Infrared and visible images registration with adaptable local-global feature integration for rail inspection

摘要: Abstract Active thermography provides infrared images that contain sub-surface defect information, while visible only reveal surface information. Mapping information to offers more comprehensive visualization for decision-making in rail inspection. However, the common registration is limited due different modalities both local and global level. For example, track which has low temperature contrast reveals rich details images, but turns blurry counterparts. This paper proposes a algorithm called Edge-Guided Speeded-Up-Robust-Features (EG-SURF) address this issue. Rather than sequentially integrating matching stage suffered from buckets effect, adaptively integrates into descriptor gather before matching. adaptability consists of two facets, an adaptable weighting factor between main direction accuracy. The extracted using SURF represented by shape context edges. Meanwhile, generation process, edges are weighted according scale decomposed bins vector decomposition manner provide accurate descriptor. proposed qualitatively quantitatively validated eddy current pulsed scene experiments. In comparison with other algorithms, better performance been achieved.