Research on 3D Face Modeling Based on 2D Images

摘要

The 3D face modeling based on 2D images is in computer vision area overlapping with computer graphics and machine learning. The main purpose of this research is how to change images (e.g. photos) to 3D model by automatically. Mostly of methods for estimate the features on the facial images in both front and side views, for create model on 3D axis system.
　　 This thesis describes an automatic system for 3D face modeling using front and side view images taken by an ordinary digital camera, their direction are orthogonal. The thesis consists of four keys. Firstly study the 3D face modeling guideline including facial features extraction from both frontal and profile images. Secondly explain the novel method‘Full Features Extraction from Images 'modified from Computer Vision and Image Processing technical. Thirdly implement ‘3D Face Modeling Based on 2D Images Software' by C＃ language, EMGU CV library and XNA framework for create 3D face model from images. Finally design experiment, test and analyze the results for measure performance of software in accuracy and time usage.