split FAST in order to reuse it in BRISK

This commit is contained in:
Vincent Rabaud 2012-08-23 14:33:11 +02:00 committed by Vadim Pisarevsky
parent a6e2acbfee
commit 228070a74c
3 changed files with 439 additions and 327 deletions

View File

@ -42,335 +42,11 @@ The references are:
*/
#include "precomp.hpp"
#include "fast_score.hpp"
namespace cv
{
static void makeOffsets(int pixel[], int row_stride, int patternSize)
{
switch(patternSize) {
case 16:
pixel[0] = 0 + row_stride * 3;
pixel[1] = 1 + row_stride * 3;
pixel[2] = 2 + row_stride * 2;
pixel[3] = 3 + row_stride * 1;
pixel[4] = 3 + row_stride * 0;
pixel[5] = 3 + row_stride * -1;
pixel[6] = 2 + row_stride * -2;
pixel[7] = 1 + row_stride * -3;
pixel[8] = 0 + row_stride * -3;
pixel[9] = -1 + row_stride * -3;
pixel[10] = -2 + row_stride * -2;
pixel[11] = -3 + row_stride * -1;
pixel[12] = -3 + row_stride * 0;
pixel[13] = -3 + row_stride * 1;
pixel[14] = -2 + row_stride * 2;
pixel[15] = -1 + row_stride * 3;
break;
case 12:
pixel[0] = 0 + row_stride * 2;
pixel[1] = 1 + row_stride * 2;
pixel[2] = 2 + row_stride * 1;
pixel[3] = 2 + row_stride * 0;
pixel[4] = 2 + row_stride * -1;
pixel[5] = 1 + row_stride * -2;
pixel[6] = 0 + row_stride * -2;
pixel[7] = -1 + row_stride * -2;
pixel[8] = -2 + row_stride * -1;
pixel[9] = -2 + row_stride * 0;
pixel[10] = -2 + row_stride * 1;
pixel[11] = -1 + row_stride * 2;
break;
case 8:
pixel[0] = 0 + row_stride * 1;
pixel[1] = 1 + row_stride * 1;
pixel[2] = 1 + row_stride * 0;
pixel[3] = 1 + row_stride * -1;
pixel[4] = 0 + row_stride * -1;
pixel[5] = -1 + row_stride * -1;
pixel[6] = 0 + row_stride * 0;
pixel[7] = 1 + row_stride * 1;
break;
}
}
/*static void testCorner(const uchar* ptr, const int pixel[], int K, int N, int threshold) {
// check that with the computed "threshold" the pixel is still a corner
// and that with the increased-by-1 "threshold" the pixel is not a corner anymore
for( int delta = 0; delta <= 1; delta++ )
{
int v0 = std::min(ptr[0] + threshold + delta, 255);
int v1 = std::max(ptr[0] - threshold - delta, 0);
int c0 = 0, c1 = 0;
for( int k = 0; k < N; k++ )
{
int x = ptr[pixel[k]];
if(x > v0)
{
if( ++c0 > K )
break;
c1 = 0;
}
else if( x < v1 )
{
if( ++c1 > K )
break;
c0 = 0;
}
else
{
c0 = c1 = 0;
}
}
CV_Assert( (delta == 0 && std::max(c0, c1) > K) ||
(delta == 1 && std::max(c0, c1) <= K) );
}
}*/
template<int patternSize>
int cornerScore(const uchar* ptr, const int pixel[], int threshold);
template<>
int cornerScore<16>(const uchar* ptr, const int pixel[], int threshold)
{
const int K = 8, N = 16 + K + 1;
int k, v = ptr[0];
short d[N];
for( k = 0; k < N; k++ )
d[k] = (short)(v - ptr[pixel[k]]);
#if CV_SSE2
__m128i q0 = _mm_set1_epi16(-1000), q1 = _mm_set1_epi16(1000);
for( k = 0; k < 16; k += 8 )
{
__m128i v0 = _mm_loadu_si128((__m128i*)(d+k+1));
__m128i v1 = _mm_loadu_si128((__m128i*)(d+k+2));
__m128i a = _mm_min_epi16(v0, v1);
__m128i b = _mm_max_epi16(v0, v1);
v0 = _mm_loadu_si128((__m128i*)(d+k+3));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+4));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+5));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+6));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+7));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+8));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
v0 = _mm_loadu_si128((__m128i*)(d+k+9));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
}
q0 = _mm_max_epi16(q0, _mm_sub_epi16(_mm_setzero_si128(), q1));
q0 = _mm_max_epi16(q0, _mm_unpackhi_epi64(q0, q0));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 4));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 2));
threshold = (short)_mm_cvtsi128_si32(q0) - 1;
#else
int a0 = threshold;
for( k = 0; k < 16; k += 2 )
{
int a = std::min((int)d[k+1], (int)d[k+2]);
a = std::min(a, (int)d[k+3]);
if( a <= a0 )
continue;
a = std::min(a, (int)d[k+4]);
a = std::min(a, (int)d[k+5]);
a = std::min(a, (int)d[k+6]);
a = std::min(a, (int)d[k+7]);
a = std::min(a, (int)d[k+8]);
a0 = std::max(a0, std::min(a, (int)d[k]));
a0 = std::max(a0, std::min(a, (int)d[k+9]));
}
int b0 = -a0;
for( k = 0; k < 16; k += 2 )
{
int b = std::max((int)d[k+1], (int)d[k+2]);
b = std::max(b, (int)d[k+3]);
b = std::max(b, (int)d[k+4]);
b = std::max(b, (int)d[k+5]);
if( b >= b0 )
continue;
b = std::max(b, (int)d[k+6]);
b = std::max(b, (int)d[k+7]);
b = std::max(b, (int)d[k+8]);
b0 = std::min(b0, std::max(b, (int)d[k]));
b0 = std::min(b0, std::max(b, (int)d[k+9]));
}
threshold = -b0-1;
#endif
#if 0
testCorner(ptr, pixel, K, N, threshold);
#endif
return threshold;
}
template<>
int cornerScore<12>(const uchar* ptr, const int pixel[], int threshold)
{
const int K = 6, N = 12 + K + 1;
int k, v = ptr[0];
short d[N];
for( k = 0; k < N; k++ )
d[k] = (short)(v - ptr[pixel[k]]);
#if CV_SSE2
__m128i q0 = _mm_set1_epi16(-1000), q1 = _mm_set1_epi16(1000);
for( k = 0; k < 16; k += 8 )
{
__m128i v0 = _mm_loadu_si128((__m128i*)(d+k+1));
__m128i v1 = _mm_loadu_si128((__m128i*)(d+k+2));
__m128i a = _mm_min_epi16(v0, v1);
__m128i b = _mm_max_epi16(v0, v1);
v0 = _mm_loadu_si128((__m128i*)(d+k+3));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+4));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+5));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+6));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
v0 = _mm_loadu_si128((__m128i*)(d+k+7));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
}
q0 = _mm_max_epi16(q0, _mm_sub_epi16(_mm_setzero_si128(), q1));
q0 = _mm_max_epi16(q0, _mm_unpackhi_epi64(q0, q0));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 4));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 2));
threshold = (short)_mm_cvtsi128_si32(q0) - 1;
#else
int a0 = threshold;
for( k = 0; k < 12; k += 2 )
{
int a = std::min((int)d[k+1], (int)d[k+2]);
if( a <= a0 )
continue;
a = std::min(a, (int)d[k+3]);
a = std::min(a, (int)d[k+4]);
a = std::min(a, (int)d[k+5]);
a = std::min(a, (int)d[k+6]);
a0 = std::max(a0, std::min(a, (int)d[k]));
a0 = std::max(a0, std::min(a, (int)d[k+7]));
}
int b0 = -a0;
for( k = 0; k < 12; k += 2 )
{
int b = std::max((int)d[k+1], (int)d[k+2]);
b = std::max(b, (int)d[k+3]);
b = std::max(b, (int)d[k+4]);
if( b >= b0 )
continue;
b = std::max(b, (int)d[k+5]);
b = std::max(b, (int)d[k+6]);
b0 = std::min(b0, std::max(b, (int)d[k]));
b0 = std::min(b0, std::max(b, (int)d[k+7]));
}
threshold = -b0-1;
#endif
#if 0
testCorner(ptr, pixel, K, N, threshold);
#endif
return threshold;
}
template<>
int cornerScore<8>(const uchar* ptr, const int pixel[], int threshold)
{
const int K = 4, N = 8 + K + 1;
int k, v = ptr[0];
short d[N];
for( k = 0; k < N; k++ )
d[k] = (short)(v - ptr[pixel[k]]);
#if CV_SSE2
__m128i q0 = _mm_set1_epi16(-1000), q1 = _mm_set1_epi16(1000);
for( k = 0; k < 16; k += 8 )
{
__m128i v0 = _mm_loadu_si128((__m128i*)(d+k+1));
__m128i v1 = _mm_loadu_si128((__m128i*)(d+k+2));
__m128i a = _mm_min_epi16(v0, v1);
__m128i b = _mm_max_epi16(v0, v1);
v0 = _mm_loadu_si128((__m128i*)(d+k+3));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+4));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
v0 = _mm_loadu_si128((__m128i*)(d+k+5));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
}
q0 = _mm_max_epi16(q0, _mm_sub_epi16(_mm_setzero_si128(), q1));
q0 = _mm_max_epi16(q0, _mm_unpackhi_epi64(q0, q0));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 4));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 2));
threshold = (short)_mm_cvtsi128_si32(q0) - 1;
#else
int a0 = threshold;
for( k = 0; k < 8; k += 2 )
{
int a = std::min((int)d[k+1], (int)d[k+2]);
if( a <= a0 )
continue;
a = std::min(a, (int)d[k+3]);
a = std::min(a, (int)d[k+4]);
a0 = std::max(a0, std::min(a, (int)d[k]));
a0 = std::max(a0, std::min(a, (int)d[k+5]));
}
int b0 = -a0;
for( k = 0; k < 8; k += 2 )
{
int b = std::max((int)d[k+1], (int)d[k+2]);
b = std::max(b, (int)d[k+3]);
if( b >= b0 )
continue;
b = std::max(b, (int)d[k+4]);
b0 = std::min(b0, std::max(b, (int)d[k]));
b0 = std::min(b0, std::max(b, (int)d[k+5]));
}
threshold = -b0-1;
#endif
#if 0
testCorner(ptr, pixel, K, N, threshold);
#endif
return threshold;
}
template<int patternSize>
void FAST_t(InputArray _img, std::vector<KeyPoint>& keypoints, int threshold, bool nonmax_suppression)
{
@ -381,8 +57,6 @@ void FAST_t(InputArray _img, std::vector<KeyPoint>& keypoints, int threshold, bo
#endif
int i, j, k, pixel[25];
makeOffsets(pixel, (int)img.step, patternSize);
for(k = patternSize; k < 25; k++)
pixel[k] = pixel[k - patternSize];
keypoints.clear();

View File

@ -0,0 +1,374 @@
/* This is FAST corner detector, contributed to OpenCV by the author, Edward Rosten.
Below is the original copyright and the references */
/*
Copyright (c) 2006, 2008 Edward Rosten
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
*Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
*Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
*Neither the name of the University of Cambridge nor the names of
its contributors may be used to endorse or promote products derived
from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*
The references are:
* Machine learning for high-speed corner detection,
E. Rosten and T. Drummond, ECCV 2006
* Faster and better: A machine learning approach to corner detection
E. Rosten, R. Porter and T. Drummond, PAMI, 2009
*/
#include "fast_score.hpp"
namespace cv
{
void makeOffsets(int pixel[25], int row_stride, int patternSize)
{
CV_Assert(pixel != 0);
switch(patternSize) {
case 16:
pixel[0] = 0 + row_stride * 3;
pixel[1] = 1 + row_stride * 3;
pixel[2] = 2 + row_stride * 2;
pixel[3] = 3 + row_stride * 1;
pixel[4] = 3 + row_stride * 0;
pixel[5] = 3 + row_stride * -1;
pixel[6] = 2 + row_stride * -2;
pixel[7] = 1 + row_stride * -3;
pixel[8] = 0 + row_stride * -3;
pixel[9] = -1 + row_stride * -3;
pixel[10] = -2 + row_stride * -2;
pixel[11] = -3 + row_stride * -1;
pixel[12] = -3 + row_stride * 0;
pixel[13] = -3 + row_stride * 1;
pixel[14] = -2 + row_stride * 2;
pixel[15] = -1 + row_stride * 3;
break;
case 12:
pixel[0] = 0 + row_stride * 2;
pixel[1] = 1 + row_stride * 2;
pixel[2] = 2 + row_stride * 1;
pixel[3] = 2 + row_stride * 0;
pixel[4] = 2 + row_stride * -1;
pixel[5] = 1 + row_stride * -2;
pixel[6] = 0 + row_stride * -2;
pixel[7] = -1 + row_stride * -2;
pixel[8] = -2 + row_stride * -1;
pixel[9] = -2 + row_stride * 0;
pixel[10] = -2 + row_stride * 1;
pixel[11] = -1 + row_stride * 2;
break;
case 8:
pixel[0] = 0 + row_stride * 1;
pixel[1] = 1 + row_stride * 1;
pixel[2] = 1 + row_stride * 0;
pixel[3] = 1 + row_stride * -1;
pixel[4] = 0 + row_stride * -1;
pixel[5] = -1 + row_stride * -1;
pixel[6] = 0 + row_stride * 0;
pixel[7] = 1 + row_stride * 1;
break;
}
for(int k = patternSize; k < 25; k++)
pixel[k] = pixel[k - patternSize];
}
/*static void testCorner(const uchar* ptr, const int pixel[], int K, int N, int threshold) {
// check that with the computed "threshold" the pixel is still a corner
// and that with the increased-by-1 "threshold" the pixel is not a corner anymore
for( int delta = 0; delta <= 1; delta++ )
{
int v0 = std::min(ptr[0] + threshold + delta, 255);
int v1 = std::max(ptr[0] - threshold - delta, 0);
int c0 = 0, c1 = 0;
for( int k = 0; k < N; k++ )
{
int x = ptr[pixel[k]];
if(x > v0)
{
if( ++c0 > K )
break;
c1 = 0;
}
else if( x < v1 )
{
if( ++c1 > K )
break;
c0 = 0;
}
else
{
c0 = c1 = 0;
}
}
CV_Assert( (delta == 0 && std::max(c0, c1) > K) ||
(delta == 1 && std::max(c0, c1) <= K) );
}
}*/
template<>
int cornerScore<16>(const uchar* ptr, const int pixel[], int threshold)
{
const int K = 8, N = 16 + K + 1;
int k, v = ptr[0];
short d[N];
for( k = 0; k < N; k++ )
d[k] = (short)(v - ptr[pixel[k]]);
#if CV_SSE2
__m128i q0 = _mm_set1_epi16(-1000), q1 = _mm_set1_epi16(1000);
for( k = 0; k < 16; k += 8 )
{
__m128i v0 = _mm_loadu_si128((__m128i*)(d+k+1));
__m128i v1 = _mm_loadu_si128((__m128i*)(d+k+2));
__m128i a = _mm_min_epi16(v0, v1);
__m128i b = _mm_max_epi16(v0, v1);
v0 = _mm_loadu_si128((__m128i*)(d+k+3));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+4));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+5));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+6));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+7));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+8));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
v0 = _mm_loadu_si128((__m128i*)(d+k+9));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
}
q0 = _mm_max_epi16(q0, _mm_sub_epi16(_mm_setzero_si128(), q1));
q0 = _mm_max_epi16(q0, _mm_unpackhi_epi64(q0, q0));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 4));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 2));
threshold = (short)_mm_cvtsi128_si32(q0) - 1;
#else
int a0 = threshold;
for( k = 0; k < 16; k += 2 )
{
int a = std::min((int)d[k+1], (int)d[k+2]);
a = std::min(a, (int)d[k+3]);
if( a <= a0 )
continue;
a = std::min(a, (int)d[k+4]);
a = std::min(a, (int)d[k+5]);
a = std::min(a, (int)d[k+6]);
a = std::min(a, (int)d[k+7]);
a = std::min(a, (int)d[k+8]);
a0 = std::max(a0, std::min(a, (int)d[k]));
a0 = std::max(a0, std::min(a, (int)d[k+9]));
}
int b0 = -a0;
for( k = 0; k < 16; k += 2 )
{
int b = std::max((int)d[k+1], (int)d[k+2]);
b = std::max(b, (int)d[k+3]);
b = std::max(b, (int)d[k+4]);
b = std::max(b, (int)d[k+5]);
if( b >= b0 )
continue;
b = std::max(b, (int)d[k+6]);
b = std::max(b, (int)d[k+7]);
b = std::max(b, (int)d[k+8]);
b0 = std::min(b0, std::max(b, (int)d[k]));
b0 = std::min(b0, std::max(b, (int)d[k+9]));
}
threshold = -b0-1;
#endif
#if 0
testCorner(ptr, pixel, K, N, threshold);
#endif
return threshold;
}
template<>
int cornerScore<12>(const uchar* ptr, const int pixel[], int threshold)
{
const int K = 6, N = 12 + K + 1;
int k, v = ptr[0];
short d[N];
for( k = 0; k < N; k++ )
d[k] = (short)(v - ptr[pixel[k]]);
#if CV_SSE2
__m128i q0 = _mm_set1_epi16(-1000), q1 = _mm_set1_epi16(1000);
for( k = 0; k < 16; k += 8 )
{
__m128i v0 = _mm_loadu_si128((__m128i*)(d+k+1));
__m128i v1 = _mm_loadu_si128((__m128i*)(d+k+2));
__m128i a = _mm_min_epi16(v0, v1);
__m128i b = _mm_max_epi16(v0, v1);
v0 = _mm_loadu_si128((__m128i*)(d+k+3));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+4));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+5));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+6));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
v0 = _mm_loadu_si128((__m128i*)(d+k+7));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
}
q0 = _mm_max_epi16(q0, _mm_sub_epi16(_mm_setzero_si128(), q1));
q0 = _mm_max_epi16(q0, _mm_unpackhi_epi64(q0, q0));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 4));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 2));
threshold = (short)_mm_cvtsi128_si32(q0) - 1;
#else
int a0 = threshold;
for( k = 0; k < 12; k += 2 )
{
int a = std::min((int)d[k+1], (int)d[k+2]);
if( a <= a0 )
continue;
a = std::min(a, (int)d[k+3]);
a = std::min(a, (int)d[k+4]);
a = std::min(a, (int)d[k+5]);
a = std::min(a, (int)d[k+6]);
a0 = std::max(a0, std::min(a, (int)d[k]));
a0 = std::max(a0, std::min(a, (int)d[k+7]));
}
int b0 = -a0;
for( k = 0; k < 12; k += 2 )
{
int b = std::max((int)d[k+1], (int)d[k+2]);
b = std::max(b, (int)d[k+3]);
b = std::max(b, (int)d[k+4]);
if( b >= b0 )
continue;
b = std::max(b, (int)d[k+5]);
b = std::max(b, (int)d[k+6]);
b0 = std::min(b0, std::max(b, (int)d[k]));
b0 = std::min(b0, std::max(b, (int)d[k+7]));
}
threshold = -b0-1;
#endif
#if 0
testCorner(ptr, pixel, K, N, threshold);
#endif
return threshold;
}
template<>
int cornerScore<8>(const uchar* ptr, const int pixel[], int threshold)
{
const int K = 4, N = 8 + K + 1;
int k, v = ptr[0];
short d[N];
for( k = 0; k < N; k++ )
d[k] = (short)(v - ptr[pixel[k]]);
#if CV_SSE2
__m128i q0 = _mm_set1_epi16(-1000), q1 = _mm_set1_epi16(1000);
for( k = 0; k < 16; k += 8 )
{
__m128i v0 = _mm_loadu_si128((__m128i*)(d+k+1));
__m128i v1 = _mm_loadu_si128((__m128i*)(d+k+2));
__m128i a = _mm_min_epi16(v0, v1);
__m128i b = _mm_max_epi16(v0, v1);
v0 = _mm_loadu_si128((__m128i*)(d+k+3));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k+4));
a = _mm_min_epi16(a, v0);
b = _mm_max_epi16(b, v0);
v0 = _mm_loadu_si128((__m128i*)(d+k));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
v0 = _mm_loadu_si128((__m128i*)(d+k+5));
q0 = _mm_max_epi16(q0, _mm_min_epi16(a, v0));
q1 = _mm_min_epi16(q1, _mm_max_epi16(b, v0));
}
q0 = _mm_max_epi16(q0, _mm_sub_epi16(_mm_setzero_si128(), q1));
q0 = _mm_max_epi16(q0, _mm_unpackhi_epi64(q0, q0));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 4));
q0 = _mm_max_epi16(q0, _mm_srli_si128(q0, 2));
threshold = (short)_mm_cvtsi128_si32(q0) - 1;
#else
int a0 = threshold;
for( k = 0; k < 8; k += 2 )
{
int a = std::min((int)d[k+1], (int)d[k+2]);
if( a <= a0 )
continue;
a = std::min(a, (int)d[k+3]);
a = std::min(a, (int)d[k+4]);
a0 = std::max(a0, std::min(a, (int)d[k]));
a0 = std::max(a0, std::min(a, (int)d[k+5]));
}
int b0 = -a0;
for( k = 0; k < 8; k += 2 )
{
int b = std::max((int)d[k+1], (int)d[k+2]);
b = std::max(b, (int)d[k+3]);
if( b >= b0 )
continue;
b = std::max(b, (int)d[k+4]);
b0 = std::min(b0, std::max(b, (int)d[k]));
b0 = std::min(b0, std::max(b, (int)d[k+5]));
}
threshold = -b0-1;
#endif
#if 0
testCorner(ptr, pixel, K, N, threshold);
#endif
return threshold;
}
}

View File

@ -0,0 +1,64 @@
/* This is FAST corner detector, contributed to OpenCV by the author, Edward Rosten.
Below is the original copyright and the references */
/*
Copyright (c) 2006, 2008 Edward Rosten
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
*Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
*Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
*Neither the name of the University of Cambridge nor the names of
its contributors may be used to endorse or promote products derived
from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*
The references are:
* Machine learning for high-speed corner detection,
E. Rosten and T. Drummond, ECCV 2006
* Faster and better: A machine learning approach to corner detection
E. Rosten, R. Porter and T. Drummond, PAMI, 2009
*/
#ifndef __OPENCV_FEATURES_2D_FAST_HPP__
#define __OPENCV_FEATURES_2D_FAST_HPP__
#ifdef __cplusplus
#include "precomp.hpp"
namespace cv
{
void makeOffsets(int pixel[25], int row_stride, int patternSize);
//static void testCorner(const uchar* ptr, const int pixel[], int K, int N, int threshold);
template<int patternSize>
int cornerScore(const uchar* ptr, const int pixel[], int threshold);
}
#endif
#endif