X_pix_data_normalize = preprocessing.normalize(X_pix_data)
pix_prediction = loaded_model.predict_classes(X_pix_data_normalize)
pix_predict_proba = pd.DataFrame.from_records(
loaded_model.predict_proba(X_pix_data_normalize))
pix_prob_arr = pix_predict_proba.as_matrix()
pix_prob_list = pix_predict_proba.max(axis=1)
for m in range(NoWidth):
pix_dir[i][m] = pix_prediction[m]
pix_dir_prob[i][m] = pix_prob_list[m]
pix_consensus[i][m] = sum(weightTBL[pix_prediction[m]] *
pix_prob_arr[m])
print()
sys.stdout.flush()
pixtri = np.full(fm.shape, 192, dtype=int)
#pixSMT=np.full(fm.shape, 192, dtype=int)
pixSMT = MyL.UT_PixTri(plt, fpfg, pix_dir, pixtri, SMLoop=3)
plt.title(sys.argv[2][-11:])
plt.imshow(pixtri, cmap=plt.cm.gray)
ImgPath = 'D:\\Fingerprint\\paper8_NN\\P8NN_Images\\' + sys.argv[
1] + '_' + sys.argv[2][-11:-4] + "_7.png"
plt.savefig(ImgPath, dpi=600, bbox_inches='tight')
#plt.show()
plt.title(sys.argv[2][-11:])
plt.imshow(pixSMT, cmap=plt.cm.gray)
ImgPath = 'D:\\Fingerprint\\paper8_NN\\P8NN_Images\\' + sys.argv[
1] + '_' + sys.argv[2][-11:-4] + "_8.png"
plt.savefig(ImgPath, dpi=600, bbox_inches='tight')
#plt.show()
