Note
Go to the endto download the full example code.
Image tutorial#
A short tutorial on plotting images with Matplotlib.
Startup commands#
First, let's start IPython. It is a most excellent enhancement to thestandard Python prompt, and it ties in especially well withMatplotlib. Start IPython either directly at a shell, or with the JupyterNotebook (where IPython as a running kernel).
With IPython started, we now need to connect to a GUI event loop. Thistells IPython where (and how) to display plots. To connect to a GUIloop, execute the%matplotlib magic at your IPython prompt. There's moredetail on exactly what this does atIPython's documentation on GUIevent loops.
If you're using Jupyter Notebook, the same commands are available, butpeople commonly use a specific argument to the %matplotlib magic:
In [1]:%matplotlib inline
This turns on inline plotting, where plot graphics will appear in yournotebook. This has important implications for interactivity. For inline plotting, commands incells below the cell that outputs a plot will not affect the plot. For example,changing the colormap is not possible from cells below the cell that creates a plot.However, for other backends, such as Qt, that open a separate window,cells below those that create the plot will change the plot - it is alive object in memory.
This tutorial will use Matplotlib's implicit plotting interface, pyplot. Thisinterface maintains global state, and is very useful for quickly and easilyexperimenting with various plot settings. The alternative is the explicit,which is more suitable for large application development. For an explanationof the tradeoffs between the implicit and explicit interfaces seeMatplotlib Application Interfaces (APIs) and theQuick start guide to start using the explicit interface.For now, let's get on with the implicit approach:
fromPILimportImageimportmatplotlib.pyplotaspltimportnumpyasnp
Importing image data into Numpy arrays#
Matplotlib relies on thePillow library to load image data.
Here's the image we're going to play with:
It's a 24-bit RGB PNG image (8 bits for each of R, G, B). Dependingon where you get your data, the other kinds of image that you'll mostlikely encounter are RGBA images, which allow for transparency, orsingle-channel grayscale (luminosity) images. Downloadstinkbug.pngto your computer for the rest of this tutorial.
We use Pillow to open an image (withPIL.Image.open), and immediatelyconvert thePIL.Image.Image object into an 8-bit (dtype=uint8) numpyarray.
img=np.asarray(Image.open('../../doc/_static/stinkbug.png'))print(repr(img))
array([[[104, 104, 104], [104, 104, 104], [104, 104, 104], ..., [109, 109, 109], [109, 109, 109], [109, 109, 109]], [[105, 105, 105], [105, 105, 105], [105, 105, 105], ..., [109, 109, 109], [109, 109, 109], [109, 109, 109]], [[107, 107, 107], [106, 106, 106], [106, 106, 106], ..., [110, 110, 110], [110, 110, 110], [110, 110, 110]], ..., [[112, 112, 112], [111, 111, 111], [110, 110, 110], ..., [116, 116, 116], [115, 115, 115], [115, 115, 115]], [[113, 113, 113], [113, 113, 113], [112, 112, 112], ..., [115, 115, 115], [114, 114, 114], [114, 114, 114]], [[113, 113, 113], [115, 115, 115], [115, 115, 115], ..., [114, 114, 114], [114, 114, 114], [113, 113, 113]]], shape=(375, 500, 3), dtype=uint8)
Each inner list represents a pixel. Here, with an RGB image, thereare 3 values. Since it's a black and white image, R, G, and B are allsimilar. An RGBA (where A is alpha, or transparency) has 4 valuesper inner list, and a simple luminance image just has one value (andis thus only a 2-D array, not a 3-D array). For RGB and RGBA images,Matplotlib supports float32 and uint8 data types. For grayscale,Matplotlib supports only float32. If your array data does not meetone of these descriptions, you need to rescale it.
Plotting numpy arrays as images#
So, you have your data in a numpy array (either by importing it, or bygenerating it). Let's render it. In Matplotlib, this is performedusing theimshow() function. Here we'll grabthe plot object. This object gives you an easy way to manipulate theplot from the prompt.
You can also plot any numpy array.
Applying pseudocolor schemes to image plots#
Pseudocolor can be a useful tool for enhancing contrast andvisualizing your data more easily. This is especially useful whenmaking presentations of your data using projectors - their contrast istypically quite poor.
Pseudocolor is only relevant to single-channel, grayscale, luminosityimages. We currently have an RGB image. Since R, G, and B are allsimilar (see for yourself above or in your data), we can just pick onechannel of our data using array slicing (you can read more in theNumpy tutorial):
lum_img=img[:,:,0]plt.imshow(lum_img)
Now, with a luminosity (2D, no color) image, the default colormap (aka lookup table,LUT), is applied. The default is called viridis. There are plenty ofothers to choose from.
plt.imshow(lum_img,cmap="hot")
Note that you can also change colormaps on existing plot objects using theset_cmap() method:
imgplot=plt.imshow(lum_img)imgplot.set_cmap('nipy_spectral')
Note
However, remember that in the Jupyter Notebook with the inline backend,you can't make changes to plots that have already been rendered. If youcreate imgplot here in one cell, you cannot call set_cmap() on it in a latercell and expect the earlier plot to change. Make sure that you enter thesecommands together in one cell. plt commands will not change plots from earliercells.
There are many other colormap schemes available. See thelist and imagesof the colormaps.
Color scale reference#
It's helpful to have an idea of what value a color represents. We cando that by adding a color bar to your figure:
Examining a specific data range#
Sometimes you want to enhance the contrast in your image, or expandthe contrast in a particular region while sacrificing the detail incolors that don't vary much, or don't matter. A good tool to findinteresting regions is the histogram. To create a histogram of ourimage data, we use thehist() function.
Most often, the "interesting" part of the image is around the peak,and you can get extra contrast by clipping the regions above and/orbelow the peak. In our histogram, it looks like there's not muchuseful information in the high end (not many white things in theimage). Let's adjust the upper limit, so that we effectively "zoom inon" part of the histogram. We do this by settingclim, the colormaplimits.
This can be done by passing aclim keyword argument in the call toimshow.
plt.imshow(lum_img,clim=(0,175))
This can also be done by calling theset_clim() method of the returned imageplot object, but make sure that you do so in the same cell as your plotcommand when working with the Jupyter Notebook - it will not changeplots from earlier cells.
imgplot=plt.imshow(lum_img)imgplot.set_clim(0,175)
Array Interpolation schemes#
Interpolation calculates what the color or value of a pixel "should"be, according to different mathematical schemes. One common placethat this happens is when you resize an image. The number of pixelschange, but you want the same information. Since pixels are discrete,there's missing space. Interpolation is how you fill that space.This is why your images sometimes come out looking pixelated when youblow them up. The effect is more pronounced when the differencebetween the original image and the expanded image is greater. Let'stake our image and shrink it. We're effectively discarding pixels,only keeping a select few. Now when we plot it, that data gets blownup to the size on your screen. The old pixels aren't there anymore,and the computer has to draw in pixels to fill that space.
We'll use the Pillow library that we used to load the image also to resizethe image.
img=Image.open('../../doc/_static/stinkbug.png')img.thumbnail((64,64))# resizes image in-placeimgplot=plt.imshow(img)
Here we use the default interpolation ("nearest"), since we did notgiveimshow() any interpolation argument.
Let's try some others. Here's "bilinear":
imgplot=plt.imshow(img,interpolation="bilinear")
and bicubic:
imgplot=plt.imshow(img,interpolation="bicubic")
Bicubic interpolation is often used when blowing up photos - peopletend to prefer blurry over pixelated.
Total running time of the script: (0 minutes 11.845 seconds)