提交 84d62b69 编写于 作者: A Andrey Kamaev

Fixed windows build of FREAK

上级 bd901eb5
// freak.cpp // freak.cpp
// //
// Copyright (C) 2011-2012 Signal processing laboratory 2, EPFL, // Copyright (C) 2011-2012 Signal processing laboratory 2, EPFL,
// Kirell Benzi (kirell.benzi@epfl.ch), // Kirell Benzi (kirell.benzi@epfl.ch),
// Raphael Ortiz (raphael.ortiz@a3.epfl.ch) // Raphael Ortiz (raphael.ortiz@a3.epfl.ch)
// Alexandre Alahi (alexandre.alahi@epfl.ch) // Alexandre Alahi (alexandre.alahi@epfl.ch)
// and Pierre Vandergheynst (pierre.vandergheynst@epfl.ch) // and Pierre Vandergheynst (pierre.vandergheynst@epfl.ch)
// //
// Redistribution and use in source and binary forms, with or without modification, // Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met: // are permitted provided that the following conditions are met:
...@@ -119,12 +119,12 @@ void FREAK::buildPattern() ...@@ -119,12 +119,12 @@ void FREAK::buildPattern()
{ {
if( patternScale == patternScale0 && nOctaves == nOctaves0 && !patternLookup.empty() ) if( patternScale == patternScale0 && nOctaves == nOctaves0 && !patternLookup.empty() )
return; return;
nOctaves0 = nOctaves; nOctaves0 = nOctaves;
patternScale0 = patternScale; patternScale0 = patternScale;
patternLookup.resize(FREAK_NB_SCALES*FREAK_NB_ORIENTATION*FREAK_NB_POINTS); patternLookup.resize(FREAK_NB_SCALES*FREAK_NB_ORIENTATION*FREAK_NB_POINTS);
double scaleStep = pow(2.0, (double)(nOctaves)/FREAK_NB_SCALES ); // 2 ^ ( (nOctaves-1) /nbScales) double scaleStep = pow(2.0, (double)(nOctaves)/FREAK_NB_SCALES ); // 2 ^ ( (nOctaves-1) /nbScales)
double scalingFactor, alpha, beta, theta = 0; double scalingFactor, alpha, beta, theta = 0;
// pattern definition, radius normalized to 1.0 (outer point position+sigma=1.0) // pattern definition, radius normalized to 1.0 (outer point position+sigma=1.0)
...@@ -147,21 +147,21 @@ void FREAK::buildPattern() ...@@ -147,21 +147,21 @@ void FREAK::buildPattern()
for( int orientationIdx = 0; orientationIdx < FREAK_NB_ORIENTATION; ++orientationIdx ) { for( int orientationIdx = 0; orientationIdx < FREAK_NB_ORIENTATION; ++orientationIdx ) {
theta = double(orientationIdx)* 2*CV_PI/double(FREAK_NB_ORIENTATION); // orientation of the pattern theta = double(orientationIdx)* 2*CV_PI/double(FREAK_NB_ORIENTATION); // orientation of the pattern
int pointIdx = 0; int pointIdx = 0;
PatternPoint* patternLookupPtr = &patternLookup[0]; PatternPoint* patternLookupPtr = &patternLookup[0];
for( size_t i = 0; i < 8; ++i ) { for( size_t i = 0; i < 8; ++i ) {
for( int k = 0 ; k < n[i]; ++k ) { for( int k = 0 ; k < n[i]; ++k ) {
beta = M_PI/n[i] * (i%2); // orientation offset so that groups of points on each circles are staggered beta = CV_PI/n[i] * (i%2); // orientation offset so that groups of points on each circles are staggered
alpha = double(k)* 2*M_PI/double(n[i])+beta+theta; alpha = double(k)* 2*CV_PI/double(n[i])+beta+theta;
// add the point to the look-up table // add the point to the look-up table
PatternPoint& point = patternLookupPtr[ scaleIdx*FREAK_NB_ORIENTATION*FREAK_NB_POINTS+orientationIdx*FREAK_NB_POINTS+pointIdx ]; PatternPoint& point = patternLookupPtr[ scaleIdx*FREAK_NB_ORIENTATION*FREAK_NB_POINTS+orientationIdx*FREAK_NB_POINTS+pointIdx ];
point.x = radius[i] * cos(alpha) * scalingFactor * patternScale; point.x = static_cast<float>(radius[i] * cos(alpha) * scalingFactor * patternScale);
point.y = radius[i] * sin(alpha) * scalingFactor * patternScale; point.y = static_cast<float>(radius[i] * sin(alpha) * scalingFactor * patternScale);
point.sigma = sigma[i] * scalingFactor * patternScale; point.sigma = static_cast<float>(sigma[i] * scalingFactor * patternScale);
// adapt the sizeList if necessary // adapt the sizeList if necessary
const int sizeMax = ceil((radius[i]+sigma[i])*scalingFactor*patternScale) + 1; const int sizeMax = static_cast<int>(ceil((radius[i]+sigma[i])*scalingFactor*patternScale)) + 1;
if( patternSizes[scaleIdx] < sizeMax ) if( patternSizes[scaleIdx] < sizeMax )
patternSizes[scaleIdx] = sizeMax; patternSizes[scaleIdx] = sizeMax;
...@@ -231,9 +231,9 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat ...@@ -231,9 +231,9 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat
return; return;
if( keypoints.empty() ) if( keypoints.empty() )
return; return;
((FREAK*)this)->buildPattern(); ((FREAK*)this)->buildPattern();
#if CV_SSSE3 #if CV_SSSE3
register __m128i operand1; register __m128i operand1;
register __m128i operand2; register __m128i operand2;
...@@ -246,7 +246,7 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat ...@@ -246,7 +246,7 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat
std::vector<int> kpScaleIdx(keypoints.size()); // used to save pattern scale index corresponding to each keypoints std::vector<int> kpScaleIdx(keypoints.size()); // used to save pattern scale index corresponding to each keypoints
const std::vector<int>::iterator ScaleIdxBegin = kpScaleIdx.begin(); // used in std::vector erase function const std::vector<int>::iterator ScaleIdxBegin = kpScaleIdx.begin(); // used in std::vector erase function
const std::vector<cv::KeyPoint>::iterator kpBegin = keypoints.begin(); // used in std::vector erase function const std::vector<cv::KeyPoint>::iterator kpBegin = keypoints.begin(); // used in std::vector erase function
const float sizeCst = FREAK_NB_SCALES/(FREAK_LOG2* nOctaves ); const float sizeCst = static_cast<float>(FREAK_NB_SCALES/(FREAK_LOG2* nOctaves));
uchar pointsValue[FREAK_NB_POINTS]; uchar pointsValue[FREAK_NB_POINTS];
int thetaIdx = 0; int thetaIdx = 0;
int direction0; int direction0;
...@@ -317,7 +317,7 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat ...@@ -317,7 +317,7 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat
direction1 += delta*(orientationPairs[m].weight_dy)/2048; direction1 += delta*(orientationPairs[m].weight_dy)/2048;
} }
keypoints[k].angle = atan2((float)direction1,(float)direction0)*(180.0/M_PI);//estimate orientation keypoints[k].angle = static_cast<float>(atan2((float)direction1,(float)direction0)*(180.0/CV_PI));//estimate orientation
thetaIdx = int(FREAK_NB_ORIENTATION*keypoints[k].angle*(1/360.0)+0.5); thetaIdx = int(FREAK_NB_ORIENTATION*keypoints[k].angle*(1/360.0)+0.5);
if( thetaIdx < 0 ) if( thetaIdx < 0 )
thetaIdx += FREAK_NB_ORIENTATION; thetaIdx += FREAK_NB_ORIENTATION;
...@@ -328,7 +328,7 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat ...@@ -328,7 +328,7 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat
// extract descriptor at the computed orientation // extract descriptor at the computed orientation
for( int i = FREAK_NB_POINTS; i--; ) { for( int i = FREAK_NB_POINTS; i--; ) {
pointsValue[i] = meanIntensity(image, imgIntegral, keypoints[k].pt.x,keypoints[k].pt.y, kpScaleIdx[k], thetaIdx, i); pointsValue[i] = meanIntensity(image, imgIntegral, keypoints[k].pt.x,keypoints[k].pt.y, kpScaleIdx[k], thetaIdx, i);
} }
#if CV_SSSE3 #if CV_SSSE3
// extracting descriptor by blocks of 128 bits using SSE instructions // extracting descriptor by blocks of 128 bits using SSE instructions
// note that comparisons order is modified in each block (but first 128 comparisons remain globally the same-->does not affect the 128,384 bits segmanted matching strategy) // note that comparisons order is modified in each block (but first 128 comparisons remain globally the same-->does not affect the 128,384 bits segmanted matching strategy)
...@@ -388,7 +388,7 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat ...@@ -388,7 +388,7 @@ void FREAK::computeImpl( const Mat& image, std::vector<KeyPoint>& keypoints, Mat
direction1 += delta*(orientationPairs[m].weight_dy)/2048; direction1 += delta*(orientationPairs[m].weight_dy)/2048;
} }
keypoints[k].angle = atan2((float)direction1,(float)direction0)*(180.0/M_PI); //estimate orientation keypoints[k].angle = static_cast<float>(atan2((float)direction1,(float)direction0)*(180.0/CV_PI)); //estimate orientation
thetaIdx = int(FREAK_NB_ORIENTATION*keypoints[k].angle*(1/360.0)+0.5); thetaIdx = int(FREAK_NB_ORIENTATION*keypoints[k].angle*(1/360.0)+0.5);
if( thetaIdx < 0 ) if( thetaIdx < 0 )
...@@ -438,8 +438,8 @@ uchar FREAK::meanIntensity( const cv::Mat& image, const cv::Mat& integral, ...@@ -438,8 +438,8 @@ uchar FREAK::meanIntensity( const cv::Mat& image, const cv::Mat& integral,
int ret_val; int ret_val;
if( radius < 0.5 ) { if( radius < 0.5 ) {
// interpolation multipliers: // interpolation multipliers:
const int r_x = (xf-x)*1024; const int r_x = static_cast<int>((xf-x)*1024);
const int r_y = (yf-y)*1024; const int r_y = static_cast<int>((yf-y)*1024);
const int r_x_1 = (1024-r_x); const int r_x_1 = (1024-r_x);
const int r_y_1 = (1024-r_y); const int r_y_1 = (1024-r_y);
uchar* ptr = image.data+x+y*imagecols; uchar* ptr = image.data+x+y*imagecols;
...@@ -451,7 +451,7 @@ uchar FREAK::meanIntensity( const cv::Mat& image, const cv::Mat& integral, ...@@ -451,7 +451,7 @@ uchar FREAK::meanIntensity( const cv::Mat& image, const cv::Mat& integral,
ret_val += (r_x*r_y*int(*ptr)); ret_val += (r_x*r_y*int(*ptr));
ptr--; ptr--;
ret_val += (r_x_1*r_y*int(*ptr)); ret_val += (r_x_1*r_y*int(*ptr));
return (ret_val+512)/1024; return static_cast<uchar>((ret_val+512)/1024);
} }
// expected case: // expected case:
...@@ -468,7 +468,7 @@ uchar FREAK::meanIntensity( const cv::Mat& image, const cv::Mat& integral, ...@@ -468,7 +468,7 @@ uchar FREAK::meanIntensity( const cv::Mat& image, const cv::Mat& integral,
ret_val -= integral.at<int>(y_top,x_right); ret_val -= integral.at<int>(y_top,x_right);
ret_val = ret_val/( (x_right-x_left)* (y_bottom-y_top) ); ret_val = ret_val/( (x_right-x_left)* (y_bottom-y_top) );
//~ std::cout<<integral.step[1]<<std::endl; //~ std::cout<<integral.step[1]<<std::endl;
return ret_val; return static_cast<uchar>(ret_val);
} }
// pair selection algorithm from a set of training images and corresponding keypoints // pair selection algorithm from a set of training images and corresponding keypoints
...@@ -583,8 +583,8 @@ void FREAKImpl::drawPattern() ...@@ -583,8 +583,8 @@ void FREAKImpl::drawPattern()
/* FREAK interface implementation */ /* FREAK interface implementation */
FREAK::FREAK( bool _orientationNormalized, bool _scaleNormalized FREAK::FREAK( bool _orientationNormalized, bool _scaleNormalized
, float _patternScale, int _nOctaves, const std::vector<int>& _selectedPairs ) , float _patternScale, int _nOctaves, const std::vector<int>& _selectedPairs )
: orientationNormalized(_orientationNormalized), scaleNormalized(_scaleNormalized), : orientationNormalized(_orientationNormalized), scaleNormalized(_scaleNormalized),
patternScale(_patternScale), nOctaves(_nOctaves), selectedPairs0(_selectedPairs) patternScale(_patternScale), nOctaves(_nOctaves), selectedPairs0(_selectedPairs)
{ {
} }
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册