ideas about improving plot module

[sturkmen72]

test code

#include <opencv2/highgui.hpp>
#include <opencv2/plot.hpp>
#include <iostream>

using namespace cv;

int main(int argc, char** argv)
{
	Mat data_x(1, 11, CV_64F);
	Mat data_y(1, 11, CV_64F);

	for (int i = 0; i < 11; i++)
	{
		data_x.at<double>(0, i) = i - 5;
		data_y.at<double>(0, i) = i - 5;
	}

	std::cout << "data_x : " << data_x << std::endl;
	std::cout << "data_y : " << data_y << std::endl;

	Mat plot_result0, plot_result1, plot_result2;

	Ptr<plot::Plot2d> plot = plot::createPlot2d(data_x, data_y);
	plot->render(plot_result0);
	plot->setPlotSize(500, 500);
	plot->render(plot_result1);

	//plot->setPlotAxisColor( Scalar(0, 0, 0) );
	plot->setPlotGridColor( Scalar(0, 0, 0) );
	plot->setPlotTextColor( Scalar(0, 0, 0) );
	plot->render(plot_result2);

	imshow("plot0", plot_result0);
	imshow("plot1", plot_result1);
	imshow("plot2", plot_result2);
	waitKey();

	return 0;
}

1

output plot0 (PlotSize is 600x400 default)

output plot1 (PlotSize is 500x500)

two graphs with same data and different size. i think there is a problem to be solved.

2

actually by setting PlotGridColor,PlotTextColor etc. you can do some customization on display like

probably adding some more functions to customization like setShowGrid(), setShowText() etc. will be fine

[catree]

I never used this plot module but I use matplotlib. My suggestions would be to add the basic features of matplotlib:

• legend, xlabel, ylabel, plot title
• xrange, yrange (already exist with setMaxX, setMaxY, ...)
• save figure
• possibility to subplot
• tick label
• documentation + examples
• 3D plot?

Example image:

Probably for nice plotting people will use dedicated tool (matplotlib, gnuplot, ...). The main purpose of this plot module should be for debugging in my opinion:

• real time plotting: when the xrange and yrange are unknown, be able to adapt accordingly
• save data in a text file to easily plot with another tool: in row, the data at time t, in column the data (xdata, ydata, ...)

Example code to import the data (here the text file is [x1,y1,z1], [x2,y2,z2], [x3,y3,z3], 3 curves, the zdata is not used):

import numpy as np
import matplotlib.pyplot as plt

data = np.loadtxt('pose_info.txt', skiprows=1)
fig = plt.figure()

for i in xrange(0, data.shape[1], 3):
  plt.plot(data[:,i], data[:,i+1])

plt.show()

Probably some features already exist.

[sturkmen72]

i impressed @leonardvandriel's work and opened leovandriel/cvplot#1

i can try to help @leonardvandriel about contributing his framework to OpenCV if it will be useful. @vpisarev @alalek Could you share your opinion on this subject

[wpalfi]

Related to #2414