Detection and description of local features in images is an essential task in robot vision. This task allows to identify and uniquely specify stable and invariant regions in a observed scene. Many successful detectors and descriptors have been proposed. However, the proper combination of a detector and a descriptor is not trivial because there is a trade-off among different performance criteria. This work presents a comparative study of successful image feature detection and description methods in the context of the simultaneous localization and mapping problem. The considered methods are exhaustively evaluated in terms of accuracy, robustness, and processing time.

© (2019) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.

Martin Gonzalez-Ruiz, Victor H. Diaz-Ramirez, and Rigoberto Juarez-Salazar, "A comparative study of image feature detection and description methods for robot vision," Proc. SPIE 11136, Optics and Photonics for Information Processing XIII, 111360P (Presented at SPIE Optical Engineering + Applications: August 14, 2019; Published: 6 September 2019); https://doi.org/10.1117/12.2528581.