update doc

Author: rflamary

README.md

@@ -80,6 +80,55 @@ Dependencies:
 This demo uses [pygame](https://www.pygame.org/) for visualization and
 [PyAudio](http://people.csail.mit.edu/hubert/pyaudio/) for microphone recording. 
 
+## Real time 2D FFT of webcam
+
+In this demonstration we perform online 2D fft of a webcam input. Both the
+absolute value and the log of the absolute values are displayed and the 0
+frequency is centered on the window (fftshift).
+
+Screenshot:
+
+![screenshot](data/screen_fft.png "screenshot")
+
+Shortcuts:
+
+* <kbd>q</kbd> : Quit demo
+* <kbd>s</kbd> : Save screen to png
+* <kbd>Space</kbd> : Pause FFT
+* <kbd>w</kbd> : Active or deactivate windowing
+
+
+Dependencies:
+
+This demo uses [opencv-python](https://github.com/skvark/opencv-python) for
+webcam access and visualization.
+
+
+
+## Real time 2D filtering of webcam
+
+In this demonstration we perform online 2D filtering of a webcam input. Both the
+original, filtered and FFT of the filtered video with 0
+frequency centered on the window (fftshift) are displayed.
+
+Screenshot:
+
+![screenshot](data/screen_2dfilter.png "screenshot")
+
+Shortcuts:
+
+* <kbd>q</kbd> : Quit demo
+* <kbd>s</kbd> : Save screen to png
+* <kbd>Space</kbd> : Apply/deactivate filter
+* <kbd>f</kbd> : Change filter (average, high pass, Prewitt, Sobel, median, ...)
+* <kbd>+</kbd> <kbd>-</kbd> : Change cutoff frequency of filter (or its size)
+* <kbd>w</kbd> : Active or deactivate windowing
+
+Dependencies:
+
+This demo uses [opencv-python](https://github.com/skvark/opencv-python) for
+webcam access and visualization.
+
 ## 2D classification demo
 
 In this demonstration we illustrate the decision function and update of 2D
@@ -103,6 +152,32 @@ Dependencies:
 This demo uses [pygame](https://www.pygame.org/) for visualization and
 [Scikit-learn](https://scikit-learn.org/) for classification.
 
+## 2D Dictionary learning demo
+
+In this demonstration we illustrate dictionary learning on 2D data. The user cla
+click on the window to add samples (left click) and to remove samples (right
+click). The dictionary is learned in real time and the atoms are displayed.
+
+
+
+![screenshot](data/screen_dico_2D.png "screenshot")
+
+* <kbd>q</kbd> : Quit demo
+* <kbd>left click</kbd> : Add sample
+* <kbd>right click</kbd> : Remove sample
+* <kbd>c</kbd> : Clear training data
+* <kbd>Space</kbd> : Show/hide dictionary atoms and color samples wrt their representation
+* <kbd>m</kbd> : Change method (PCA, ICA KMeans, NMF, SparsePCA, DictionaryLearning)
+* <kbd>p</kbd> : Show projected and reconstructed samples
+* <kbd>Up</kbd> : Increment the number of atoms
+* <kbd>Down</kbd> : Decrement the number of atoms
+* <kbd>s</kbd> : save screenshot
+
+Dependencies:
+
+This demo uses [pygame](https://www.pygame.org/) for visualization and
+[Scikit-learn](https://scikit-learn.org/) for classification.
+
 
 ## Style Transfer Demo
 

demos.ini

@@ -9,6 +9,16 @@ text=Real time Total Variation demo on webcam
 depends=opencv-python\nprox_tv
 file=demos/demo_tv.py
 
+[fft]
+text=Real time FFT of webcam 
+depends=opencv-python
+file=demos/demo_fft.py
+
+[2dfilter]
+text=Real time 2D filtering of webcam 
+depends=opencv-python
+file=demos/demo_2d_filter.py
+
 [ot]
 text=Demo of Optimal Transport between discrete dstributions
 depends=pygame>1.9.1\nPOT\nnumpy\nscipy
@@ -19,6 +29,11 @@ text=Demo of 2D classification (LDA, SVM)
 depends=numpy\nscipy\nscikit-learn
 file=demos/demo_classif_2D.py
 
+[dicolearn_2D]
+text=Demo of 2D Dictionary learning (PCA, ICA, NMF, Sparse DL, KMeans) 
+depends=numpy\nscipy\nscikit-learn
+file=demos/demo_dicolearn_2D.py
+
 [style_transfer]
 text=Demo style transfer between images
 depends=numpy\ntorch\ntorchvision\nopencv-python

demos/demo_2d_filter.py

@@ -101,17 +101,17 @@ def apply(x):
         def apply(x):
             h=np.array([[-1,0,1],[-1,0,1],[-1,0,1]])/3
             return np.maximum(np.minimum(sp.signal.convolve(x,h,'same'),1),0)
-        txt= 'Prewit Horiz. Abs.'
+        txt= 'Prewitt Horiz. Abs.'
     if idfilt==5:
         def apply(x):
             h=h=np.array([[-1,0,1],[-1,0,1],[-1,0,1]]).T/3
             return np.maximum(np.minimum(sp.signal.convolve(x,h,'same'),1),0)
-        txt= 'Prewit Vert. Abs.'
+        txt= 'Prewitt Vert. Abs.'
     if idfilt==6:
         def apply(x):
             h=h=np.array([[-1,0,1],[-1,0,1],[-1,0,1]]).T/3
             return abs(sp.signal.convolve(x,h,'same'))**.5+abs(sp.signal.convolve(x,h.T,'same'))**.5
-        txt= 'Prewit Both Abs.'       
+        txt= 'Prewitt Both Abs.'       
     if idfilt==7:
         def apply(x):
             h=np.array([[-1,0,1],[-2,0,2],[-1,0,1]])/4
@@ -228,13 +228,8 @@ def apply(x):
         cut=0.5        
     if (key & 0xFF) in [ ord('F')]:
         idfilt-=1         
-    if (key & 0xFF) in [ ord('h')]:
-        if not half:
-            do_tv=True
-            half=True
-        else:
-            half=False
-        
+        s=1
+        cut=0.5
 
 # When everything done, release the capture
 cap.release()