English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

The 3D reconstruction problem in computer vision is to acquire the third dimension information of a target object from its multiple 2D images. It is a naturally ill-posed problem. As a result, although intensive researching works have been done on this field, the 3D reconstruction is still an open problem. The 3D reconstruction problem normally contains a series of steps including camera calibration, image matching and reconstruction. There are specific problems in each step, however, the results of previous step will also have strong effect on the ones in late stages. For example, the whole reconstruction system accuracy performance is depending on the step with worst error. This leads to the idea of considering the steps jointly. Moreover, most of existing image matching and reconstruction methods work on 2D images because they are the only information available. However, this approach will have difficulty to deal with the variations due to changes of viewpoint. So the task to overcome the variations and obtain dense and accurate reconstruction is another key problem for us. In addition, our research is also trying to develop algorithms for reconstruction with less user intervention, and a fully automatic one will be the ultimate goal. In this thesis, we will firstly study the cooperation between feature detection and feature matching in order to improve the overall performance. The feature selected is the corner. To meet this objective, we redesign the standard corner detector to improve its accuracy and robustness. Also, we redesign the core function of corner NANYANG TECHNOLOGICAL UNIVERSITY SINGAPORE ATTENTION: The Singapore Copyright Act applies to the use of this document. Nanyang Technological University Library

Automatic 3D object reconstruction from multiple 2D image views