Method and device for three-dimensional reconstruction of an object

Abstract

The invention discloses a three dimensional reconstruction method and equipment for an object; the method comprises the following steps: obtaining initial local TSDF values of voxels in a three dimensional space, wherein each initial local TSDF value is matched with one depth map of various depth maps; grouping the initial local TSDF values according to corresponding depth maps, wherein at least one group of initial local TSDF values are matched with more than one depth map; obtaining a global TSDF value of the group according to each group of initial local TSDF values; using the obtained global TSDF values of each group as initial local TSDF values, and solving the optimization problem so as to obtain the final global TSDF value; reconstructing the object in a three dimensional mode according to the obtained final global TSDF value. In the optimization problem, the variables comprise voxel global TSDF values and changing parameters, and a cost function is related to the following factors: the squared weighted summation of the difference between the global TSDF value of a specific voxel and the initial local TSDF value of the converted voxel corresponding to the specific voxel, wherein the weight refers to the weight of a corresponding group of the voxel converted from the specific voxel.

Description

technical field [0001] The present invention relates generally to the field of three-dimensional imaging. In particular, the present invention relates to a method and device capable of accurately performing three-dimensional reconstruction of an object. Background technique [0002] In recent years, with the development of 3D imaging technology, many related applications have emerged, such as augmented reality, digital museum, 3D printing and so on. An important aspect of 3D imaging technology is 3D reconstruction technology. Mainstream 3D reconstruction methods reconstruct 3D objects based on a set of depth images. [0003] Traditional 3D reconstruction methods mainly focus on how to estimate camera parameters more accurately, such as introducing deformation information to estimate camera distortion. However, in the traditional 3D reconstruction method, when calculating the global TSDF (Truncated Signed Distance Function, Truncated Signed Distance Function) representatio...

AI Technical Summary

Problems solved by technology

However, in the traditional 3D reconstruction method, when calculating the global TSDF (Truncated Signed Distance Function, Truncated Signed Distance Function) representation, it simply performs the arithmetic mean of all local TSDF representations, and does not fully explore the local TSDF representation and the global TSDF representation. The relationship between

Since the effect of 3D reconstruction is only improved by improving the estimation of camera parameters, the effect of traditional 3D reconstruction methods is limited, and the accuracy of 3D reconstruction objects needs to be further improved

Images