Live video background replacement method

Abstract

The invention discloses a live video background replacement method, which comprises the following steps: dividing a whole image into a foreground and a background, the background being basically stationary, and the foreground being changing; the method comprises the following steps: by taking a pixel point as a unit, judging whether the pixel point belongs to a foreground or a background by comparing pixel values of pixel points at the same position in a plurality of adjacent frames of images and mathematical statistical information of pixel point change values in a region, classifying each pixel point to obtain a background region, and replacing an original image with contents of the same region of a background picture. According to the method, the model is simple, the live broadcast video background replacement can be completed only through the pixel value of each pixel point in the live broadcast room video and the background picture YUV color space, the flexibility is high, ideal effects can be obtained for various different live broadcast scenes, and the universality is good.

Description

technical field [0001] The invention belongs to the technical field of digital image processing, and in particular relates to a live video background replacement method. Background technique [0002] With the rapid development of computer and network communication technology, webcasting has become an important way of information dissemination, and is widely used in online shopping, online education, game entertainment and other scenarios, generating a lot of economic benefits, and has become an important Internet profit points. It can replace the traditional teaching and working mode, realize remote teaching and conference communication, and minimize the distance cost and information cost in communication, which has very important practical application value. In network live broadcasting, since the live broadcast environment is generally fixed, the live broadcast background is also fixed; however, due to the requirements of application scenarios, the live broadcast backgrou...

AI Technical Summary

Problems solved by technology

In the actual live broadcast, the foreground includes the human body and its hand-held objects, dynamic posters, various display products, etc. This method only takes the human body area as the foreground, which cannot meet the actual application needs of the live broadcast room

[0004] In terms of theoretical research, most of the current research results are oriented to background extraction in surveillance videos, such as literature [Cao Lei. Surveillance video background extraction and moving target detection based on SIFT algorithm [D]. Xiamen University, 2018], literature [Tang Jialin, Wang Zhenbo , Chen Huachun. Background Extraction in Video Traffic Flow Detection[J]. Shanxi Electronic Technology, 2011(04):39-40+46] and literature [Yang Hui, Han Qirui. Background Extraction and Update of Moving Object Detection in Video Surveillance[ J]. Computer Technology and Development, 2012,22(12):80-82+86] have put forward corresponding technical solutions, but the live video is different from the surveillance video, the foreground part of the scene is generally mostly portraits, and the proportion of The area is large, and the frequency and amplitude of body movements are large. Using ordinary background detection methods suitable for monitoring scenes, the accuracy is low

Images