Adaptive Pca-Based Models To Reconstruct 3d Faces From Single 2d Images

Maghari, Ashraf Y. A. (2014) Adaptive Pca-Based Models To Reconstruct 3d Faces From Single 2d Images. PhD thesis, Universiti Sains Malaysia.

Download (292kB) | Preview


Example-based statistical face models using Principle Component Analysis (PCA) have been widely used for 3D face reconstruction and face recognition. The main concern of this thesis is to improve the accuracy and the efficiency of the PCA-based 3D face shape reconstruction. More precisely, this thesis addresses the challenge of increasing the Representational Power (RP) of the PCA-based model in accordance with the encouraging results of the conducted empirical study. A limited set of training data is utilized towards enhancing the accuracy of 3D reconstruction. Concerning the empirical study, it examines the effect of phenomenal factors (i.e. size of the training set and the variation of the selected training examples) on the RP of 3D PCA-based face models. A regularized 3D face reconstruction algorithm has also been examined to find out how common factors such as the regularization matrix, the number of feature points, and the regularization parameter l affect the accuracy of the 3D face reconstruction based on the PCA model. Importantly, an adaptive PCA-based model is proposed to increase the RP of the 3D face reconstruction model by deforming a set of examples in the training dataset. By adding these deformed samples together with the original training samples, it has been shown that the improvement in the RP can be achieved. Comprehensive experimental validations have been carried out to demonstrate that the proposed model considerably improves the RP of the standard PCA-based model with reduced face shape reconstruction errors.

Item Type: Thesis (PhD)
Subjects: Q Science > QA Mathematics > QA75.5-76.95 Electronic computers. Computer science
Divisions: Pusat Pengajian Sains Komputer (School of Computer Sciences) > Thesis
Depositing User: HJ Hazwani Jamaluddin
Date Deposited: 26 Apr 2021 07:30
Last Modified: 26 Apr 2021 07:30

Actions (login required)

View Item View Item