Surface matching is a fundamental task in 3D computer vision, typically tackled by describing and matching local features computed from the 3D surface. As a result, description of local features lays the foundations for a variety of applications processing 3D data, such as 3D object recognition, 3D registration and reconstruction, and SLAM. A variety of algorithms for 3D feature description exists in the scientific literature. The majority of them are based on different, handcrafted ways to encode and exploit the geometric properties of a given surface. Recently, the success of deep neural networks for processing images has fueled also a data-driven approach to learn descriptive features from 3D data. This chapter provides a comprehensive review of the main proposals in the field.
Spezialetti, R., Salti, S., Di Stefano, L., Tombari, F. (2020). 3D Local Descriptors—from Handcrafted to Learned. Cham : Springer [10.1007/978-3-030-44070-1_7].
3D Local Descriptors—from Handcrafted to Learned
Spezialetti, Riccardo;Salti, Samuele;Di Stefano, Luigi;
2020
Abstract
Surface matching is a fundamental task in 3D computer vision, typically tackled by describing and matching local features computed from the 3D surface. As a result, description of local features lays the foundations for a variety of applications processing 3D data, such as 3D object recognition, 3D registration and reconstruction, and SLAM. A variety of algorithms for 3D feature description exists in the scientific literature. The majority of them are based on different, handcrafted ways to encode and exploit the geometric properties of a given surface. Recently, the success of deep neural networks for processing images has fueled also a data-driven approach to learn descriptive features from 3D data. This chapter provides a comprehensive review of the main proposals in the field.File | Dimensione | Formato | |
---|---|---|---|
2020_book_3D_features__surface_descriptors_and_object_descriptors.pdf
accesso aperto
Tipo:
Postprint
Licenza:
Licenza per accesso libero gratuito
Dimensione
1.88 MB
Formato
Adobe PDF
|
1.88 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.