def similarity(self, pict): # distance of sizes #dim=sum(map(lambda (x,y):(x-y)**2, zip(self.size, pict.size))) #dim/=self.size[0]**2+self.size[1]**2 msr=[] #dimensions=zip(self.size, pict.size) #widths=sorted(dimensions[0]) #heights=sorted(dimensions[1]) #msr.append(sqr(1.*widths[0]/widths[1]*heights[0]/heights[1])) #FIXME: hier irgendwo bleibts haengen, wenn ca. bilder # beim instanziieren ihre local copies nicht laden konnten dimcor = stats.pearson(self.size, pict.size) if dimcor: msr.append(dimcor) #hst=sum(map(lambda (x,y):(x-y)**2, zip(self.histogram, pict.histogram))) hstcor=measure.image_histograms(self, pict) if hstcor: msr.extend(hstcor) mood=measure.image_histmediandist(self, pict) if mood: msr.append(1.-mood) #colorful=measure.image_histrelcor(self, pict) #if colorful: #msr.extend(colorful) #dist = measure.image_hist_dist(self, pict) #if dist: #msr.append(1.-dist/255.) res = 1. #while len(msr)>0: #res *= msr.pop() return sum(msr)/len(msr)
def get_distances(self, other): distances=[] for p in self.members: for q in other.members: dist = measures.image_hist_dist(p,q) for corr in measures.image_histograms(p,q): dist *= 2-corr distances.append(dist) return distances
def similarity(self, pict): # distance of sizes #dim=sum(map(lambda (x,y):(x-y)**2, zip(self.size, pict.size))) #dim/=self.size[0]**2+self.size[1]**2 msr=[] dimensions=zip(self.size, pict.size) widths=sorted(dimensions[0]) heights=sorted(dimensions[1]) msr.append(sqr(1.*widths[0]/widths[1]*heights[0]/heights[1])) #hst=sum(map(lambda (x,y):(x-y)**2, zip(self.histogram, pict.histogram))) hstcor=measure.image_histograms(self, pict) msr.extend(hstcor) mood=measure.image_histmediandist(self, pict) msr.append(1-mood) #colorful=measure.image_histrelcor(self, pict) #msr.extend(colorful) return sum(msr)/len(msr)
