3D Face Reconstruction System Research Based On The Network

Based on two-dimensional images three-dimensional face reconstruction is a important technology. There are many scenarios in real life for it. Many scholars have put forward a number of effective solutions for3D face reconstruction. But there are many details for the human face reconstruction can be optimized.This thesis tries to research web-based3D face reconstruction system. We research on feature points identification, standard face model, model fitting, texture mapping, and proposed optimal algorithm. Finally, the system is set up on the top of the Django web framework. We carried out the following work.(1) AdaBoost algorithm is used with the shape constraint model to build a facial feature automatically identify classifier, to enable them to quickly and automatically locate the feature points of the human face, the feature points extracted information will be used for specific3D face model adaptation.(2) The Stretching factor is defined. Then we use it to adjust different regions of the feature point and the non-feature point respectively, to assure the feature conformance at utmost.(3) View-dependent texture mapping function is proposed to enhance the effect of the face texture mapping. Different weights are assigned according to the corresponding relations between the corresponding angle and the region on the3D face model.(4) Django web framework is used to build a three-dimensional face reconstruction system.Then we design the front end page.