Using Deep Autoencoders to Investigate Image Matching in Visual Navigation

摘要: This paper discusses the use of deep auto encoder networks to find a compressed representation an image, which can be used for visual navigation. Images reconstructed from are tested see if they retain enough information as compass (in image is matched with another recall bearing/movement direction) this ability at heart route navigation algorithm. We show that both images and representations different layers in way, suggesting compact code sufficient hold promise finding optimal encodings task.