Expression Detail Synthesis Based on Wavelet-Based Image Fusion

Expression details are texture changes caused by facial expression, such as wrinkles in the corner of the mouth when smiling and wrinkles on the forehead when surprising. Expression details can help to enhance the realistic experience of synthesized face image. In this paper, we propose to synthesize expression details by using the method of wavelet-based image fusion. We try to mine the texture feature of expression details for natural expression generation. In order to meet the requirements of individual expression detail synthesis, we use different wavelet transforms, such as the traditional wavelet transform and dual-tree complex wavelet transform, and kinds of fusion operators to get rich results. To seamlessly integrate the synthesized image of expression details to the output expressive face image, we select the optimal replacement for both images by clustering and graph cut method. Our proposed approach is applied to not only grayscale images, but also color images by the color space conversion. The experimental results show that the proposed method is effective on the expression detail synthesis, which can enhance the realistic experience of synthesized face image.