Merge pull request #1380 from woodychow:simplify_sift_avx2

alalek · alalek · commit 666ec5be838d · 2017-09-26T12:47:06.000Z
diff --git a/modules/xfeatures2d/src/sift.cpp b/modules/xfeatures2d/src/sift.cpp
@@ -378,7 +378,7 @@ static float calcOrientationHist( const Mat& img, Point pt, int radius,
         float CV_DECL_ALIGNED(32) w_mul_mag_buf[8];
         for ( ; k <= len - 8; k+=8 )
         {
-            __m256i __bin = _mm256_cvtps_epi32(_mm256_round_ps(_mm256_mul_ps(__nd360, _mm256_loadu_ps(&Ori[k])), _MM_FROUND_TO_NEAREST_INT |_MM_FROUND_NO_EXC));
+            __m256i __bin = _mm256_cvtps_epi32(_mm256_mul_ps(__nd360, _mm256_loadu_ps(&Ori[k])));
 
             __bin = _mm256_sub_epi32(__bin, _mm256_andnot_si256(_mm256_cmpgt_epi32(__n, __bin), __n));
             __bin = _mm256_add_epi32(__bin, _mm256_and_si256(__n, _mm256_cmpgt_epi32(_mm256_setzero_si256(), __bin)));

Original file line number	Diff line number	Diff line change
`@@ -378,7 +378,7 @@ static float calcOrientationHist( const Mat& img, Point pt, int radius,`
`378`	`378`	`float CV_DECL_ALIGNED(32) w_mul_mag_buf[8];`
`379`	`379`	`for ( ; k <= len - 8; k+=8 )`
`380`	`380`	`{`
`381`		`- __m256i __bin = _mm256_cvtps_epi32(_mm256_round_ps(_mm256_mul_ps(__nd360, _mm256_loadu_ps(&Ori[k])), _MM_FROUND_TO_NEAREST_INT \|_MM_FROUND_NO_EXC));`
	`381`	`+ __m256i __bin = _mm256_cvtps_epi32(_mm256_mul_ps(__nd360, _mm256_loadu_ps(&Ori[k])));`
`382`	`382`
`383`	`383`	`__bin = _mm256_sub_epi32(__bin, _mm256_andnot_si256(_mm256_cmpgt_epi32(__n, __bin), __n));`
`384`	`384`	`__bin = _mm256_add_epi32(__bin, _mm256_and_si256(__n, _mm256_cmpgt_epi32(_mm256_setzero_si256(), __bin)));`