def load_and_f(path,files): #Mapping for lbp mapping = IP.getmapping(8) for k in range(len(files)): #Load file file = os.path.join(path,files[k]) try: file = sio.loadmat(file) Mz = file['Mz'] sz = file['sz'] except NotImplementedError: file = h5py.File(file) Mz = file['Mz'][()] sz = file['sz'][()] #images #Combine mean and sd images image = Mz+sz #Grayscale normalization image = IP.localstandard(image,23,5,5,1) #image = image[20:-20,20:-20] #Feature extraction dict = {'R':9,'r':3,'wc':5,'wr':(5,5)} f1,f2,f3,f4 = IP.MRELBP(image,8,dict['R'],dict['r'],dict['wc'],dict['wr']) #Normalization and mapping of the features f2(large neighbourhood lbp) and f4(radial lbp) #f1 = 1/np.linalg.norm(f1)*f1 f2 = IP.maplbp(f2,mapping) #f2 = 1/np.linalg.norm(f2)*f2 f3 = IP.maplbp(f3,mapping) #f3 = 1/np.linalg.norm(f3)*f3 f4 = IP.maplbp(f4,mapping) #f4 = 1/np.linalg.norm(f4)*f4 #Concatenate features f = np.concatenate((f1.T,f2.T,f3.T,f4.T),axis=0) try: features = np.concatenate((features,f),axis=1) except NameError: features = f return features