colorview2d Readme

Use colorview2d to visualize and analize 2d data with (linear) axes.

Features:

• Interactive colorbar adjustment.
• Wide range of adjustable filters (mods) using routines from numpy, scipy and scikit.images:
  • interpolation,
  • Gaussian and median filters,
  • scale, rotate, flip, crop
  • thresholding to extract features,
  • absolute value, natural logarithm, derivation
  • something missing? Add a mod easily.
• Plot to pdf or just use the matplotlib figure.
• Annoyed of matplotlib.pyplots 2d colorplot interface? Simple and convenient plot configuration.
  • Adjust axis labels, their size and font as well as the plot size.
  • Easily adapt the colorbar to your needs.
• Mass extract linetraces (to depict feature evolution).
• Save cv2d config files and restore any modifications easily.
• Save and load data to and from plain text files (gnplot format).

Installation

You can use the python package index via pip

sudo pip2.7 install --upgrade colorview2d

Note: If you receive a 'Could not find a version that satisfies...' error, try to upgrade pip, pip install --upgrade pip

If you are considering writing your own mods then installation into the userspace is preferable (access to colorview2d/mods to place the mod file).

pip2.7 install --user <username> --upgrade colorview2

Usage

I stronlgy recommend to use ipython interactive shell for this tutorial. We initialize some random data with x and y ranges:

import numpy as np
data = np.random.random((100, 100))
xrange = (0., np.random.random())
yrange = (0., np.random.random())

Obtain a :class:`colorview2d.Data` instance to initialize the :class:`colorview2d.View` object:

import colorview2d
data = colorview2d.Data(data, (yrange, xrange))
view = colorview2d.View(data)

Note that the order of the ranges (y range first) is not a typo. It is reminiscent of the rows-first order of the 2d array.

What is the data about? We add some labels:

view.config['Xlabel'] = 'foo (f)'
view.config['Ylabel'] = 'bar (b)'
view.config['Cblabel'] = 'nicyness (n)'

Let us have a look.

view.show_plt_fig()

You should see two figures opening, one containing the plot, the other two simple matplotlib slider widgets to control the colorbar interactively.

We do not like the font and the ticks labels are too small

view.config.update({'Font': 'Ubuntu', 'Fontsize': 16})

Also, the colormap, being default matplotlib's jet, is not greyscale-compatible, so we change to 'Blues' (have a look at the matplotlib documentation to get a list of colormaps).

view.config['Colormap'] = 'Blues'

Its time to plot a pdf and save the config

view.plot_pdf('Nice_unmodified.pdf')
view.save_config('Nice_unmodified.cv2d')

Note: Have a look at the plain text Nice_unmodified.cv2d. The config is just read as a dict. If you modify this file, changes get applied accordingly upon calling load_config if you do not misspell parameter names or options.

If you want to reuse the config next time, just use it upon initialization of the view:

view = cv2d.View(original_data, cfgfile='Nice_unmodified.cv2d')

We realize that there is some (unphysical :) noise in the data. Nicyness does not fluctuate so much along foo or bar and our cheap nice-intstrument produced some additional fluctuations.

view.add_Smooth(1, 1)

This call is a shortcut to view.add_mod('Smooth', (1, 1)). Note that all mods found in the colorview2d/mods folder can be called by add_<Modname>(arg1, arg2, ...). Now we are interested more in the change of our nice landscape and not in its absolute values so we derive along the bar axis

view.add_Derive()

Have a look at the mods/ folder for other mods and documentation on the arguments. It is also straightforward to create your own mod there. Just have a look at the other mods in the folder.

We are interested especially in the nicyness between 0.0 and 0.1.

view.config.update({'Cbmin':0.0, 'Cbmax':0.1})

Alternatively, just use the slider in the second matplotlib figure to control the colorbar limits.

To re-use this data later (without having to invoke colorview2d again), we can store the data to a gnuplot-style plain text file.

colorview2d.fileloaders.save_gpfile('Nice_smooth_and_derived.dat', view.data)

Extending colorview2d

fileloaders

Have a look at the :class:`colorview2d.Data` definition in the :module:`colorview2d.data` module. To create Data we have to provide the 2d array and the bounds of the y and x ranges.

data = colorview2d.Data(
    array,
    ((bottom_on_y_axis, top_on_y_axis),
    (left_on_x_axis, right_on_x_axis)))

To save data, just use the Data attributes, e.g.

my_array = my_view.data.zdata # 2d numpy.array
my_x_range = my_view.data.x_range # 1d numpy.array (left-to-right)
my_y_range = my_view.data.y_range # 1d numpy.array (bottom-to-top)

mods

If you want to apply your own modifications to the data, just put a module inside the colorview2d/mods directory (or package, if you wish). The module should contain a class (with the class name becoming the name of the mod) which inherits from :class:`colorview2d.IMod` and implements the method do_apply(self, data, modargs).

This method is also the right place to document your mods usage, i.e., the required arguments. The docstring of <Modname>.do_apply, where <Modname> is the class's name, is displayed when you call

help(view.add_<Modname>())

In do_apply(self, data, modargs) you can modifiy the datafile freely, there is no error-checking done on the consistency of the data (axes bounds, dimensions). Have a look at the mods/Derive.py module for a minimal example.

To see if your mod is added successfully, have a look at my_view.modlist.

6.10.2015, A. Dirnaichner