在 PyQt 中嵌入 matplotlib 颜色条

Question

I am developping an app with PyQt4 where I need to display graphs with Matplotlib. To do so, I use the following code, which works well :

self.PhaFig = fig.Figure() #Creation of the Figure
self.PhaBeamCanvas = FigureCanvasQTAgg(self.PhaFig) #Creation of the display canvas
self.AxesPhaInit = self.PhaFig.add_subplot(111, title = "Input beam phase", aspect = 'equal') #Creation of the axes object inside the Figure
imgPha = self.AxesPhaInit.imshow(self.wf_mask,extent =[-xx/2+xx/N,xx/2,-xx/2+xx/N,xx/2])
self.PhaFig.colorbar(imgPha) #adding a colorbar to the graph
self.PhaBeamCanvas.draw() #drawing on the canvas

This graph is updated when a button is pushed. The update of the graph works well. The problem comes from the colorbar. When a new graph is drawn, the previous colorbar is not deleted and the new one is drawn near to the old one.

I have tried something like :

self.PhaFig.delaxes(self.AxesPhaInit)

But this does not work. The graph is deleted but not the colorbar, and the next graph display does not work anymore.

Does someone have an idea?

Answer 1

colorbar ( docs ) has a kwarg cax which in the axis into which the colorbar should be drawn. The default behaver is to 'steal' space in the figure from (iirc) the axes in which the colormappable is living.

Once you have handles on the Axes objects there is no difference between embedding or not embedding so I will demonstrate with non-embedded (which is really just pre-embedded, but I digress).

fig, (ax, cax) = plt.subplots(1, 2)
img = ax.imshow(np.random.rand(42, 42))
fig.colorbar(img, cax=cax)

You will obviously need to size your cax more sensibly and some of the other kwargs on colorbar can help with that.

Answer 2

I guess I'm a little late to this party, but I found a super simple way to solve this problem (which I was encountering myself in my own app).

Each time you update or redraw the plot in the pyqt app, call:

self.PhaFig.clf()

Then add your suplots and draw etc.

Unbelievably enough...It's as simple as that! just using clf refreshes the figure by getting rid of what it contained before.

At least, this worked for me in PyQt5.