Facial Illumination Migration Method Based on Generative Adversarial Network

Abstract

The invention discloses a face illumination migration method based on a generative confrontation network, comprising the steps of: (1) obtaining sample data of a frontal face image; (2) selecting two images from the sample data as target face images and The reference illuminated image is used as the input of the generator, the generator outputs the re-illuminated image, the discriminator feeds back the error of the real image and the re-illuminated image to the generator, and the classifier feeds back the error of the identity information of the target face image and the re-illuminated image to the generator. The generator, the discriminator and the classifier conduct repeated confrontation training to obtain the optimal face illumination transfer model; (3) face illumination transfer. The invention adopts the structure of generative confrontation network and the loss function, and utilizes the attention mechanism to enable the model to effectively process the details of local illumination. In the network training, there is no need to use the 3D information of the face, and there is no need to align the face, and the end-to-end training has a good re-lighting result.

Description

technical field [0001] The invention relates to the field of face illumination migration and generation network, in particular to a face illumination migration method based on generation confrontation network. Background technique [0002] The lighting effect of face images is widely used in film production, virtual reality, game entertainment, etc., but it takes many years of practical experience to master the lighting processing technology of images. Although image editing techniques can also be used to achieve face lighting processing, professional image editing tools inevitably involve many complex operations. Therefore, it is necessary to develop a more intelligent face illumination processing method. [0003] In the face lighting processing method, one processing method is face lighting migration, that is, the lighting effect on the reference image is transferred to the target image. For the problem of face light transfer, many scholars have proposed a variety of met...

AI Technical Summary

Problems solved by technology

[0004] The existing face light migration methods still have the following shortcomings: face alignment or 3D information of the face is needed to solve the problem, which greatly increases the complexity of the problem

Moreover, for face alignment and the 3D structure of the face, it is necessary to detect the key points of the face, and the algorithm may fail under some extreme lighting conditions, which is also a point to be considered

Images