Kernel Local Descriptors with Implicit Rotation Matching

Andrei Bursuc 1 Giorgos Tolias 1 Hervé Jégou 1
1 LinkMedia - Creating and exploiting explicit links between multimedia fragments
Inria Rennes – Bretagne Atlantique , IRISA-D6 - MEDIA ET INTERACTIONS
Abstract : In this work we design a kernelized local feature descriptor and propose a matching scheme for aligning patches quickly and automatically. We analyze the SIFT descriptor from a kernel view and identify and reproduce some of its underlying benefits. We overcome the quantization artifacts of SIFT by encoding pixel attributes in a continuous manner via explicit feature maps. Experiments performed on the patch dataset of Brown et al. [3] show the superiority of our descriptor over methods based on supervised learning.
Document type :
Conference papers
Complete list of metadatas

Cited literature [12 references]  Display  Hide  Download

https://hal.inria.fr/hal-01145656
Contributor : Giorgos Tolias <>
Submitted on : Saturday, April 25, 2015 - 12:05:00 AM
Last modification on : Thursday, November 15, 2018 - 11:58:51 AM
Long-term archiving on : Wednesday, April 19, 2017 - 6:19:28 AM

File

BursucToliasJegou_ICMR2015_Ker...
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01145656, version 1

Citation

Andrei Bursuc, Giorgos Tolias, Hervé Jégou. Kernel Local Descriptors with Implicit Rotation Matching. ACM International Conference on Multimedia Retrieval, 2015, Shanghai, China. ⟨hal-01145656⟩

Share

Metrics

Record views

579

Files downloads

1121