Abstract

We examine methods used for the construction of image descriptors for the global visual indoor localization problem based on the aggregation of local features and deep-learning convolutional neural networks. We propose a criterion for estimating the quality of image matching results and a technique for selecting reference frames in a test sample consisting of images obtained by the sequential motion of the camera. Additionally, we present an evaluation of the efficiency and speed of the considered methods, which determined that methods based on convolutional neural networks provide more reliable image matching than techniques based on local features, although the former exhibit lower processing speed.

© 2017 Optical Society of America

Feature matching based on curve descriptor and local D-Nets for fish-eye images

Yakun Zhang, Hongsheng Zhang, and Wenming Zhang
J. Opt. Soc. Am. A 37(5) 787-796 (2020)

Regularity based descriptor computed from local image oscillations

Leonardo Trujillo, Gustavo Olague, Pierrick Legrand, and Evelyne Lutton
Opt. Express 15(10) 6140-6145 (2007)

ResNet-based image inpainting method for enhancing the imaging speed of single molecule localization microscopy

Zhiwei Zhou, Weibing Kuang, Zhengxia Wang, and Zhen-Li Huang
Opt. Express 30(18) 31766-31784 (2022)

DSCNet: lightweight and efficient self-supervised network via depthwise separable cross convolution blocks for speckle image matching

Lin Li, Peng Wang, Lingrui Wang, Changku Sun, and Luhua Fu
Opt. Express 32(6) 10715-10731 (2024)

Registration of 3D point clouds using a local descriptor based on grid point normal

Jiang Wang, Bin Wu, and Jiehu Kang
Appl. Opt. 60(28) 8818-8828 (2021)