Last weekend the Python Edinburgh users group hosted a mini-conference. Saturday morning was kicked off with a series of talks followed by sessions introducing and then focusing on contributing to django prior to sprints which really got going on the Sunday.
The slides for my talk on, "Images and Vision in Python" are now available in pdf format here.
The slide deck I used is relatively lightweight with my focus being on demonstrating using the different packages available. The code I went through is below.
fromPILimportImage
#Openanimageandshowitpil1=Image.open('filename')pil1.show()
#Getitssizepil1.size
#Resizepil1s=pil1.resize((100,100))
#or-thumbnailpil1.thumbnail((100,100),Image.ANTIALIAS)
#Newimagebg=Image.new('RGB',(500,500),'#ffffff')
#Twowaysofaccessingthepixels
#getpixel/putpixelandload
#loadisfasterpix=bg.load()forainrange(100, 200):forbinrange(100,110):pix[a,b]=(0,0,255)bg.show()
#DrawingshapesisslightlymoreinvolvedfromPILimportImageDrawdraw=ImageDraw.Draw(bg)draw.ellipse((300,300,320,320),fill='#ff0000')bg.show()fromPILimportImageFontfont=ImageFont.truetype("/usr/share/fonts/truetype/freefont/FreeSerif.ttf", 72)draw.text((10,10),"Hello",font=font,fill='#00ff00')bg.show()
#Demo'sforvisionfromscipyimportndimageimportmahotas
#Createasampleimagev1=np.zeros((10,10),bool)v1[1:4,1:4]=Truev1[4:7,2:6]=Trueimshow(v1,interpolation="Nearest")imshow(mahotas.dilate(v1),interpolation="Nearest")imshow(mahotas.erode(v1),interpolation="Nearest")imshow(mahotas.thin(v1),interpolation="Nearest")
#Opening,closingandtop-hatascombinationsofdilateanderode
#Labeling
#Latestversionofmahotashasalabelfuncv1[8:,8:]=Trueimshow(v1)labeled,nr_obj=ndimage.label(v1)nr_objimshow(labeled,interpolation="Nearest")pylab.jet()
#Thresholding
#Convertagrayscaleimagetoabinaryimagev2=mahotas.imread("/home/jonathan/openplaques/blueness_images/1.jpg")T=mahotas.otsu(v2)imshow(v2)imshow(v2>T)
#DistanceTransformsdist=mahotas.distance(v2>T)imshow(dist)