Open Source Computer Vision Library https://opencv.org/
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

182 lines
5.8 KiB

/*M///////////////////////////////////////////////////////////////////////////////////////
//
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
//
// By downloading, copying, installing or using the software you agree to this license.
// If you do not agree to this license, do not download, install,
// copy or use the software.
//
//
// License Agreement
// For Open Source Computer Vision Library
//
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
// Third party copyrights are property of their respective owners.
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
//
// * Redistribution's of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
//
// * Redistribution's 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.
//
// * The name of the copyright holders may not 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 Intel Corporation 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.
//
//M*/
#include "test_precomp.hpp"
#include <string>
#include <iostream>
#include <fstream>
#include <iterator>
#include <limits>
#include <numeric>
using namespace cv;
using namespace std;
class CV_RigidTransform_Test : public cvtest::BaseTest
{
public:
CV_RigidTransform_Test();
~CV_RigidTransform_Test();
protected:
void run(int);
bool testNPoints(int);
bool testImage();
};
CV_RigidTransform_Test::CV_RigidTransform_Test()
{
}
CV_RigidTransform_Test::~CV_RigidTransform_Test() {}
struct WrapAff2D
{
const double *F;
WrapAff2D(const Mat& aff) : F(aff.ptr<double>()) {}
Point2f operator()(const Point2f& p)
{
return Point2d( p.x * F[0] + p.y * F[1] + F[2],
p.x * F[3] + p.y * F[4] + F[5]);
}
};
bool CV_RigidTransform_Test::testNPoints(int from)
{
cv::RNG rng = ts->get_rng();
int progress = 0;
int k, ntests = 10000;
for( k = from; k < ntests; k++ )
{
ts->update_context( this, k, true );
progress = update_progress(progress, k, ntests, 0);
Mat aff(2, 3, CV_64F);
rng.fill(aff, CV_RAND_UNI, Scalar(-2), Scalar(2));
int n = (unsigned)rng % 100 + 10;
Mat fpts(1, n, CV_32FC2);
Mat tpts(1, n, CV_32FC2);
rng.fill(fpts, CV_RAND_UNI, Scalar(0,0), Scalar(10,10));
transform(fpts.ptr<Point2f>(), fpts.ptr<Point2f>() + n, tpts.ptr<Point2f>(), WrapAff2D(aff));
Mat noise(1, n, CV_32FC2);
rng.fill(noise, CV_RAND_NORMAL, Scalar::all(0), Scalar::all(0.001*(n<=7 ? 0 : n <= 30 ? 1 : 10)));
tpts += noise;
Mat aff_est = estimateRigidTransform(fpts, tpts, true);
double thres = 0.1*norm(aff);
double d = norm(aff_est, aff, NORM_L2);
if (d > thres)
{
double dB=0, nB=0;
if (n <= 4)
{
Mat A = fpts.reshape(1, 3);
Mat B = A - repeat(A.row(0), 3, 1), Bt = B.t();
B = Bt*B;
dB = cv::determinant(B);
nB = norm(B);
if( fabs(dB) < 0.01*nB )
continue;
}
ts->set_failed_test_info(cvtest::TS::FAIL_BAD_ACCURACY);
ts->printf( cvtest::TS::LOG, "Threshold = %f, norm of difference = %f", thres, d );
return false;
}
}
return true;
}
bool CV_RigidTransform_Test::testImage()
{
Mat img;
Mat testImg = imread( string(ts->get_data_path()) + "shared/graffiti.png", 1);
if (testImg.empty())
{
ts->printf( ts->LOG, "test image can not be read");
ts->set_failed_test_info(cvtest::TS::FAIL_INVALID_TEST_DATA);
return false;
}
pyrDown(testImg, img);
Mat aff = cv::getRotationMatrix2D(Point(img.cols/2, img.rows/2), 1, 0.99);
aff.ptr<double>()[2]+=3;
aff.ptr<double>()[5]+=3;
Mat rotated;
warpAffine(img, rotated, aff, img.size());
Mat aff_est = estimateRigidTransform(img, rotated, true);
const double thres = 0.03;
if (norm(aff_est, aff, NORM_INF) > thres)
{
ts->set_failed_test_info(cvtest::TS::FAIL_BAD_ACCURACY);
ts->printf( cvtest::TS::LOG, "Threshold = %f, norm of difference = %f", thres,
norm(aff_est, aff, NORM_INF) );
return false;
}
return true;
}
void CV_RigidTransform_Test::run( int start_from )
{
cvtest::DefaultRngAuto dra;
if (!testNPoints(start_from))
return;
if (!testImage())
return;
ts->set_failed_test_info(cvtest::TS::OK);
}
TEST(Video_RigidFlow, accuracy) { CV_RigidTransform_Test test; test.safe_run(); }