Image processing method and device

Abstract

The invention discloses an image processing method and device. The image processing method comprises the following steps: extracting a character head portrait from a real scene image and inputting the character head portrait to an encoder to obtain a first vector; the training editor modifies the first vector to obtain a second vector, the modification refers to a description text in a natural language format constructed on the basis of user requirements, and the closeness degree of a character head portrait generated on the basis of the second vector and the description text meets a set requirement; providing the second vector for generating a specified image processing model to obtain a corrected head portrait; and fusing the corrected head portrait and the character head portrait in the live-action image to obtain a new live-action image. Compared with the prior art, the hidden variables of the head portraits are edited through the natural language supervision editor to generate the character head portraits meeting the requirements, and the natural degree of the new live-action image is obtained through head portrait fusion.

Description

technical field [0001] The present disclosure relates to the application field of the combination of neural network and image processing, in particular, to an image processing method and device. Background technique [0002] For enterprises, it will take a lot of manpower and material resources to re-shoot the materials (including posters, videos, leaflets, etc.) that have been photographed, but in some cases, they have to shoot new materials, such as When a product enters a new country, it is often necessary to use local models to shoot new promotional posters. [0003] With the rapid application of neural networks, image processing models have been able to generate high-definition face images, but the generation and replacement of real-life models still face the following difficulties: 1) The models in the original materials often have specific Pose, in order to ensure a good replacement effect, how to controllably generate faces in the same pose is a difficulty; 2) Diffe...

AI Technical Summary

Problems solved by technology

[0003] With the rapid application of neural networks, image processing models have been able to generate high-definition face images, but the generation and replacement of real-life models still face the following difficulties: 1) The models in the original materials often have specific Pose, in order to ensure a good replacement effect, how to controllably generate faces in the same pose is a difficulty; 2) Different demanders have different requirements for character images, such as face shape, hair color, eyes, etc., how to generate corresponding The required profile picture is another difficulty

Images