260 lines
6.9 KiB
C++
260 lines
6.9 KiB
C++
#include <iostream>
|
|
#include <fstream>
|
|
|
|
#include "opencv2/core.hpp"
|
|
#include <opencv2/core/utility.hpp>
|
|
#include "opencv2/highgui.hpp"
|
|
#include "opencv2/cudaoptflow.hpp"
|
|
#include "opencv2/cudaarithm.hpp"
|
|
|
|
using namespace std;
|
|
using namespace cv;
|
|
using namespace cv::cuda;
|
|
|
|
inline bool isFlowCorrect(Point2f u)
|
|
{
|
|
return !cvIsNaN(u.x) && !cvIsNaN(u.y) && fabs(u.x) < 1e9 && fabs(u.y) < 1e9;
|
|
}
|
|
|
|
static Vec3b computeColor(float fx, float fy)
|
|
{
|
|
static bool first = true;
|
|
|
|
// relative lengths of color transitions:
|
|
// these are chosen based on perceptual similarity
|
|
// (e.g. one can distinguish more shades between red and yellow
|
|
// than between yellow and green)
|
|
const int RY = 15;
|
|
const int YG = 6;
|
|
const int GC = 4;
|
|
const int CB = 11;
|
|
const int BM = 13;
|
|
const int MR = 6;
|
|
const int NCOLS = RY + YG + GC + CB + BM + MR;
|
|
static Vec3i colorWheel[NCOLS];
|
|
|
|
if (first)
|
|
{
|
|
int k = 0;
|
|
|
|
for (int i = 0; i < RY; ++i, ++k)
|
|
colorWheel[k] = Vec3i(255, 255 * i / RY, 0);
|
|
|
|
for (int i = 0; i < YG; ++i, ++k)
|
|
colorWheel[k] = Vec3i(255 - 255 * i / YG, 255, 0);
|
|
|
|
for (int i = 0; i < GC; ++i, ++k)
|
|
colorWheel[k] = Vec3i(0, 255, 255 * i / GC);
|
|
|
|
for (int i = 0; i < CB; ++i, ++k)
|
|
colorWheel[k] = Vec3i(0, 255 - 255 * i / CB, 255);
|
|
|
|
for (int i = 0; i < BM; ++i, ++k)
|
|
colorWheel[k] = Vec3i(255 * i / BM, 0, 255);
|
|
|
|
for (int i = 0; i < MR; ++i, ++k)
|
|
colorWheel[k] = Vec3i(255, 0, 255 - 255 * i / MR);
|
|
|
|
first = false;
|
|
}
|
|
|
|
const float rad = sqrt(fx * fx + fy * fy);
|
|
const float a = atan2(-fy, -fx) / (float)CV_PI;
|
|
|
|
const float fk = (a + 1.0f) / 2.0f * (NCOLS - 1);
|
|
const int k0 = static_cast<int>(fk);
|
|
const int k1 = (k0 + 1) % NCOLS;
|
|
const float f = fk - k0;
|
|
|
|
Vec3b pix;
|
|
|
|
for (int b = 0; b < 3; b++)
|
|
{
|
|
const float col0 = colorWheel[k0][b] / 255.0f;
|
|
const float col1 = colorWheel[k1][b] / 255.0f;
|
|
|
|
float col = (1 - f) * col0 + f * col1;
|
|
|
|
if (rad <= 1)
|
|
col = 1 - rad * (1 - col); // increase saturation with radius
|
|
else
|
|
col *= .75; // out of range
|
|
|
|
pix[2 - b] = static_cast<uchar>(255.0 * col);
|
|
}
|
|
|
|
return pix;
|
|
}
|
|
|
|
static void drawOpticalFlow(const Mat_<float>& flowx, const Mat_<float>& flowy, Mat& dst, float maxmotion = -1)
|
|
{
|
|
dst.create(flowx.size(), CV_8UC3);
|
|
dst.setTo(Scalar::all(0));
|
|
|
|
// determine motion range:
|
|
float maxrad = maxmotion;
|
|
|
|
if (maxmotion <= 0)
|
|
{
|
|
maxrad = 1;
|
|
for (int y = 0; y < flowx.rows; ++y)
|
|
{
|
|
for (int x = 0; x < flowx.cols; ++x)
|
|
{
|
|
Point2f u(flowx(y, x), flowy(y, x));
|
|
|
|
if (!isFlowCorrect(u))
|
|
continue;
|
|
|
|
maxrad = max(maxrad, sqrt(u.x * u.x + u.y * u.y));
|
|
}
|
|
}
|
|
}
|
|
|
|
for (int y = 0; y < flowx.rows; ++y)
|
|
{
|
|
for (int x = 0; x < flowx.cols; ++x)
|
|
{
|
|
Point2f u(flowx(y, x), flowy(y, x));
|
|
|
|
if (isFlowCorrect(u))
|
|
dst.at<Vec3b>(y, x) = computeColor(u.x / maxrad, u.y / maxrad);
|
|
}
|
|
}
|
|
}
|
|
|
|
static void showFlow(const char* name, const GpuMat& d_flow)
|
|
{
|
|
GpuMat planes[2];
|
|
cuda::split(d_flow, planes);
|
|
|
|
Mat flowx(planes[0]);
|
|
Mat flowy(planes[1]);
|
|
|
|
Mat out;
|
|
drawOpticalFlow(flowx, flowy, out, 10);
|
|
|
|
imshow(name, out);
|
|
}
|
|
|
|
int main(int argc, const char* argv[])
|
|
{
|
|
string filename1, filename2;
|
|
if (argc < 3)
|
|
{
|
|
cerr << "Usage : " << argv[0] << " <frame0> <frame1>" << endl;
|
|
filename1 = "../data/basketball1.png";
|
|
filename2 = "../data/basketball2.png";
|
|
}
|
|
else
|
|
{
|
|
filename1 = argv[1];
|
|
filename2 = argv[2];
|
|
}
|
|
|
|
Mat frame0 = imread(filename1, IMREAD_GRAYSCALE);
|
|
Mat frame1 = imread(filename2, IMREAD_GRAYSCALE);
|
|
|
|
if (frame0.empty())
|
|
{
|
|
cerr << "Can't open image [" << filename1 << "]" << endl;
|
|
return -1;
|
|
}
|
|
if (frame1.empty())
|
|
{
|
|
cerr << "Can't open image [" << filename2 << "]" << endl;
|
|
return -1;
|
|
}
|
|
|
|
if (frame1.size() != frame0.size())
|
|
{
|
|
cerr << "Images should be of equal sizes" << endl;
|
|
return -1;
|
|
}
|
|
|
|
GpuMat d_frame0(frame0);
|
|
GpuMat d_frame1(frame1);
|
|
|
|
GpuMat d_flow(frame0.size(), CV_32FC2), d_flowxy;
|
|
|
|
Ptr<cuda::BroxOpticalFlow> brox = cuda::BroxOpticalFlow::create(0.197f, 50.0f, 0.8f, 10, 77, 10);
|
|
Ptr<cuda::DensePyrLKOpticalFlow> lk = cuda::DensePyrLKOpticalFlow::create(Size(7, 7));
|
|
Ptr<cuda::FarnebackOpticalFlow> farn = cuda::FarnebackOpticalFlow::create();
|
|
Ptr<cuda::OpticalFlowDual_TVL1> tvl1 = cuda::OpticalFlowDual_TVL1::create();
|
|
Ptr<cuda::NvidiaOpticalFlow_1_0> nvof = cuda::NvidiaOpticalFlow_1_0::create(
|
|
frame0.size().width, frame0.size().height, NvidiaOpticalFlow_1_0::NVIDIA_OF_PERF_LEVEL::NV_OF_PERF_LEVEL_FAST);
|
|
|
|
{
|
|
GpuMat d_frame0f;
|
|
GpuMat d_frame1f;
|
|
|
|
d_frame0.convertTo(d_frame0f, CV_32F, 1.0 / 255.0);
|
|
d_frame1.convertTo(d_frame1f, CV_32F, 1.0 / 255.0);
|
|
|
|
const int64 start = getTickCount();
|
|
|
|
brox->calc(d_frame0f, d_frame1f, d_flow);
|
|
|
|
const double timeSec = (getTickCount() - start) / getTickFrequency();
|
|
cout << "Brox : " << timeSec << " sec" << endl;
|
|
|
|
showFlow("Brox", d_flow);
|
|
}
|
|
|
|
{
|
|
const int64 start = getTickCount();
|
|
|
|
lk->calc(d_frame0, d_frame1, d_flow);
|
|
|
|
const double timeSec = (getTickCount() - start) / getTickFrequency();
|
|
cout << "LK : " << timeSec << " sec" << endl;
|
|
|
|
showFlow("LK", d_flow);
|
|
}
|
|
|
|
{
|
|
const int64 start = getTickCount();
|
|
|
|
farn->calc(d_frame0, d_frame1, d_flow);
|
|
|
|
const double timeSec = (getTickCount() - start) / getTickFrequency();
|
|
cout << "Farn : " << timeSec << " sec" << endl;
|
|
|
|
showFlow("Farn", d_flow);
|
|
}
|
|
|
|
{
|
|
const int64 start = getTickCount();
|
|
|
|
tvl1->calc(d_frame0, d_frame1, d_flow);
|
|
|
|
const double timeSec = (getTickCount() - start) / getTickFrequency();
|
|
cout << "TVL1 : " << timeSec << " sec" << endl;
|
|
|
|
showFlow("TVL1", d_flow);
|
|
}
|
|
|
|
{
|
|
//The timing displayed below includes the time taken to copy the input buffers to the OF CUDA input buffers
|
|
//and to copy the output buffers from the OF CUDA output buffer to the output buffer.
|
|
//Hence it is expected to be more than what is displayed in the NVIDIA Optical Flow SDK documentation.
|
|
const int64 start = getTickCount();
|
|
|
|
nvof->calc(d_frame0, d_frame1, d_flowxy);
|
|
|
|
const double timeSec = (getTickCount() - start) / getTickFrequency();
|
|
cout << "NVIDIAOpticalFlow : " << timeSec << " sec" << endl;
|
|
|
|
nvof->upSampler(d_flowxy, frame0.size().width, frame0.size().height,
|
|
nvof->getGridSize(), d_flow);
|
|
|
|
showFlow("NVIDIAOpticalFlow", d_flow);
|
|
}
|
|
|
|
imshow("Frame 0", frame0);
|
|
imshow("Frame 1", frame1);
|
|
waitKey();
|
|
|
|
return 0;
|
|
}
|