Three dimensional reconstruction of scenes from multiple views using active vision

Gibbs, Jonathon (2019) Three dimensional reconstruction of scenes from multiple views using active vision. PhD thesis, University of Nottingham.

[img] PDF (Thesis - as examined) - Repository staff only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Download (3MB)


The need to understand the mechanisms underlying the growth of plants and crops (plant phenotyping) is becoming increasingly important in society, particularly as the quantity of food and biofuel will need to double to meet the demands of the expanding global population, which is likely to exceed nine billion by 2050.

The practical aim of this research is to contribute to reducing the bottleneck associated with plant phenotyping by generating a fully automated response to photometric data acquisition and the recovery of three-dimensional models of plants from multiple views without the dependency of botanical expertise, ensuring a non-intrusive and non-destructive approach.

Plants are complex objects displaying high degrees of concavity and self-occlusion and can be considered examples of crowded scenes. This unavoidable complexity makes careful camera placement a necessity. If a complete 3D reconstruction is to be achieved, viewpoints must be chosen to reflect the broad 3D structure of the plant.

Within this thesis, an Active Vision Cell (AVC), consisting of a camera-mounted robot arm, turntable and automatic image acquisition technique is proposed, along with a novel surface reconstruction algorithm. This approach provides a robust, flexible and accurate approach to automating 3D reconstruction of plants. The active vision method exploits volumetric shape representations to provide a compact image set well-suited to multi-view stereo. The reconstruction algorithm can reduce noise and provides a promising and extendable framework, improving on the current state-of-the-art. Furthermore, the pipeline can be applied to any plant species or form due to its application of an active vision framework combined with the automatic detection of key parameters for surface reconstruction.

Item Type: Thesis (University of Nottingham only) (PhD)
Supervisors: Pridmore, Tony
Murchie, Erik
French, Andrew
Wells, Darren
Keywords: Plant phenotyping; Photometric data acquisition; Three-dimensional models of plants; Image processing; 3D modelling
Subjects: Q Science > QA Mathematics > QA 75 Electronic computers. Computer science
T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800 Electronics > TK8300 Photoelectronic devices
Faculties/Schools: UK Campuses > Faculty of Science > School of Computer Science
Item ID: 56489
Depositing User: Gibbs, Jonathon
Date Deposited: 18 Mar 2020 13:53
Last Modified: 06 May 2020 09:52

Actions (Archive Staff Only)

Edit View Edit View