Least squares multiple images matching for large coverage aerial image and small coverage UAV image

Kai Zhi Zhan, Tee-Ann Teo

Research output: Contribution to conferencePaperpeer-review

Abstract

Unmanned aerial vehicle (UAV) is widely used to acquire high resolution imagery at multiple viewing angles. The benefit of multi-view images is to provide better intersection geometry. To compare the UAV and traditional aerial photogrammetry, UAV derives 3D structure from different view angles but traditional aerial photogrammetry usually takes photo at vertical view. The integration of these two platforms may improve the viewing geometry. However, there are some difficulties to integrate two platform images as different image-scales, occlusions, illumination changes and acquisition geometry. In this study, we propose a robust image matching method based on least squares matching. The multi-view least squares matching (MVLSM) combines multi-view geometry and least squares matching method to determine the conjugate points. The initial tie points and images scale are obtained manually. Then, we use the MVLSM in precise matching. The test images are UltraCam aerial image and sensefly eBee UAV images. The test area is located at National Chiao Tung University, Taiwan. The MVLSM may improve the matching accuracy at sub-pixels level. Moreover, integrating aerial photo and UAV images matching strategy will be beneficial to the data fusion, data analysis and other applications.

Original languageEnglish
StatePublished - 1 Jan 2014
Event35th Asian Conference on Remote Sensing 2014: Sensing for Reintegration of Societies, ACRS 2014 - Nay Pyi Taw, Myanmar
Duration: 27 Oct 201431 Oct 2014

Conference

Conference35th Asian Conference on Remote Sensing 2014: Sensing for Reintegration of Societies, ACRS 2014
CountryMyanmar
CityNay Pyi Taw
Period27/10/1431/10/14

Keywords

  • Geometric constraint
  • Least squares matching
  • Multi-view image

Fingerprint Dive into the research topics of 'Least squares multiple images matching for large coverage aerial image and small coverage UAV image'. Together they form a unique fingerprint.

Cite this