Live broadcast method and device and live broadcast server

Abstract

The invention discloses a live broadcast method and device and a live broadcast server, and belongs to the technical field of live broadcast. The live broadcast method comprises the steps: in the livebroadcast process of an anchor, a depth camera collects an image signal of the anchor, and the form of a virtual character displayed in a live broadcast interface is adjusted according to the image signal. The coordinate of each pixel point in the image signal acquired by the depth camera is the coordinate of the corresponding position point in the space coordinate system established by taking the depth camera as the original point, so that the three-dimensional coordinates of each position point of the human body of the anchor can be directly acquired; the three-dimensional coordinates of each position point of the human body do not need to be obtained through a machine learning mode; and the directly acquired three-dimensional coordinates of the position points can obviously represent the real spatial positions of the position points, so that the form of the anchor can be more accurately represented through the form of the virtual character adjusted by the acquired three-dimensionalcoordinates of each position point of the human body of the anchor.

Description

technical field [0001] The present application relates to the field of live broadcast technology, in particular to a live broadcast method, device and live broadcast server. Background technique [0002] With the development of live broadcast technology, for an anchor who is unwilling to appear on camera, the anchor can choose a virtual character from the multiple virtual characters provided by the live broadcast server to replace the anchor for live broadcast, so that the anchor can show his talent. [0003] In the related technology, after the anchor end selects a virtual character, the anchor end collects a plane image of the anchor's human body during the live broadcast process on the anchor end. The plane image of the human body includes two-dimensional coordinates of each position point of the human body. The anchor end determines the three-dimensional coordinates of each position point of the human body based on the plane image of the human body through machine learn...

AI Technical Summary

Problems solved by technology

[0004] Since machine learning is based on experience to determine the three-dimensional coordinates of each position of the human body, the determined three-dimensional coordinates may not be consistent with the actual three-dimensional coordinates, resulting in that the virtual character displayed on the live broadcast interface cannot accurately represent the actions of the anchor

Images