 < v0;
+ v1 = thresh2 < v1;
+ v0 = v0 & maxval2;
+ v1 = v1 & maxval2;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 2, v1 );
}
-#endif
for( ; j < roi.width; j++ )
dst[j] = src[j] > thresh ? maxval : 0;
for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
j = 0;
-#if CV_SSE
- if( useSIMD )
- {
- __m128 thresh4 = _mm_set1_ps(thresh), maxval4 = _mm_set1_ps(maxval);
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128 v0, v1;
- v0 = _mm_loadu_ps( src + j );
- v1 = _mm_loadu_ps( src + j + 4 );
- v0 = _mm_cmple_ps( v0, thresh4 );
- v1 = _mm_cmple_ps( v1, thresh4 );
- v0 = _mm_and_ps( v0, maxval4 );
- v1 = _mm_and_ps( v1, maxval4 );
- _mm_storeu_ps( dst + j, v0 );
- _mm_storeu_ps( dst + j + 4, v1 );
- }
- }
-#elif CV_NEON
- float32x4_t v_thresh = vdupq_n_f32(thresh);
- uint32x4_t v_maxval = vreinterpretq_u32_f32(vdupq_n_f32(maxval));
-
for( ; j <= roi.width - 4; j += 4 )
{
- float32x4_t v_src = vld1q_f32(src + j);
- uint32x4_t v_dst = vandq_u32(vcleq_f32(v_src, v_thresh), v_maxval);
- vst1q_f32(dst + j, vreinterpretq_f32_u32(v_dst));
+ v_float64x2 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 2 );
+ v0 = v0 <= thresh2;
+ v1 = v1 <= thresh2;
+ v0 = v0 & maxval2;
+ v1 = v1 & maxval2;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 2, v1 );
}
-#endif
for( ; j < roi.width; j++ )
dst[j] = src[j] <= thresh ? maxval : 0;
for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
j = 0;
-#if CV_SSE
- if( useSIMD )
+ for( ; j <= roi.width - 4; j += 4 )
{
- __m128 thresh4 = _mm_set1_ps(thresh);
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128 v0, v1;
- v0 = _mm_loadu_ps( src + j );
- v1 = _mm_loadu_ps( src + j + 4 );
- v0 = _mm_min_ps( v0, thresh4 );
- v1 = _mm_min_ps( v1, thresh4 );
- _mm_storeu_ps( dst + j, v0 );
- _mm_storeu_ps( dst + j + 4, v1 );
- }
+ v_float64x2 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 2 );
+ v0 = v_min( v0, thresh2 );
+ v1 = v_min( v1, thresh2 );
+ v_store( dst + j, v0 );
+ v_store( dst + j + 2, v1 );
}
-#elif CV_NEON
- float32x4_t v_thresh = vdupq_n_f32(thresh);
-
- for( ; j <= roi.width - 4; j += 4 )
- vst1q_f32(dst + j, vminq_f32(vld1q_f32(src + j), v_thresh));
-#endif
for( ; j < roi.width; j++ )
- dst[j] = std::min(src[j], thresh);
+ dst[j] = std::min( src[j], thresh );
}
break;
for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
j = 0;
-#if CV_SSE
- if( useSIMD )
- {
- __m128 thresh4 = _mm_set1_ps(thresh);
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128 v0, v1;
- v0 = _mm_loadu_ps( src + j );
- v1 = _mm_loadu_ps( src + j + 4 );
- v0 = _mm_and_ps(v0, _mm_cmpgt_ps(v0, thresh4));
- v1 = _mm_and_ps(v1, _mm_cmpgt_ps(v1, thresh4));
- _mm_storeu_ps( dst + j, v0 );
- _mm_storeu_ps( dst + j + 4, v1 );
- }
- }
-#elif CV_NEON
- float32x4_t v_thresh = vdupq_n_f32(thresh);
-
for( ; j <= roi.width - 4; j += 4 )
{
- float32x4_t v_src = vld1q_f32(src + j);
- uint32x4_t v_dst = vandq_u32(vcgtq_f32(v_src, v_thresh),
- vreinterpretq_u32_f32(v_src));
- vst1q_f32(dst + j, vreinterpretq_f32_u32(v_dst));
+ v_float64x2 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 2 );
+ v0 = ( thresh2 < v0 ) & v0;
+ v1 = ( thresh2 < v1 ) & v1;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 2, v1 );
}
-#endif
for( ; j < roi.width; j++ )
{
- float v = src[j];
+ double v = src[j];
dst[j] = v > thresh ? v : 0;
}
}
for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
j = 0;
-#if CV_SSE
- if( useSIMD )
- {
- __m128 thresh4 = _mm_set1_ps(thresh);
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128 v0, v1;
- v0 = _mm_loadu_ps( src + j );
- v1 = _mm_loadu_ps( src + j + 4 );
- v0 = _mm_and_ps(v0, _mm_cmple_ps(v0, thresh4));
- v1 = _mm_and_ps(v1, _mm_cmple_ps(v1, thresh4));
- _mm_storeu_ps( dst + j, v0 );
- _mm_storeu_ps( dst + j + 4, v1 );
- }
- }
-#elif CV_NEON
- float32x4_t v_thresh = vdupq_n_f32(thresh);
-
for( ; j <= roi.width - 4; j += 4 )
{
- float32x4_t v_src = vld1q_f32(src + j);
- uint32x4_t v_dst = vandq_u32(vcleq_f32(v_src, v_thresh),
- vreinterpretq_u32_f32(v_src));
- vst1q_f32(dst + j, vreinterpretq_f32_u32(v_dst));
+ v_float64x2 v0, v1;
+ v0 = v_load( src + j );
+ v1 = v_load( src + j + 2 );
+ v0 = ( v0 <= thresh2 ) & v0;
+ v1 = ( v1 <= thresh2 ) & v1;
+ v_store( dst + j, v0 );
+ v_store( dst + j + 2, v1 );
}
-#endif
+
for( ; j < roi.width; j++ )
{
- float v = src[j];
+ double v = src[j];
dst[j] = v <= thresh ? v : 0;
}
}
break;
default:
- return CV_Error( CV_StsBadArg, "" );
+ return CV_Error(CV_StsBadArg, "");
+ }
}
-}
-
-static void
-thresh_64f(const Mat& _src, Mat& _dst, double thresh, double maxval, int type)
-{
- int i, j;
- Size roi = _src.size();
- roi.width *= _src.channels();
- const double* src = _src.ptr<double>();
- double* dst = _dst.ptr<double>();
- size_t src_step = _src.step / sizeof(src[0]);
- size_t dst_step = _dst.step / sizeof(dst[0]);
-
-#if CV_SSE2
- volatile bool useSIMD = checkHardwareSupport(CV_CPU_SSE2);
+ else
#endif
-
- if (_src.isContinuous() && _dst.isContinuous())
- {
- roi.width *= roi.height;
- roi.height = 1;
- }
-
- switch (type)
{
- case THRESH_BINARY:
- for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step)
- {
- j = 0;
-#if CV_SSE2
- if( useSIMD )
- {
- __m128d thresh2 = _mm_set1_pd(thresh), maxval2 = _mm_set1_pd(maxval);
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128d v0, v1, v2, v3;
- v0 = _mm_loadu_pd( src + j );
- v1 = _mm_loadu_pd( src + j + 2 );
- v2 = _mm_loadu_pd( src + j + 4 );
- v3 = _mm_loadu_pd( src + j + 6 );
- v0 = _mm_cmpgt_pd( v0, thresh2 );
- v1 = _mm_cmpgt_pd( v1, thresh2 );
- v2 = _mm_cmpgt_pd( v2, thresh2 );
- v3 = _mm_cmpgt_pd( v3, thresh2 );
- v0 = _mm_and_pd( v0, maxval2 );
- v1 = _mm_and_pd( v1, maxval2 );
- v2 = _mm_and_pd( v2, maxval2 );
- v3 = _mm_and_pd( v3, maxval2 );
- _mm_storeu_pd( dst + j, v0 );
- _mm_storeu_pd( dst + j + 2, v1 );
- _mm_storeu_pd( dst + j + 4, v2 );
- _mm_storeu_pd( dst + j + 6, v3 );
- }
- }
-#elif CV_NEON && defined(__aarch64__)
- float64x2_t v_thresh = vdupq_n_f64(thresh);
- uint64x2_t v_maxval = vreinterpretq_u64_f64(vdupq_n_f64(maxval));
-
- for( ; j <= roi.width - 4; j += 4 )
- {
- float64x2_t v_src0 = vld1q_f64(src + j);
- float64x2_t v_src1 = vld1q_f64(src + j + 2);
- uint64x2_t v_dst0 = vandq_u64(vcgtq_f64(v_src0, v_thresh), v_maxval);
- uint64x2_t v_dst1 = vandq_u64(vcgtq_f64(v_src1, v_thresh), v_maxval);
- vst1q_f64(dst + j, vreinterpretq_f64_u64(v_dst0));
- vst1q_f64(dst + j + 2, vreinterpretq_f64_u64(v_dst1));
- }
-#endif
-
- for (; j < roi.width; j++)
- dst[j] = src[j] > thresh ? maxval : 0;
- }
- break;
-
- case THRESH_BINARY_INV:
- for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step)
+ switch( type )
{
- j = 0;
-
-#if CV_SSE2
- if( useSIMD )
- {
- __m128d thresh2 = _mm_set1_pd(thresh), maxval2 = _mm_set1_pd(maxval);
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128d v0, v1, v2, v3;
- v0 = _mm_loadu_pd( src + j );
- v1 = _mm_loadu_pd( src + j + 2 );
- v2 = _mm_loadu_pd( src + j + 4 );
- v3 = _mm_loadu_pd( src + j + 6 );
- v0 = _mm_cmple_pd( v0, thresh2 );
- v1 = _mm_cmple_pd( v1, thresh2 );
- v2 = _mm_cmple_pd( v2, thresh2 );
- v3 = _mm_cmple_pd( v3, thresh2 );
- v0 = _mm_and_pd( v0, maxval2 );
- v1 = _mm_and_pd( v1, maxval2 );
- v2 = _mm_and_pd( v2, maxval2 );
- v3 = _mm_and_pd( v3, maxval2 );
- _mm_storeu_pd( dst + j, v0 );
- _mm_storeu_pd( dst + j + 2, v1 );
- _mm_storeu_pd( dst + j + 4, v2 );
- _mm_storeu_pd( dst + j + 6, v3 );
- }
- }
-#elif CV_NEON && defined(__aarch64__)
- float64x2_t v_thresh = vdupq_n_f64(thresh);
- uint64x2_t v_maxval = vreinterpretq_u64_f64(vdupq_n_f64(maxval));
-
- for( ; j <= roi.width - 4; j += 4 )
+ case THRESH_BINARY:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- float64x2_t v_src0 = vld1q_f64(src + j);
- float64x2_t v_src1 = vld1q_f64(src + j + 2);
- uint64x2_t v_dst0 = vandq_u64(vcleq_f64(v_src0, v_thresh), v_maxval);
- uint64x2_t v_dst1 = vandq_u64(vcleq_f64(v_src1, v_thresh), v_maxval);
- vst1q_f64(dst + j, vreinterpretq_f64_u64(v_dst0));
- vst1q_f64(dst + j + 2, vreinterpretq_f64_u64(v_dst1));
+ j = 0;
+ for( ; j < roi.width; j++ )
+ dst[j] = src[j] > thresh ? maxval : 0;
}
-#endif
- for (; j < roi.width; j++)
- dst[j] = src[j] <= thresh ? maxval : 0;
- }
- break;
-
- case THRESH_TRUNC:
- for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step)
- {
- j = 0;
+ break;
-#if CV_SSE2
- if( useSIMD )
+ case THRESH_BINARY_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- __m128d thresh2 = _mm_set1_pd(thresh);
- for( ; j <= roi.width - 8; j += 8 )
- {
- __m128d v0, v1, v2, v3;
- v0 = _mm_loadu_pd( src + j );
- v1 = _mm_loadu_pd( src + j + 2 );
- v2 = _mm_loadu_pd( src + j + 4 );
- v3 = _mm_loadu_pd( src + j + 6 );
- v0 = _mm_min_pd( v0, thresh2 );
- v1 = _mm_min_pd( v1, thresh2 );
- v2 = _mm_min_pd( v2, thresh2 );
- v3 = _mm_min_pd( v3, thresh2 );
- _mm_storeu_pd( dst + j, v0 );
- _mm_storeu_pd( dst + j + 2, v1 );
- _mm_storeu_pd( dst + j + 4, v2 );
- _mm_storeu_pd( dst + j + 6, v3 );
- }
+ j = 0;
+ for( ; j < roi.width; j++ )
+ dst[j] = src[j] <= thresh ? maxval : 0;
}
-#elif CV_NEON && defined(__aarch64__)
- float64x2_t v_thresh = vdupq_n_f64(thresh);
+ break;
- for( ; j <= roi.width - 4; j += 4 )
+ case THRESH_TRUNC:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- float64x2_t v_src0 = vld1q_f64(src + j);
- float64x2_t v_src1 = vld1q_f64(src + j + 2);
- float64x2_t v_dst0 = vminq_f64(v_src0, v_thresh);
- float64x2_t v_dst1 = vminq_f64(v_src1, v_thresh);
- vst1q_f64(dst + j, v_dst0);
- vst1q_f64(dst + j + 2, v_dst1);
+ j = 0;
+ for( ; j < roi.width; j++ )
+ dst[j] = std::min( src[j], thresh );
}
-#endif
- for (; j < roi.width; j++)
- dst[j] = std::min(src[j], thresh);
- }
- break;
-
- case THRESH_TOZERO:
- for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step)
- {
- j = 0;
+ break;
-#if CV_SSE2
- if( useSIMD )
+ case THRESH_TOZERO:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- __m128d thresh2 = _mm_set1_pd(thresh);
- for( ; j <= roi.width - 8; j += 8 )
+ j = 0;
+ for( ; j < roi.width; j++ )
{
- __m128d v0, v1, v2, v3;
- v0 = _mm_loadu_pd( src + j );
- v1 = _mm_loadu_pd( src + j + 2 );
- v2 = _mm_loadu_pd( src + j + 4 );
- v3 = _mm_loadu_pd( src + j + 6 );
- v0 = _mm_and_pd( v0, _mm_cmpgt_pd(v0, thresh2));
- v1 = _mm_and_pd( v1, _mm_cmpgt_pd(v1, thresh2));
- v2 = _mm_and_pd( v2, _mm_cmpgt_pd(v2, thresh2));
- v3 = _mm_and_pd( v3, _mm_cmpgt_pd(v3, thresh2));
- _mm_storeu_pd( dst + j, v0 );
- _mm_storeu_pd( dst + j + 2, v1 );
- _mm_storeu_pd( dst + j + 4, v2 );
- _mm_storeu_pd( dst + j + 6, v3 );
+ double v = src[j];
+ dst[j] = v > thresh ? v : 0;
}
}
-#elif CV_NEON && defined(__aarch64__)
- float64x2_t v_thresh = vdupq_n_f64(thresh);
-
- for( ; j <= roi.width - 4; j += 4 )
- {
- float64x2_t v_src0 = vld1q_f64(src + j);
- float64x2_t v_src1 = vld1q_f64(src + j + 2);
- uint64x2_t v_dst0 = vandq_u64(vcgtq_f64(v_src0, v_thresh),
- vreinterpretq_u64_f64(v_src0));
- uint64x2_t v_dst1 = vandq_u64(vcgtq_f64(v_src1, v_thresh),
- vreinterpretq_u64_f64(v_src1));
- vst1q_f64(dst + j, vreinterpretq_f64_u64(v_dst0));
- vst1q_f64(dst + j + 2, vreinterpretq_f64_u64(v_dst1));
- }
-#endif
- for (; j < roi.width; j++)
- {
- double v = src[j];
- dst[j] = v > thresh ? v : 0;
- }
- }
- break;
-
- case THRESH_TOZERO_INV:
- for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step)
- {
- j = 0;
+ break;
-#if CV_SSE2
- if( useSIMD )
+ case THRESH_TOZERO_INV:
+ for( i = 0; i < roi.height; i++, src += src_step, dst += dst_step )
{
- __m128d thresh2 = _mm_set1_pd(thresh);
- for( ; j <= roi.width - 8; j += 8 )
+ j = 0;
+ for( ; j < roi.width; j++ )
{
- __m128d v0, v1, v2, v3;
- v0 = _mm_loadu_pd( src + j );
- v1 = _mm_loadu_pd( src + j + 2 );
- v2 = _mm_loadu_pd( src + j + 4 );
- v3 = _mm_loadu_pd( src + j + 6 );
- v0 = _mm_and_pd( v0, _mm_cmple_pd(v0, thresh2));
- v1 = _mm_and_pd( v1, _mm_cmple_pd(v1, thresh2));
- v2 = _mm_and_pd( v2, _mm_cmple_pd(v2, thresh2));
- v3 = _mm_and_pd( v3, _mm_cmple_pd(v3, thresh2));
- _mm_storeu_pd( dst + j, v0 );
- _mm_storeu_pd( dst + j + 2, v1 );
- _mm_storeu_pd( dst + j + 4, v2 );
- _mm_storeu_pd( dst + j + 6, v3 );
+ double v = src[j];
+ dst[j] = v <= thresh ? v : 0;
}
}
-#elif CV_NEON && defined(__aarch64__)
- float64x2_t v_thresh = vdupq_n_f64(thresh);
-
- for( ; j <= roi.width - 4; j += 4 )
- {
- float64x2_t v_src0 = vld1q_f64(src + j);
- float64x2_t v_src1 = vld1q_f64(src + j + 2);
- uint64x2_t v_dst0 = vandq_u64(vcleq_f64(v_src0, v_thresh),
- vreinterpretq_u64_f64(v_src0));
- uint64x2_t v_dst1 = vandq_u64(vcleq_f64(v_src1, v_thresh),
- vreinterpretq_u64_f64(v_src1));
- vst1q_f64(dst + j, vreinterpretq_f64_u64(v_dst0));
- vst1q_f64(dst + j + 2, vreinterpretq_f64_u64(v_dst1));
- }
-#endif
- for (; j < roi.width; j++)
- {
- double v = src[j];
- dst[j] = v <= thresh ? v : 0;
- }
+ break;
+ default:
+ return CV_Error(CV_StsBadArg, "");
}
- break;
- default:
- return CV_Error(CV_StsBadArg, "");
}
}
#ifdef HAVE_IPP
static bool ipp_getThreshVal_Otsu_8u( const unsigned char* _src, int step, Size size, unsigned char &thresh)
{
-#if IPP_VERSION_X100 >= 810 && !HAVE_ICV
+ CV_INSTRUMENT_REGION_IPP()
+
+#if IPP_VERSION_X100 >= 810
int ippStatus = -1;
IppiSize srcSize = { size.width, size.height };
CV_SUPPRESS_DEPRECATED_START
- ippStatus = ippiComputeThreshold_Otsu_8u_C1R(_src, step, srcSize, &thresh);
+ ippStatus = CV_INSTRUMENT_FUN_IPP(ippiComputeThreshold_Otsu_8u_C1R, _src, step, srcSize, &thresh);
CV_SUPPRESS_DEPRECATED_END
if(ippStatus >= 0)
#ifdef HAVE_IPP
unsigned char thresh;
- CV_IPP_RUN(IPP_VERSION_X100 >= 810 && !HAVE_ICV, ipp_getThreshVal_Otsu_8u(_src.ptr(), step, size, thresh), thresh);
+ CV_IPP_RUN(IPP_VERSION_X100 >= 810, ipp_getThreshVal_Otsu_8u(_src.ptr(), step, size, thresh), thresh);
#endif
const int N = 256;
#endif
+
+#ifdef HAVE_OPENVX
+#define IMPL_OPENVX_TOZERO 1
+static bool openvx_threshold(Mat src, Mat dst, int thresh, int maxval, int type)
+{
+ Mat a = src;
+
+ int trueVal, falseVal;
+ switch (type)
+ {
+ case THRESH_BINARY:
+#ifndef VX_VERSION_1_1
+ if (maxval != 255)
+ return false;
+#endif
+ trueVal = maxval;
+ falseVal = 0;
+ break;
+ case THRESH_TOZERO:
+#if IMPL_OPENVX_TOZERO
+ trueVal = 255;
+ falseVal = 0;
+ if (dst.data == src.data)
+ {
+ a = Mat(src.size(), src.type());
+ src.copyTo(a);
+ }
+ break;
+#endif
+ case THRESH_BINARY_INV:
+#ifdef VX_VERSION_1_1
+ trueVal = 0;
+ falseVal = maxval;
+ break;
+#endif
+ case THRESH_TOZERO_INV:
+#ifdef VX_VERSION_1_1
+#if IMPL_OPENVX_TOZERO
+ trueVal = 0;
+ falseVal = 255;
+ if (dst.data == src.data)
+ {
+ a = Mat(src.size(), src.type());
+ src.copyTo(a);
+ }
+ break;
+#endif
+#endif
+ case THRESH_TRUNC:
+ default:
+ return false;
+ }
+
+ try
+ {
+ ivx::Context ctx = ivx::Context::create();
+
+ ivx::Threshold thh = ivx::Threshold::createBinary(ctx, VX_TYPE_UINT8, thresh);
+ thh.setValueTrue(trueVal);
+ thh.setValueFalse(falseVal);
+
+ ivx::Image
+ ia = ivx::Image::createFromHandle(ctx, VX_DF_IMAGE_U8,
+ ivx::Image::createAddressing(a.cols*a.channels(), a.rows, 1, (vx_int32)(a.step)), src.data),
+ ib = ivx::Image::createFromHandle(ctx, VX_DF_IMAGE_U8,
+ ivx::Image::createAddressing(dst.cols*dst.channels(), dst.rows, 1, (vx_int32)(dst.step)), dst.data);
+
+ ivx::IVX_CHECK_STATUS(vxuThreshold(ctx, ia, thh, ib));
+#if IMPL_OPENVX_TOZERO
+ if (type == THRESH_TOZERO || type == THRESH_TOZERO_INV)
+ {
+ ivx::Image
+ ic = ivx::Image::createFromHandle(ctx, VX_DF_IMAGE_U8,
+ ivx::Image::createAddressing(dst.cols*dst.channels(), dst.rows, 1, (vx_int32)(dst.step)), dst.data);
+ ivx::IVX_CHECK_STATUS(vxuAnd(ctx, ib, ia, ic));
+ }
+#endif
+ }
+ catch (ivx::RuntimeError & e)
+ {
+ CV_Error(CV_StsInternal, e.what());
+ return false;
+ }
+ catch (ivx::WrapperError & e)
+ {
+ CV_Error(CV_StsInternal, e.what());
+ return false;
+ }
+
+ return true;
+}
+#endif
+
}
double cv::threshold( InputArray _src, OutputArray _dst, double thresh, double maxval, int type )
{
+ CV_INSTRUMENT_REGION()
+
CV_OCL_RUN_(_src.dims() <= 2 && _dst.isUMat(),
ocl_threshold(_src, _dst, thresh, maxval, type), thresh)
src.copyTo(dst);
return thresh;
}
+
+#ifdef HAVE_OPENVX
+ if (openvx_threshold(src, dst, ithresh, imaxval, type))
+ return thresh;
+#endif
+
thresh = ithresh;
maxval = imaxval;
}
void cv::adaptiveThreshold( InputArray _src, OutputArray _dst, double maxValue,
int method, int type, int blockSize, double delta )
{
+ CV_INSTRUMENT_REGION()
+
Mat src = _src.getMat();
CV_Assert( src.type() == CV_8UC1 );
CV_Assert( blockSize % 2 == 1 && blockSize > 1 );
projects / platform / upstream / opencv.git / blobdiff