A Scene Flow Estimation Method Based on 3D Local Rigidity and Depth Map Guided Anisotropic Smoothing

Abstract

The invention relates to a scene flow estimation method based on 3D local rigidity and depth map-guided anisotropic smoothing. S1 uses the RGB‑D sensor to acquire aligned texture images and depth images simultaneously; constructs a scene flow estimation energy functional, and solves the dense scene flow by combining the 3D local rigid surface hypothesis and the global constraint method. The form of the scene flow energy function is: using texture images and depth images, combining the 3D local rigid surface hypothesis to design data items; combining the anisotropic diffusion tensor driven by the depth map and total variation regularization to design the smoothing terms; creating an image pyramid, using a coarse-to-fine solution strategy; using duals The method solves scene flow and introduces scene flow auxiliary variables. This invention uses the inter-pixel color difference of the color image and the positional relationship between pixels to jointly determine the spatial filter weight, thereby solving the problem of edge distortion during the repair process. In order to reduce the repair error, the color information and the structural similarity coefficient are combined to jointly determine the spatial filter weight. Determine the range filter weights.

Description

technical field [0001] The present invention relates to a scene flow estimation method based on 3D local rigidity and depth map guided anisotropic smoothing. Background technique [0002] The scene flow is a 3D dense motion field representing the 3D motion of each point in the actual scene. The scene flow represents the real motion field of the scene and contains depth information, so the scene flow has broad application prospects in the fields of intelligent human-computer interaction, 3D reconstruction and vehicle assisted driving. For scene flow to be widely applicable, accurate estimation of scene flow is a crucial issue. The scene flow solution is an ill-conditioned problem that requires additional assumptions and constraints to solve. [0003] The concept of scene flow was first proposed by Vedula of Carnegie Mellon University in 1999. Scene flow estimation is mainly divided into: scene flow estimation based on binocular stereo vision and scene flow estimation based...

AI Technical Summary

Problems solved by technology

However, in the existing technology, few people pay attention to solving the accuracy of the scene flow and also pay attention to the preservation effect of the moving edge.

Images