Method for constructing three-dimensional model of target object and related apparatus

A computer device, comprising:

• a processor, and a memory, the memory storing program instructions that, when executed by the processor, cause the computer device to perform a method for constructing a three-dimensional (3D) model of a target object including:
• obtaining at least two initial images of a moving target object using a camera from a plurality of camera poses at a reference position, the at least two initial images respectively including depth information of the target object, and the depth information indicating distances between a plurality of points of the target object and the reference position;
• removing, from the at least two initial images, an image whose camera pose is similar to a camera pose of another image to be retained, wherein a similarity of the two camera poses is greater than a preset value;
• obtaining first point cloud information corresponding to the at least two initial images respectively according to the depth information in the at least two initial images;
• fusing the first point cloud information respectively corresponding to the at least two initial images into second point cloud information according to their corresponding camera poses;
• performing Poisson reconstruction on the second point cloud information to obtain a 3D mesh of the target object, the 3D mesh being a non-porous surface connecting points in the second point cloud information;
• and
• tailoring and smoothing the 3D mesh to obtain the 3D model of the target object.

A non-transitory computer-readable storage medium, storing program instructions that, when executed by a processor of a computer device, cause the computer device to perform a method for constructing a three-dimensional (3D) model of a target object including:

• obtaining at least two initial images of a moving target object using a camera from a plurality of camera poses at a reference position, the at least two initial images respectively including depth information of the target object, and the depth information indicating distances between a plurality of points of the target object and the reference position;
• removing, from the at least two initial images, an image whose camera pose is similar to a camera pose of another image to be retained, wherein a similarity of the two camera poses is greater than a preset value;
• obtaining first point cloud information corresponding to the at least two initial images respectively according to the depth information in the at least two initial images;
• fusing the first point cloud information respectively corresponding to the at least two initial images into second point cloud information according to their corresponding camera poses;
• performing Poisson reconstruction on the second point cloud information to obtain a 3D mesh of the target object, the 3D mesh being a non-porous surface connecting points in the second point cloud information;
• and
• tailoring and smoothing the 3D mesh to obtain the 3D model of the target object.