Cross-spectral local descriptors via quadruplet network

Cristhian A. Aguilera, Angel D. Sappa, Cristhian Aguilera, Ricardo Toledo

    Research output: Contribution to journalArticleResearchpeer-review

    11 Citations (Scopus)


    © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This paper presents a novel CNN-based architecture, referred to as Q-Net, to learn local feature descriptors that are useful for matching image patches from two different spectral bands. Given correctly matched and non-matching cross-spectral image pairs, a quadruplet network is trained to map input image patches to a common Euclidean space, regardless of the input spectral band. Our approach is inspired by the recent success of triplet networks in the visible spectrum, but adapted for cross-spectral scenarios, where, for each matching pair, there are always two possible non-matching patches: one for each spectrum. Experimental evaluations on a public cross-spectral VIS-NIR dataset shows that the proposed approach improves the state-of-the-art. Moreover, the proposed technique can also be used in mono-spectral settings, obtaining a similar performance to triplet network descriptors, but requiring less training data.
    Original languageEnglish
    Article number873
    JournalSensors (Switzerland)
    Issue number4
    Publication statusPublished - 15 Apr 2017


    • CNN
    • Cross-spectral
    • Descriptor
    • Infrared

    Fingerprint Dive into the research topics of 'Cross-spectral local descriptors via quadruplet network'. Together they form a unique fingerprint.

    Cite this