Unlock instant, AI-driven research and patent intelligence for your innovation.

A two-person three-dimensional human body shape optimization reconstruction method in an interactive scene

A technology of human body shape and scene, applied in the field of computer vision, can solve problems such as short effective distance, and achieve the effect of preventing mutual penetration, low environmental requirements, and high precision

Active Publication Date: 2021-11-23
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Using the RGB-D depth camera and using the Fusion4D method proposed by Microsoft, it mainly relies on the algorithm to realize the reconstruction of the human body shape, limited by the sensor, and the effective distance is short

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A two-person three-dimensional human body shape optimization reconstruction method in an interactive scene
  • A two-person three-dimensional human body shape optimization reconstruction method in an interactive scene
  • A two-person three-dimensional human body shape optimization reconstruction method in an interactive scene

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035] The test data of the experiment is a monocular video containing two people interacting in the 3DPW dataset. According to the method of this patent, the three-dimensional shape of the two people interacting in each video frame is reconstructed. Initialized parameters include: T 0 , T 1 , f, c x and c y . The initialization of camera internal parameters is f=2000, c x = w / 2 and c y =h / 2. Among them, w and h represent the height and width of the image, respectively. T 0 and T 1 Represents the translation parameters of the extrinsic parameters of the two cameras, initialized to (0,0,4) T .

[0036] Among them, the weight coefficient λ in step 5 s , lambda 3D , lambda cs , lambda cp and lambda reg Set to 0.01, 1000, 10000, 1000 and 0.005 respectively; in step 6, the weight coefficient λ s , lambda 3D , lambda cs , lambda cp , lambdareg and lambda col Set to 0.01, 1000, 10000, 1000, 0.005 and 1, respectively.

[0037] The metrics evaluated are the averag...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a two-person three-dimensional human body shape optimization reconstruction method in an interactive scene, belongs to the field of computer vision, and particularly relates to a multi-person three-dimensional human body shape reconstruction method. Compared with the prior art, the technology provided by the invention only needs a single video stream for reconstruction; meanwhile, the environment requirement is low, and the system can be suitable for indoor and outdoor scenes; compared with a previous collision detection method, the collision detection method provided by the invention is higher in precision and can better prevent two persons in a space from mutual penetration; compared with a method based on deep learning, the human body three-dimensional shape reconstruction method based on optimization does not need a large amount of training data; according to the hierarchical optimization algorithm provided by the invention, global rotation of the human body can be better recovered, and human body shape reconstruction failure caused by inaccurate initial 3D postures is avoided.

Description

technical field [0001] The invention belongs to the field of computer vision, in particular to a multi-person three-dimensional human body shape reconstruction method. Background technique [0002] Human body shape reconstruction has a wide range of needs in the fields of live sports events, virtual fitting and video monitoring. Existing human body shape reconstruction methods are mainly aimed at single-person reconstruction. Commonly used human body shape reconstruction methods need to use 3D laser scanners, multi-camera systems or RGB-D depth cameras. Among them, the human body shape reconstructed by 3D laser scanner has the highest accuracy, which is usually suitable for indoor scenes. The reconstruction process mainly uses the principle of TOF, and its disadvantages are high cost and poor real-time performance. Methods using multi-camera systems such as Mocap and Vicon systems have high reconstruction accuracy, but they are mainly aimed at the reconstruction of key poi...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06T17/00G06T19/20G06T7/246
CPCG06T17/00G06T19/20G06T7/246G06T2207/10016G06T2207/30196G06T2210/21Y02T10/40
Inventor 郑亚莉李伟
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA