Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.491290
Title: Recovery of surface reflectance and lighting parameters from image data for 3D rendering
Author: Xu, Shida
Awarding Body: Heriot-Watt University
Current Institution: Heriot-Watt University
Date of Award: 2008
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
We present a method to recover the reflectance of objects and the parameters of multiple lights illuminating the scene using a 3D image acquired by a depth sensor and a stereo intensity pair. The methodology has three components: the object geometric model, the surface reflection properties and the light source parameters. In the simpler case, we allow calibration of the point light sources using a Lambertian sphere; this can be used subsequently to estimate the surface reflection properties of an object with known geometry, acquired by an active sensor and represented as a triangulation mesh. However, the main contribution of the thesis is to remove the need for pre-calibration, and estimate the light source parameters, the diffuse and specular surface reflection parameters and the surface geometry from image data alone. By adapting a welldetermined iterative regression process on the illumination equation, a Gaussian Subtraction (GS) algorithm is used to separate the diffuse and specular components of the surface reflectance across the object surface. Then, the locations and intensities of the several light sources are recovered. We demonstrate the approach on both synthetic and real data acquired from a series of objects of reasonable complexity, using up to three light sources. No pre-calibration of the surfaces or of the light positions is necessary. For evaluation, we compare the results to ground truth when available, but also render these objects in dynamic X3DNRML format for a subjective comparison.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.491290  DOI: Not available
Share: