Depth integration and curved-surface evolution based multi-viewpoint three-dimensional reconstruction method

Abstract

The invention discloses a depth integration and curved-surface evolution based multi-viewpoint three-dimensional reconstruction method. The method comprises the following steps: firstly, calibrating a camera array consisting of depth cameras and visible light cameras, projecting depth maps obtained by the depth cameras to a viewpoint in the array by using calibration parameters so as to obtain an initial depth map of the viewpoint, and then obtaining an initial depth curved-surface according to the initial depth map; secondary, defining a curved-surface evolution based energy function, wherein the definition of the energy function comprises the color consistency cost of a plurality of cameras, and the uncertainty and curved-surface smoothing items of the initial depth curved-surface; and finally, driving the depth curved-surface evolution through convexification and minimization on the energy function so as to obtain a final depth curved-surface, namely, three-dimensional scene information. According to the invention, the initial depth information and the color consistency cost of a plurality of cameras are simultaneously fused, therefore, the method is a three-dimensional reconstruction method by which scene space information can be effectively obtained, and can meet various application requirements in three-dimensional televisions, virtual realities and the like.

Description

technical field [0001] The invention relates to a three-dimensional depth information acquisition system, in particular to a multi-viewpoint three-dimensional reconstruction method based on depth fusion and surface evolution. Background technique [0002] With the development of information technology and computer technology, three-dimensional information acquisition technology has become a key technology in the fields of biomedicine, virtual reality, film and television entertainment, etc. Reconnaissance and other fields have broad application prospects. In the above fields, people's requirements for 3D acquisition technology are getting higher and higher. At present, three-dimensional acquisition technologies such as multi-eye stereo vision technology, structured light technology, DFM (depth from motion), SFM (structure from motion) and TOF (Time of Flight) are mainly used. None of the techniques can well meet the requirements of real-time, high precision, high resolutio...

AI Technical Summary

Problems solved by technology

This type of method is usually based on the assumption that the depth edge and the color edge are consistent, and it is difficult to deal with complex scenes with similar foreground and background colors.

In addition, the energy function is usually not convex, which makes it easy to converge to a local optimal solution when the initial value deviates far from the true value, and the complexity of completely solving the energy function is high

Images