08-17-2017, 12:10 AM
A Coupled Statistical Model for Face Shape
Recovery From Brightness Images
a statistical model that can be used to recover facial shape from brightness images of faces is developed here. principal components analysis on sets of parameters describing the contents of the intensity images and the facial shape representations are done to construct this model. This coupled model is able to generate accurate shape from out-of-training-sample intensity images. the image irradiance equation is commonly used to extract a ?eld of surface normals and then surface normals are integrated to get surface height function. The
idea is to construct a statistical model in which variations in
image brightness are linked to variations in facial shape
Facial Shape-From-Shading
SFS may be viewed as an integrated process in which surface height is recovered from an input image. The problems are that when inte-
grated, the concave/convex ambiguities in the needle-map can
lead to the distortion of the topography of the reconstructed face. a shape reconstruction method for bi-
laterally symmetric surfaces from a single image has been presented earlier. a novel equal-height contour propagation method for
solving the SFS problem was developed.
SURFACE SHAPE REPRESENTATIONS
Based on directional information,and based on height information, construction of statistical models of facial shape is done. We take orthogonal projection where the viewed surface is assumed to have
been projected into 2-D space of the image plane such that the
direction of the projection axis is opposite to that of the viewer.
for a full report with images refer:
[attachment=3277]
also visit:
http://znu.ac.ir/data/members/fazli_saei...130415.pdf