Digital Bas-relief Optimization And Planar Design

Bas-relief is a semi-stereoscopic sculpture that is suitable for decoration in various environments.Since the rise of 3D printing technology,digital bas-relief modeling has attracted much more researchers' attention.However,little attention has been paid to the repair and optimization of existing bas-relief.In this work,we focus on the optimization of lossy bas-relief and the planar design of bas-relief.The main research content is as follows:(1)Based on sparse representation and low-rank approximation,we propose a bas-relief optimization algorithm.The initial image is first divided into patches of 8 × 8 size in steps of 1,and the edge patches are extracted for denoising.Then,patches are respectively processed by sparse representation and low-rank approximation.During the sparse representation,the K-SVD algorithm is first used to learn a dictionary.Afterwards,noise is filtered by representing each patch as a linear combination of a few dictionary atoms.During the low-rank approximation,block matching is performed for each patch based on an external dictionary library.And each denoised patch is then obtained by treating each similarity matrix with low-rank approximation.Finally,a new height field is reconstructed to generate the optimized bas-relief.Experiment shows that a large amount of noise on the bas-relief surface has been removed,but the overall smoothness and continuity remain to be improved.(2)A twice bas-relief optimization algorithm is proposed to further optimize the reconstructed height field.The second phase of the algorithm is somewhat different from the first phase.First,the approximate representation of noisy patches is completed by using the local similarity of the reconstructed height field and the low-rank property of the similarity matrix.Second,feature enhancement technology is further performed on the approximation results to highlight the interesting features of bas-relief.Compared with image denoising methods and image-based 3D reconstruction method,it is found that our proposed algorithm has better performance in terms of smoothing,denoising and feature enhancement.And the reconstructed bas-relief has a more clear and intuitive visual effect.(3)Based on the idea of feature separation and migration,we combine the weighted least squares filter and the full least squares filter to propose a planar bas-relief modeling method,which not only keeps the surface details effectively,but also realizes the edge enhancement and optimization.In the experiment,a variety of planar bas-relief effects are shown by adjusting the structures and details separated from initial bas-relief.In view of the powerful decorative effect of planar bas-relief,it is of practical significance to realize planar bas-relief modeling with detail preservation.