MDGHM-SURF: A robust local image descriptor based on modified discrete Gaussian–Hermite moment

作者:

Highlights:

• We propose a novel family of the SURF descriptor.

摘要

• We propose a novel family of the SURF descriptor.• To solve the sensitivity of the SURF, we redefine the SURF using MDGHM (MDGHM-SURF).• The MDGHM offers more feature information than the first-order derivative-based SURF.• The evaluations are conducted with six types of deformations and panoramic image generation.• The results indicate that our method outperforms other SURF related algorithms.

论文关键词:Modified discrete Gaussian–Hermite moment,Local feature extraction,SURF algorithm

论文评审过程:Received 17 December 2013, Revised 10 May 2014, Accepted 23 June 2014, Available online 2 July 2014.

论文官网地址:https://doi.org/10.1016/j.patcog.2014.06.022