CurveFigureAnalysis

1.introduction

It's a project target at extracting the line figure from the picture

For example

Change the pic to a binary pic:

We tend to get a picture containing targeted line info, and after process we could get a binary picture.

The picture could be multicolored and having strong noise.

See below two example:

pic type	input	output
the normal background color
the black background color
the real word background

2.basic function

(1)read in the pics of Curve or BrokenLine.

After we get the binary pictures, we have several method

(2)return the functions or points of the figure.

1)in excel mode

(3)figure extraction.

(4)Other mode

1.Extraction of scatters

2.Extraction of hists

3.tech used

(1)basic introduce of figure extraction

We use three techs to detect the figure,for different compute power

1)Using traditional color divide method

Basically we calc the color info and decide which color is the line color.

And through canny and sobel method we also get the processed pic. Finally, we judge the quality of several pic, and return.

2)Using machine learn the filter color

3)Using HED to detect the edge

(2)Points extraction

1)scipy.signal process

We assume that our curve is a signal, and using the low filter to smooth the curve.

4.tutorial

(1)Using python API

from LineFigure import LineFgure
from BrokenLineFigure import BrokenLinrFigure

We define the class of "LineFigure",and subclasses such as BrokenLineFigure and CurveFigure for more detail process

(2)Using website

We build a website for a more portable usage.

5.Acknowledgement

papers:

For more information:

contact us via e-mail: 8203200527@csu.edu.cn