New geometry denoising method uses normal vector priors

A new method for geometry denoising has been proposed, utilizing prior knowledge of surface normal vectors. This approach incorporates a set of "label vectors" to guide the denoising process and naturally embeds a segmentation problem. The segmentation is determined by the similarity of normal vectors to these label vectors, with regularization achieved through a total variation term. A split Bregman (ADMM) approach is employed to solve the optimization problem, featuring a vertex update step based on second-order shape calculus. The method has been demonstrated on examples such as denoising an eroded medieval gravestone inscription. AI

RANK_REASON The cluster contains an academic paper published on arXiv. [lever_c_demoted from research: ic=1 ai=0.4]

Read on arXiv cs.CV →

• arXiv
• Roland Herzog

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →