Multi-view human body image synthesis method and device based on human body posture

Abstract

The invention belongs to the field of image synthesis technology, and specifically provides a multi-view human body image synthesis method and device based on human body posture. It aims to solve the problem that the existing technology cannot effectively synthesize multi-view images of people and keep the characteristics of people from obvious changes. The invention provides a multi-view human body image synthesis method based on human body posture, which includes posture conversion of the original human body posture based on a posture conversion model and according to the original human body posture and the target perspective; based on the foreground conversion model and based on the original foreground image, original Human body posture and target perspective human posture, perform foreground conversion on the original foreground image; based on the multi-view synthesis model and based on the original image and the target foreground image, perform multi-view synthesis on the original image to obtain a multi-view human body composite image. The method provided by the present invention has the beneficial effects of maintaining human characteristics in synthetic images and synthesizing high-quality synthetic images.

Description

technical field [0001] The invention belongs to the technical field of image synthesis, and in particular relates to a multi-view human body image synthesis method and device based on human body posture. Background technique [0002] Human multi-view image synthesis has important application value in the understanding of human behavior. Using human multi-view synthetic images can effectively solve cross-view problems in computer vision, such as cross-view behavior recognition, cross-view pedestrian re-identification, etc. question. [0003] The existing multi-view image synthesis is mainly aimed at the multi-view synthesis of rigid objects (such as cars, chairs, etc.). . Although the posture of a person is invariant in the three-dimensional space, when performing multi-view synthesis, the human posture in the three-dimensional space is mapped to the two-dimensional space. Perspective human body image synthesis needs to synthesize a two-dimensional pose image corresponding...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the problem that the prior art cannot effectively synthesize multi-view images of people and keep the characteristics of people from changing significantly, the first aspect of the present invention provides a human-based A multi-view human body image synthesis method for gestures, including:

Images