Evaluation of Global Descriptor Methods for Appearance-Based Visual Place Recognition