Research On Modeling And Editing Of Portrait Relief Based On Depth Transformation

As an art form between painting and sculpture,relief has been widely used in narratives and decorations.Traditional relief design relies on professional skills and experiences,which is time-consuming.In recent years,automatic or semi-automated methods of relief modeling on three-dimensional objects or two-dimensional images have become a hot research area.Existing relief modeling methods are mainly designed for general types of objects,and are not suitable for processing three-dimensional portrait relief with obvious portrait features.This article focuses on the problem of portrait relief modeling,and makes comprehensive research on depth field transform and inverse transform,which includes two main aspects:(1)A novel method on portrait relief modeling from 3D portrait model is proposed.First,the input 3D portrait model is divided into multiple sub-regions to flexibly process different portrait features.In order to strengthen the important visual features in the relief,the saliency of different regions of the 3D portrait is defined and normal enhanced.After the viewpoint direction is positioned,the input 3D model is converted into a 2.5D height field,and Laplacian-based mesh deformation is applied to optimize the height field.The results show that the geometric details,visual saliency features and depth structure levels of the portrait model can be effectively highlighted,which has a better modeling effect than the existing methods.(2)A novel method on depth inverse transformation problem,e.g.portrait bas-relief thickness editing,is proposed.First,the geometric information of the input bas-relief model such as normal and contour lines is extracted.Then,a normal map inflation algorithm is constructed based on the contour lines to estimate the overall geometry of the height field.At the same time,novel normal composition and decomposition algorithms are proposed and the relief normals are transferred to the inflated normal map.Finally,the thickness of the relief is edited by using Poisson-based surface reconstruction.Experimental results show that the proposed method can effectively edit the thickness of the bas-relief of the portrait,and results are capable to preserve geometric details and visual effects of the original bas-relief.