World Library  

QR link for On the Use of Ins to Improve Feature Matching : Volume Xl-1, Issue 1 (07/11/2014)
Add to Book Shelf
Flag as Inappropriate
Email this Book

On the Use of Ins to Improve Feature Matching : Volume Xl-1, Issue 1 (07/11/2014)

By Masiero, A.

Click here to view

Book Id: WPLBN0004014259
Format Type: PDF Article :
File Size: Pages 6
Reproduction Date: 2015

Title: On the Use of Ins to Improve Feature Matching : Volume Xl-1, Issue 1 (07/11/2014)  
Author: Masiero, A.
Volume: Vol. XL-1, Issue 1
Language: English
Subject: Science, Isprs, International
Collections: Periodicals: Journal and Magazine Collection, Copernicus Publications
Historic
Publication Date:
2014
Publisher: Copernicus Publications, Göttingen, Germany
Member Page: Copernicus Publications

Citation

APA MLA Chicago

Pirotti, F., Vettore, A., Guarnieri, A., & Masiero, A. (2014). On the Use of Ins to Improve Feature Matching : Volume Xl-1, Issue 1 (07/11/2014). Retrieved from http://community.worldlibrary.net/


Description
Description: Interdepartmental Research Center of Geomatics (CIRGEO), University of Padova, Viale dellUniversit 16, Legnaro (PD) 35020, Italy. The continuous technological improvement of mobile devices opens the frontiers of Mobile Mapping systems to very compact systems, i.e. a smartphone or a tablet. This motivates the development of efficient 3D reconstruction techniques based on the sensors typically embedded in such devices, i.e. imaging sensors, GPS and Inertial Navigation System (INS). Such methods usually exploits photogrammetry techniques (structure from motion) to provide an estimation of the geometry of the scene.

Actually, 3D reconstruction techniques (e.g. structure from motion) rely on use of features properly matched in different images to compute the 3D positions of objects by means of triangulation. Hence, correct feature matching is of fundamental importance to ensure good quality 3D reconstructions.

Matching methods are based on the appearance of features, that can change as a consequence of variations of camera position and orientation, and environment illumination. For this reason, several methods have been developed in recent years in order to provide feature descriptors robust (ideally invariant) to such variations, e.g. Scale-Invariant Feature Transform (SIFT), Affine SIFT, Hessian affine and Harris affine detectors, Maximally Stable Extremal Regions (MSER).

This work deals with the integration of information provided by the INS in the feature matching procedure: a previously developed navigation algorithm is used to constantly estimate the device position and orientation. Then, such information is exploited to estimate the transformation of feature regions between two camera views. This allows to compare regions from different images but associated to the same feature as seen by the same point of view, hence significantly easing the comparison of feature characteristics and, consequently, improving matching. SIFT-like descriptors are used in order to ensure good matching results in presence of illumination variations and to compensate the approximations related to the estimation process.


Summary
On the use of INS to improve Feature Matching

 

Click To View

Additional Books


  • Relative Pose Estimation Using Image Fea... (by )
  • Segmentation and Crown Parameter Extract... (by )
  • Pleiades Hr in Flight Geometrical Calibr... (by )
  • Assesment of the Influence of Uav Image ... (by )
  • Geospatial Data Sharing, Online Spatial ... (by )
  • Assessment of Coastal Erosion Along Indi... (by )
  • Automated and Accurate Orientation of Co... (by )
  • Urban Archaeology: How to Communicate a ... (by )
  • Prospects of Photon Counting Lidar for S... (by )
  • Using Gps-tracking Technology for Urban ... (by )
  • Ancient Rome Worldwide Links: Sharing Kn... (by )
  • Transmission Line Theory Based Two Layer... (by )
Scroll Left
Scroll Right

 



Copyright © World Library Foundation. All rights reserved. eBooks from World Library are sponsored by the World Library Foundation,
a 501c(4) Member's Support Non-Profit Organization, and is NOT affiliated with any governmental agency or department.