一种人脸图像中眼镜去除方法及装置制造方法及图纸

一种人脸图像中眼镜去除的方法及装置，用以解决现有技术中存在的戴眼镜的人脸图像在眼镜去除后，眼镜所在的点的像素值为0，人脸信息丢失的问题。所述人脸图像中眼镜去除方法包括：获取待检测图像(S101)；通过第一卷积层提取所述待检测图像中的人脸面部特征(S102)；通过第二卷积层确定所述人脸面部特征中的眼睛区域特征，并过滤所述眼睛区域特征中的眼镜区域特征(S103)；通过第一反卷积层修正经过过滤的人脸面部特征(S104)；通过第二反卷积层生成修正后的所述人脸面部特征中的眼镜区域对应的人脸面部特征，得到去除眼镜的人脸图像(S105)。

A Method and Device for Eyeglasses Removal in Face Image

A method and device for eyeglasses removal in face image is provided to solve the problem that the pixel value of the point where the eyeglasses are located is 0 and the face information is lost after the eyeglasses are removed from the eyeglasses of the face image in the prior art. The eyeglasses removal method in the face image includes: acquiring the image to be detected (S101); extracting the facial features (S102) from the image to be detected by the first convolution layer; determining the eye area features in the face features by the second convolution layer, and filtering the eyeglasses area features (S103) in the eye area features; and correcting the filtered eye area features by the first convolution layer. Facial features (S104) are generated by the second deconvolution layer to generate the corresponding facial features of the eyeglasses area in the corrected facial features, and then the face image (S105) with the eyeglasses removed is obtained.

【技术实现步骤摘要】
【国外来华专利技术】PCT国内申请，说明书已公开。

【技术保护点】
PCT国内申请，权利要求书已公开。

【技术特征摘要】
【国外来华专利技术】PCT国内申请，...

【专利技术属性】
技术研发人员：谭文伟，林倞，张冬雨，
申请(专利权)人：华为技术有限公司，
类型：发明
国别省市：广东,44