Automatic registration of conjugate points for mobile mapping image sequences

Ren Yu Li, Yi-Hsing Tseng

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

A mobile mapping system (MMS) which equipped with digital cameras is capable of capturing overlapped image sequences with known image orientations referring to the ground coordinate system. The 3D coordinates of a feature point can be determined through an intersection of overlapped image sequences. This calculation needs the input of conjugate image points of the target feature. This paper proposes an automatic method of conjugate point registration for MMS image sequences. The key concept of the proposed method is image matching with the constraint of epipolar geometry. Based on one image point given by an operator, the algorithm is designed to trace conjugate point along the epipolar line of each overlapped images. The employed image matching methods is cross correlation. The matching quality is evaluated using the correlation coefficient of the multiple conjugate image patches finalized. The proposed method was tested with some image sequences of a real MMS case. The results show the feasibility of the method. This algorithm can largely reduce the manual workload of image measurement for MMS image processing.

Original languageEnglish
Title of host publication29th Asian Conference on Remote Sensing 2008, ACRS 2008
Pages1251-1256
Number of pages6
Publication statusPublished - 2008 Dec 1
Event29th Asian Conference on Remote Sensing 2008, ACRS 2008 - Colombo, Sri Lanka
Duration: 2008 Nov 102008 Nov 14

Publication series

Name29th Asian Conference on Remote Sensing 2008, ACRS 2008
Volume2

Other

Other29th Asian Conference on Remote Sensing 2008, ACRS 2008
CountrySri Lanka
CityColombo
Period08-11-1008-11-14

All Science Journal Classification (ASJC) codes

  • Computer Networks and Communications

Fingerprint Dive into the research topics of 'Automatic registration of conjugate points for mobile mapping image sequences'. Together they form a unique fingerprint.

Cite this