Gaussian depth regularization new viewpoint synthesis method based on hierarchical mask

By using a Gaussian depth regularization method based on layered masks, the problems of geometric blurring and texture discontinuity in complex scenes of existing 3DGS methods are solved, achieving high-precision new viewpoint synthesis and improving the stability and rendering quality of the model.

CN121962469AActive Publication Date: 2026-05-01NANJING UNIV OF POSTS & TELECOMM
4 Cites 1 Cited by

Patent Information

Application Number
CN202610396822.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-03-30
Publication Date
2026-05-01
Estimated Expiration
2046-03-30

AI Technical Summary

Technical Problem

Existing 3DGS methods are prone to problems such as blurred geometry, depth drift, and texture discontinuity in real-world scenes with sparse viewpoints, complex occlusion, or mixed foreground and background. Furthermore, monocular depth noise affects model stability, and multi-view depth continuity and texture consistency are not adequately modeled, making it difficult to achieve high-precision new viewpoint synthesis.

Method used

A Gaussian depth regularization method based on hierarchical masks is adopted. By applying strong supervision to the foreground target and weak constraints to the background region, and combining multi-view depth and texture consistency constraints, hierarchical weight masks are extracted using a semantically guided object detection and segmentation model. Depth alignment is performed by combining sparse point cloud and camera parameters, a random dropout mechanism is introduced for depth rendering, and a multi-view loss function is calculated for optimization.

Benefits of technology

It improves the geometric accuracy and visual consistency of 3D Gaussian model scenes, reduces depth error in the foreground area, reduces depth drift and texture breakage, and improves the stability and rendering quality of the model.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a Gaussian depth regularization new viewpoint synthesis method based on layered masks. Firstly, a layered depth optimization mechanism with foreground strong supervision and background weak constraint is constructed, and multi-view depth continuity and texture consistency collaborative constraint is introduced to realize stable optimization of three-dimensional Gaussian parameters. According to the method, monocular depth noise propagation is effectively inhibited, three-dimensional reconstruction precision and visual consistency are improved, training efficiency and engineering feasibility are considered, and the method is suitable for application scenes such as virtual reality, digital twinning and three-dimensional content generation; the innovation point of the invention is that the mask-guided hierarchical supervision and multi-view collaborative optimization mechanism solves the problem that the existing 3DGS lacks structural constraints, and has clear technical improvement; the method is based on a mature depth estimation and segmentation model, is high in technical maturity, and has an implementation basis.
Need to check novelty before this filing date? Find Prior Art